Cinematic Hyperobject : the Man from London

A film is typically experienced as a temporal flow or stream of images. But a film can also be understood spatially. Every frame is a selection from a set or “space” of possible images. The temporal and spatial aspects are both essential constituents of cinema. The term “hyperobject” in this context highlights, first of all, this twofold aspect of cinema, and secondly, the high-dimensional nature of the space of any movie, or a group of movies. This project employs methods from manifold learning, a family of unsupervised machine learning algorithms used for dimensionality reduction, to visualize films as hyperobjects.

The visualization used here allows the viewer to experience a movie both as a temporal stream of images and as a spatial object. The temporal flow of the movie can be seen as a “walk” within this spatial object.