Blade Runner Gets Trippy When Auto-Encoded by an AI
Ashley Allen / 7 years ago
A trippy recreation of 1982 sci-fi classic Blade Runner – rendered from a simple set of digital data by an auto-encoding computer system – was so eerily close to the original that the resulting Vimeo video was hit by a DMCA takedown notice from Warner Bros.
So, How do Androids Interpret Electric Sheep? That’s what computer researcher Terence Broad wanted to find out. As part of his Master’s degree dissertation in Creative Computing, Broad developed a neural network to interpret video data and auto-encode it, and what better to feed a computer system designed to interpret and define reality than the work of Philip K. Dick?
The work of Dick – author of cerebral sci-fi novel Do Androids Dream of Electric Sheep, which served as the inspiration for Blade Runner – deals with themes of perception and the subjective nature of reality.
Broad’s artificial neural network is left to work out how to encode compressed video data on its own, without help from its creator. Here is the result:
“The type of neural network used is an autoencoder,” Broad wrote on Medium. “An autoencoder is a type of neural net with a very small bottleneck, it encodes a data sample into a much smaller representation (in this case a 200 digit number), then reconstructs the data sample to the best of its ability. The reconstructions are in no way perfect, but the project was more of a creative exploration of both the capacity and limitations of this approach.”
Broad trained the neural network to look at the selected frames from Blade Runner, compare it to “false” data – unrelated to the film – in order to help it develop its own understanding of what should be “true,” effectively quantifying what the film should look like.
The neural network also tackled fellow Dick adaptation A Scanner Darkly:
After Broad explained the project to Warner Bros., his (or rather, the neural network’s) Blade Runner video was reinstated. “No one has ever made a video like this before, so I guess there is no precedent for this and no legal definition of whether these reconstructed videos are an infringement of copyright,” Broad told Vox.