This film uses a detection algorithm trained to identify the arrangement of bodies on narrative TV. The program sets to work analysing non-narrative footage of a collection of antelope skulls at the Oxford Museum of Natural History. The animate and inanimate compete for the algorithm’s eyes. As skulls invite us to imagine what the animal once looked like, so the detection algorithm attempts to retrieve something lost. The residual here is doubled: These are bones, but the program watching the action understands their function differently. Is there a non-anthropocentric mode of editing that can emerge between these two modes of seeing?