Man Lost in the Convergence of Time, Avebury (2022): Reconfiguring film through human figure removal and collage

This article outlines the processes taken in composing the piece man lost in the convergence of time, a series of prints in which generative diffusion models are used to abstract video through language, by initially removing the human from a film shoot at Avebury’s ceremonial path, Wiltshire (2022). With this work we seek to reveal intrinsic bias in high- resolution image models released to the public for appropriation, introducing sets of text prompts which anchor video-to-video translation allegorical to science fiction. We developed a collage methodology where a secondary model opposed to the diffusion process predicts masks using dichotomous image segmentation, allowing us to composite the footage and mask to further diffusion steps recursively. Our series illustrates different attempts where we were successful in creating formal abstraction with semantic consistency and carry on a discussion on possible futures of experimental compositing techniques with CLIP-guidance conditioned by field recordings. We acknowledge our work promotes reinterpretation of space in moving image, as it has been in different generative model contexts.