Project Esper: The Filmic Fascia, The Cinematic Backface and Backface Dreaming

This annotated portfolio, which presents three connected pieces of Research through Creative Practice, synthesises findings from tests, experiments, prototypes and outcomes, which is developed to explore the nascent methods and processes needed to respatialising cinema. Project Esper applies computer vision workflows, to reconstruct cinematic space from existing story worlds for use in virtual production and immersive media.
By feeding shots, sequences and scenes from existing filmic narratives into these computational models, flat cinematic images are reconstituted into spatial, explorable environments. The camera's original flattening of space into image and sequence becomes a kind of sedimentary record - noisy, partial, and layered - from which computational logics reassembles and reconstitutes depth, and extrudes form.
In this process, the cinematic shot is both training data and geological trace. As scenes are parsed by the algorithms, and passed between processing nodes in the workflow, they generate 3D environments composed from the visible surfaces - the Filmed Fascia - while occluded zones, untraced by the camera, remain unresolved. These blank zones become cinematic cavities: gaps, escarpments, and fissures in the reconstructed cinematic surfaces, where the geometry of the world drops away. Some of these absences reflect missing data; others reveal spaces never constructed on set – negative space, non-places made visible only in their absence – the Cinematic Backface.
To resolve these voids, video diffusion models are introduced as a synthetic imagination, generating plausible continuations of the cinematic world. This process - which we call Backface Dreaming - imagining, inferring and extrapolating the unseen, filling in occlusions, to create a seamless, explorable cinematic space. The result is a kind of cinematic geology: layers of visual data transformed into immersive strata, where spatial depth is mined from flattened narrative, and synthetic sediment fills the gaps left by production’s blind spots.
This work raises theoretical questions about the limits of cinematic vision, the materiality of cinematic space, how critical making and Research through Practice can be used as a form of critical enquiry to surface new approaches, new positionings to the text, and to explore the poetics of reconstruction through AI.