0 / 60 seg.

We've processed the footage using an algorithm based on Google's Deep Dream to simulate the effects of overly strong perceptual predictions.