Cycle

Cycle

Lei Han and Lorraine Walsh
CYCLES, 2021
Animation, drawings
Duration 3:16, dimensions variable

This animation coalesces various elements from the artists’ projects in the exhibition, depicting drawings from The In-Between set against a background of machine learned images from the NASA Earth Observatory. Shifting through a palette of colors to represent a warming climate, the video of the Earth’s changing surface is generated from a model trained using StyleGAN2– a Generative Adversarial Network (GAN) effective at generating synthetic images from massive image databases such as those at NASA. 

GANs are not supervised. Instead they “teach” itself how to imitate or mimic any given database–in this case thousands of images–to create new, artificial content based on machine learning algorithms. This distribution or training of data is often called the latent space of the model, hence model trained, with algorithms. 

October 27, 2022