What will we leave to future generations? portraits of machine dreaming about animals that once inhabited Earth? Is this the only way to remember them, just a residual memory?
An exploration of anxieties around the wildlife such as global warming, destroyed habitat, shortage of food, and human predatory practices. Our exclusive artist, diavlex, uses AI as a tool of expression to represent the residual image of species affected by human predatory actions.
TL;DR OF OUR APPROACH
The core AI technique for this collection is our Neural Painter implementation from scratch. Neural Painters, introduced in , are Generative Models models able to paint using brushstrokes, in contrast to common pixel-level ones. We consider Neural Painters to be a more natural choice for Art+AI artwork and, therefore, our preferred choice.
We also use a tailor-made Non-Adversarial Super-Resolution model based on a dynamic U-Net architecture . The approach is inspired by fast.ai’s ‘Decrappification’ technique .
 Neural Painters: A Learned Differentiable Constraint for Generating Brushstroke Paintings. Reiichiro Nakano. Preprint. 2019. https://arxiv.org/abs/1904.08410
 U-Net: Convolutional Networks for Biomedical Image Segmentation. Olaf Ronneberger, Philipp Fischer, Thomas Brox. Medical Image Computing and Computer-Assisted Intervention (MICCAI), Springer, LNCS, Vol.9351: 234–241, 2015, available at arXiv:1505.04597
 Decrappification, DeOldification, and Super Resolution. Jason Antic (Deoldify), Jeremy Howard (fast.ai), and Uri Manor (Salk Institute). 2019. https://www.fast.ai/2019/05/03/decrappify/