Diffusions in Architecture: Artificial Intelligence and Image Generators
by Matias del Campo, Lev Manovich
Preface
Lev Manovich
Introduction to Diffusion in Architecture
This book is a unique document of the beginning of a true revolution in cultural imagination and creation. This revolution has been in development for over 20 years. The first AI papers proposing that the web universe of texts, images, and other cultural artifacts can be used to train computers to do various tasks appeared already in 1991–2001. In 2015, Google’s deep dream and style transfer methods attracted lots of attention: suddenly computers could create new artistic images mimicking the styles of many famous artists. The release of DALL‐E 2 in 2021 was another milestone: now computers could synthesize images from text descriptions. MidJourney, Stable Diffusion, and DALL‐E 2 all contributed to the acceleration of this evolution in 2022. Synthetic images could not have many aesthetics that range from photo realism to any kind of physical or digital medium, including mosaics, oil paintings, street photography, or 3D CG rendering. The code for producing such images (also referred to as a model in the field of artificial intelligence) was made public in August 2022, sparking a flurry of experiments and accelerating development.
I’ve been using computer tools for art and design since 1984, and I’ve seen a few major media revolutions, including the introduction of Mac computers, the development of photorealistic 3D computer graphics and animation, the rise of the web after 1993, and the rise of social media sites ...