POLAE
Polæ (2021) directed by Steven Schardt, is an interactive sci-fi film for mobile devices that premiered in the 2021 Tribeca Film Festival’s Immersive category.
Polæ uses machine learning and augmented reality to deliver a personalized, interactive storyline that involves each participant.
PRESS: Tribeca News & No Proscenium
Machine Learning in Polæ
A render I created for the film. This is a noise ramp in StyleGAN. The video begins with complete colorful noise, and converges on the participant’s face. Technically: it begins with a random vector in the latent space, then the random noise is gradually ramped down to uncover the faces in the latent space of the FFHQ Flickr Faces StyleGAN model.
I developed the real-time machine learning system and ML special effects for Polæ, using generative adversarial networks hosted on AWS servers.
When playing Polæ, the participant is asked to take a photo of themselves. Their photo is then sent to a generative ML model which generates five images of human faces that bear an uncanny resemblance to the original face. Subsequently, amidst the storyline’s imminent world disaster, the participant is posed with an uncomfortable choice: In the coming post-Earth world, we may be able to save you through cloning, but we cannot replicate your biological form perfectly. Would you accept living as any of these recreated human forms?
Film Description
Polæ is an interactive science-fiction series for mobile devices depicting the future history of Adelaide Engelsen and her company. What begins with a cryptic series of articles and social media posts leads to a multimedia breaking news experience. The viewer’s decisions will determine their chances for survival when they come into direct communication with the artificial intelligence deciding the future.