For this project I created a dataset of ~700 images that I had saved from Instagram of digital art. Using RunwayML I trained a StyleGAN using transfer learning from it's original landscape parameters to creating art using my dataset. As it trained I continued to find more art on Instagram and built a larger dataset of ~1300 images (which included the original ~700 images). After the StyleGAN was trained with 3000 steps I retrained it with another 3000 steps using the newer, larger dataset. I had it generate a latent image walk as well as 50 images which are previewed below.