Exploring a Digital Art Collection through Drawing Interactions with a Deep Generative Model.

New Snow is an interactive drawing table that investigates human interaction with a deep generative model based on Edvard Munch’s sketching practice. Through drawings with pen and paper, you can interact with the model which will return synthetic sketches based on the input drawings in near real time. The StyleGAN 2 model is trained on 5800 of Edvard Munch’s sketches, and is thus constrained to representing images within the latent space of his sketching practice. As you draw on the tracing paper an infrared sensitive camera records the lines your produce. A TouchDesigner and OpenCV program cleans up the sketches and sends them to a pixel2style2pixel encoder that maps the line drawings into the latent space of the StyleGAN 2 model. This returns a synthetic Edvard Munch sketch that is projected back onto the paper. This process is repeated multiple times per second. As you draw and familiarize yourself with the model, you get to experience the visual aesthetic belonging to Munch’s sketching practice. This potential for familiarizing yourself with the aesthetic of a dataset through drawing has implications for human-AI interaction and non-verbal art mediation.


Project members: Christian Sivertsen, René Haas, Halfdan Hauch Jensen, MUNCH

A picture of the process, from drawing to output