This work shows a latent space walk through a trained model of the TreeGAN. The generative adversarial network has been trained on 76 different types of 3D trees that I made, and in this animation, we can see the network shifting between a single young tree and a small group of trees.
These works use a Generative Adversarial Network in 3D voxel space which has been trained on a dataset of 3D trees, which I made in reference to Chinese literati and German Romantic painting as well as paintings of trees by Mondrian, Schiele and Constable. The resulting animations show the machine learning network trying to make 3D trees. I see this as a continuation of the artistic question of form and how artists throughout history have struggled to represent the world around them.