Bob Gan Output 01
Bob Gan Output 02
Bob Gan Output 03
Bob Gan Output 04
Bob Gan Output 05
Bob Gan Output 06
Bob Gan Output 07
Bob Gan Output 08
Bob Gan Output 09
Bob Gan Output 10
Bob Gan Output 11
Bob Gan Output 12
Bob Gan Output 13
Bob Gan Output 14
Bob Gan Output 15
Bob Gan Output 16

Ideal Worker (Bob)

The first project, Ideal Worker (Bob), is a set of images generated from a StyleGAN2 generative adversarial network trained (with RunwayML) on a dataset of 5,000 images of people named Bob collected from LinkedIn resume profiles. None of these people are real. This project was an investigation of the notion of an “ideal worker,” and what such a person might look like based on aggregate LinkedIn data mining. The categorization of people by basic nickname (I collected multiple datasets of people with three-letter-long names) feels paradoxically relevant and ridiculous in our racially and politically polarized American landscape.

Car Fire Gan Output 01
Car Fire Gan Output 02
Car Fire Gan Output 03
Car Fire Gan Output 04
Car Fire Gan Output 05
Car Fire Gan Output 06
Car Fire Gan Output 07
Car Fire Gan Output 08
Car Fire Gan Output 09
Car Fire Gan Output 10
Car Fire Gan Output 11
Car Fire Gan Output 12
Car Fire Gan Output 13
Car Fire Gan Output 14
Car Fire Gan Output 15
Car Fire Gan Output 16

Conflagration

The second project, Conflagration, is a set of images generated from a StyleGAN2 model trained (with Google Colab) on a dataset of 2,500 images of cars on fire collected from across the internet. The truncation of these images has been adjusted to skew away from accurate representations, towards an idea of what a burning vehicle might be. The intent here was to assess how to the network would handle to conflicting aspects of these images--the geometric and stylistically repeating car bodies with the fractal nature of flames and smoke.