Returning to the Preserving Machine: Mutations and the Probabilistic Image

Milnes, Tom ORCID: https://orcid.org/0000-0002-5181-2063, Ainsworth, Peter and Plagerson, Sam (2026) Returning to the Preserving Machine: Mutations and the Probabilistic Image. Visual Resources, n/a. pp. 1-14. ISSN 1477-2809

Preview

Text (Journal article in peer-reviewed journal Visual Resources) Returning to the Preserving Machine Mutations and the Probabilistic Image.pdf - Published Version Available under License Creative Commons Attribution. Download (491kB) | Preview

Abstract / Summary

This commentary revisits the article ‘The Photogrammetric Image and Black Boxed Mutative Automation’ from the shared perspective of The Preserving Machine collective, returning to our original text to examine how its concerns have transformed within the contemporary ecology of computational imaging. Re-engaging with Philip K. Dick’s 1953 short story ‘The Preserving Machine’ as a conceptual hinge, we explore how themes of preservation, mutation, machinic agency and infrastructural opacity resurface across today’s imaging systems. Since we wrote our earlier work, photogrammetry’s once-visible ruptures and torn surfaces (while still present within contemporary image capture) have shifted into the hidden operations of probabilistic volumetric imaging methods. Techniques such as neural radiance fields (NeRFs) and Gaussian splatting use trained models to interpret sense data in real time. Instability resides within optimisation procedures, weighting schemes and dataset-derived assumptions, shifting the onto-epistemological stakes of the image away from sensory capture as evidence and towards computational inference. Across this landscape, generative synthetic processes including text-to-three-dimensional production methods create images and volumes assembled from statistical residue rather than sense data inputs, embedding the biases, omissions and extractive conditions of their training data. Simultaneously, imaging technologies remain entangled with militarised and surveillance-driven infrastructures that govern sensing, prediction and operational visibility, shaping how images act within systems rather than how they appear to human viewers. Collectively, we argue that these infrastructures demand a continual return to the question of what an image is now, as computational imaging systems mutate beyond inherited conceptual frameworks, and ontological assumptions about representation, evidence and agency.

Item Type:	Article
ISSN:	1477-2809
Subjects:	Art History & Theory Computing & Data Science Philosophy & Psychology Photography Research
Department:	Academy of Innovation and Research
Depositing User:	Tom Milnes
Date Deposited:	24 Sep 2026 15:56
Last Modified:	24 Sep 2026 15:56
URI:	https://repository.falmouth.ac.uk/id/eprint/6610

View Record (staff only)