The emergence of text-to-image generative models (e.g., Midjourney, DALL-E 2, Stable Diffusion) in the summer of 2022 impacted architectural visual culture suddenly, severely, and seemingly out of nowhere. To contextualize this phenomenon, in an issue of The International Journal of Architectural Computing, I offer a socio-technical history of contemporary text-to-image generative systems.

The "Architecture After AI" exhibition and symposium was reviewed in the March 10th edition of the Architect's Newspaper. The article by Aaron Seward is titled "A symposium at UT Austin stakes out initial signposts for architecture’s future with artificial intelligence".

In a chapter of Diffusions in Architecture - Artificial Intelligence and Image Generators, edited by Matias del Campo, I argue that in the context of prompt-based architectural visualization, designers must learn to navigate landscapes beyond the technical.

The "Origins and Destinations Beyond Midjourney" panel was reviewed in the November 11th edition of the Architect's Newspaper. The article by Davis Richardson is titled "2022 ACADIA conference, Hybrids & Haecceities, covers extra-disciplinary collaboration, the importance of tiny things, and the ethics of artificial intelligence".

In a chapter of Artificial Intelligence and Architecture: From Research to Practice, edited by Stanislas Chaillou, I argue that while recent developments in artificial intelligence threaten to dislodge some of our basic assumptions about the nature and purpose of computational design tools, creative designers ought to welcome the disruption.

I was invited to contribute the first chapter of The Routledge Companion to Artificial Intelligence in Architecture, an edited volume that surveys the state of the art in Artificial Intelligence (AI) as it relates to architecture. In this chapter, I sketch out of a number of models for practice that I see emerging in the area of machine-augmented architectural design, each defined in terms of how it situates the use of ML in design differently.

This paper presents a technical account of the development of an the augmented architectural drawing tool Sketch2Pix: an interactive application that sup-ports architectural sketching augmented by automated image-to-image translation processes.

This paper documents the approach taken by and the work produced in an undergraduate research studio conducted at UC Berkeley in the Spring of 2020. Here, a series of small design projects examine the applicability of machine-augmented sketching tools to early-stage architectural design.

In work exhibited at the NeurIPS 2017 Machine Learning for Creativity and Design, we developed a process for relating depthmaps extracted from Google Street View panoramas with the corresponding photographic information. A number of separate models were produced using limited geographic areas of selected cites. With these depthmap-to-panoramic cityscape models trained, we are able to generate new images from unrelated depthmaps which resembled photographic images of the selected cities. This is demonstrated using a javascript app (not currently online) and documented in still images and a series of videos.

This paper proposes the production of synthetic images of cities using generative adversarial networks (or GANs) represents the first computational approach to documenting the Genius Loci of a city, which is understood to include those forms, textures, colors, and qualities of light that exemplify a particular urban location and that set it apart from similar places. Presented here are methods for the collection of urban image data, for the necessary processing and formatting of this data, and for the training of two known computational statistical models (StyleGAN and Pix2Pix) that identify visual patterns distinct to a given site and that reproduce these patterns to generate new images.

This post on the Towards Data Science blog concisely summarizes the GAN Loci project - a project which applies techniques in machine learning in order to produce synthetic images intended to capture the predominant visual properties of urban places.

This paper presents a framework for the application of Machine Learning (ML) to Generative Architectural Design (GAD), and illustrates this framework through a description of a series of projects completed at the Smart Geometry conference in May of 2018 (SG 2018) in Toronto.

In work exhibited at the University of Toronto in the context of the 2018 Smart Geometry Conference, three-dimensional architectural massings for single-family homes are generated by a generative adversarial network. This GAN is trained on a small dataset of three-dimensional models of homes falling into seventeen architectural styles, and that are represented as multi-view heightfield images.

This workshop cluster offered at the 2018 Smart Geometry Conference in Toronto was the initial catalyst for the Fresh Eyes project, and was the first incorporation of user-generated image recognition models into the evaluation step of a traditional generative design workflow.

This project uniquely links the familiar parametric environment of Grasshopper with cloud-hosted models trained using Lobe.ai: a user-friendly ML graphic programming environment that runs Tensorflow.

This workshop offered at the Design Modeling Symposium in Berlin presents tools and techniques for the application of Machine Learning (ML) to Generative Architectural Design (GAD).

A toolkit for connecting parametric models in Grasshopper with hosted image classification machine learning models.

In work exhibited at the NeurIPS 2018 Machine Learning for Creativity and Design, three-dimensional architectural massings for single-family homes are generated by a generative adversarial network. This GAN is trained on a small dataset of three-dimensional models of homes falling into seventeen architectural styles, and that are represented as multi-view heightfield images. A process is developed for converting from 3d CAD model to 2d tiled heightfield image, and from the 2d heightfield images generated by GAN back to three-dimensions in voxel format.

This paper argues that prevailing approaches to CAD software have been fashioned to support modes of reasoning only of secondary importance to design activity, and that, due to some recent developments in computer vision, this state of affairs may be about to change. Surveying the current state of CAD tools, a critical position is developed based upon the best current understanding of the cognitive processes related to design.

The nature of design tools is related to the social relationships they serve. This paper speculates on the emergence of a new professional configuration - the synthesis of architect and engineer - and on the nature of new computational tools and methods that will be required to support such a reconfiguration.