Let Me Dream Again

Let Me Dream Again is a series of experiments that use machine learning to try to recreate lost films from the fragments of early cinema from Hollywood and Western Europe that still exist and can be found on the internet. While previous work I have made has tried to use machine learning as a disintegration loop, destroying what was there until no meaning was left, I wanted to see if it was possible to do the opposite: to remake something that has disappeared. Some of these experiments are on this page, including an endlessly evolving, algorithmically generated film which exists only on this website. There are strong parallels between early cinema and imagery generated by machine learning. Early film-makers had to invent a new visual language, much like artists working with machine learning are doing today. Similarly, machine learning work often places heavy emphasis on the machine that created the visuals, rather than the visuals themselves, also an echo of early cinema. Both were considered niche technologies in their infancies, and both try to record and reflect the world as seen by those in control, at times exposing implicit bias or more critical views. GANs are a way to experiment with this space that overlays past with present, real with dreamlike. GANs generate images which, when put together to create a film, move in a strange, unearthly way which defy the rules of logic on how objects and people should behave. However these same characteristics make it very challenging to build a film with a clear narrative out of what was generated. The gifs that are now playing came from an experiment where a new film was made, which contains a shadow film of stills from the original cinema clips, that sits behind the GAN generated film, moving in and out of the foreground. The process blends the past and present, real and dreamlike, recutting and reconfiguring one hundred year old footage to join together things that shouldn't be joined, yet at the same time creating a structure and a coherence.