Intelligent Design : Mutation[‘]s Fruit


3150 x 6000 px 

GAN, plastic puzzle  

In this series of image transformations I seek to explore the ideas of metamorphosis and mutation through the use of Generative Adversarial Networks that take imagery and use a large dataset of many thousands of images of all types of things pulled from the internet. GANs provided me an interesting opportunity to visually explore the question of entropy, and experiment with mutation, since they function as simulated natural systems. The dataset which my GAN was trained on was composed of a highly diverse set of images ranging from people, and animals to landscapes, and inanimate objects like technology, or furniture and most things in between. The GAN pits two neural networks against each other to generate artificial, and entirely unique instances of data that are rated, and passed or rejected by a discriminator as ‘“real” or “fake”. By using a controlled, modifiable input which can be reconfigured (the puzzle) I have been able to explore the various mutations possible from a single starting place. The puzzle is a cheap plastic child’s toy that features 4 conventional fruits: an apple, orange, banana, and strawberry and can be completely mixable as all the pieces are identical and can be moved anywhere as long as they are connected to the same side. In addition to the ability to shuffle the pieces, I appreciated the connection between fruits and mutation/evolution. The output images often feature vaguely or even highly recognizable elements in strange configurations, textures, or contexts (such as synthesized environments, the atmosphere surrounding the elements). They seem both familiar and strange to a viewer and have the ability to represent many different things to different people as the images all contain familiar ‘visual DNA’*, sort of how a Rorschach test requires a viewer to reflect on the response to a given stimulus

Using Format