An algorithm explores the web by indexing images showing faces. These, which represent millions, are classified by the proximity of facial features using a neural network. Displayed one after the other, the images found produce an animation that evolves across the entire frame: anonymous people merge with past and contemporary celebrities through surprising transitions, and a meta-face seems to emerge in a new heterogeneous and disturbing space, which continues to generate itself. The visual component of big data is thus playfully transposed in the form of image-by-image animation, mirroring the cinematographic device in its origins.