GENERATIVE ADVERSARIAL NETWORKS // FALL 2022 CUSTOM GAN / GANPAIGN CHAIR

fashion

An exploration of the use of custom-trained AI models — in this case a Generative Adversarial Network. The dataset for this model was a set of 500+ images from Thomas Parsons's book "Designer's Guide to Scandinavian Patterns". RunwayML was then used to train a StyleGAN — and then used to generated a set of new glyph-like pattern elements.

These elements were curated and arranged into a pattern, then woven into uphostery fabric. The fabric was used to make a sling for a campaign-style sling chair.

diana
diana
diana
Inputs from the training dataset
diana
diana
diana
Outputs from the trained model
diana
abuelo
tricycle
building

CONNECT

LinkedIn Instagram