Seascapes in the Style of Monet and Degas

This is my final project for Computational Portraiture.

I used BigGan with Google Collab in order to create AI's interpretation of Seascapes by famous impressionist artists. I wanted to create an interactive gallery experience that could be played through and infinite, looping as a GAN does. There are many dogs in the exhibit, the player will notice. These are all based on different seeds input into the Generator. These first iterations, some of which are birds and some of which are dogs, are then refined into the second iteration, which are at the later half of the exhibit. As the player moves through the exhibit, the images in the frames change, showing the progression of how a GAN moves through latent space. I also allowed the project to loop in order to add continuity to the space. In order to create the space I created a small room with four picture frames. Only two invisible triggers control an array which contains all of the possible images that the frames can draw from, as the player moves through the triggers, the frames change, and the frames that change will always be on the opposite side of the player. This space was inspired by games such as Superliminal, impossible spaces that could not exist in reality thrive in the digital world. An infinite gallery would never exist in the real world, due to constraints placed upon us by the real world, but using Unity, with a little clever code, they can exist with relative ease.