AI image exploration. All images and videos algorithmically created with the generative adversarial network BigGAN.
The concept was inspired by Bernd and Hilla Becher’s typological studies of industrial structures. The project images resemble human-made objects, but unlike the Becher’s 20th century typologies, these are not a photographic record of constructions or a study of form. Instead, these images were generated digitally using artificial intelligence. The final images were human-selected and grouped together by classification.
There are seven sets of images arranged in grids of nine. Additionally, each set was used to generate a latent walk video, meaning the images were used as keyframes while the video frames in between are interpolated.