I experimented with training AI to create images through working with the website Playform.io, to which I uploaded a set of 108 photos of landscapes — using my own photographs and found imagery. Specifically, I wanted to see what it would mean to utilize machine learning to create “artificial landscapes”, if we gave AI images of our natural environments. Below is a final selection of the images and a video showing the evolutions between them.
During the training process of three hours, there is an interesting thing about seeing the image output change over time akin to the development of analog film, as the digital images gradually reveal themselves.
Moving back toward the ideas of natural and artificial environments, how can these images prompt us to think about nature in new ways? I am curious about how these new landscapes relate to our “real” natural and built environments, as the images seem in-between the language of natural topography and the architectural structures of our urban spaces.