Introduction
Ephemer(e)ality Capture: Glitch Practices in Photogrammetry details glitch practices using optical phenomena to disturb imaging algorithms. In this project, reflective, transparent, specular, and patterned/repetitive objects are used to confuse the imaging algorithm to produce spikes, holes, and glitches in the mesh and textures of the 3D objects.
The research tests the limits of photogrammetry in an effort toward new image-making methods. It builds upon ‘Ripping Reality: Blind spots and wrecked data in 3D’, in which Hito Steyerl (2012b) outlines the errors of 3D-scanning media in her work. Ephemer(e)ality Capture explores gaps in Steyerl’s approach, building upon investigations into 3D scanning’s ‘constructed imagery.’ It engages ‘fractional space’ more thoroughly through glitches caused by capturing optical phenomena.
The exposition centres on an initial research residency at JOYA: arte + ecología in Spain in 2018. During this period, a number of experiments within the landscape triggered the further development of glitch-based works. The exposition presents a diary of explorations and reflections that emerged when ‘reality captures’ were created of the landscape, and the inevitable issues of trying to image ephemera.
‘Reality capture’ or ‘Recap’ has become the expression used in the creative industries for the process of photogrammetry having emerged as an umbrella term for photogrammetric and lidar-based scanning technologies. Both technologies construct 3D models by finding similar recurring visual elements within inputted images. Photogrammetry (using photographs) and lidar (commonly using infra-red laser scans) assemble a vision of 3D space by reaching a consensus on the position of recurring points, imaged from multiple angles, and then processing this assemblage as a digital 3D model or ‘reality capture’.
Photogrammetry struggles to capture certain objects and environments due to their optical nature. Transparent, repetitive, patterned, indistinct, plain, reflective, and ephemeral objects cause problems. As briefly explained above, photogrammetry works by using quantifiable data – such as shapes, textures, or colours – to plot key points (‘cloud points’) in multiple images in order to construct how source objects might have been situated three-dimensionally and make a 3D model from them. Since ephemeral qualities make it harder for the technology to do that, the photogrammetry technology has to make things up; it gets things wrong when it estimates the forms of the objects based on the information it has. This leads to glitches and errors, such as stretched images and warped textures, as well as phantom forms or holes and spikes in the 3D mesh.
I have developed a series of artworks using these processes to explore mimetic visuality in 3D capturing technologies and issues that unpick their representational nature as being heavily mediated. As a result, this research is an assessment of the use of photogrammetry and proposes a methodology to expose the mediation of the technologies – a mediation by algorithms, which make decisions about 3D space based on 2D images. Through a series of artworks, I detail practical techniques for exploring the errors generated by the technology’s perceptions of space. The artworks become a critical investigation of visual capturing technologies by forcing them to visualise the invisible, temporal, and ephemeral. This is not meant in a metaphorical or figurative sense; the technology is literally being used to visualise objects that are transparent or that change during the process of recording them. The deliberate negation of technical instructions forces the technology to speculate, and from these speculations, we get a glimpse of the technology’s decisions and how it constructs the images. This reveals how 3D media imagery is mediated and constructed using automation, although this is often disguised or unclear until the point of error. Errors become useful in understanding the workings of a technology that are otherwise inaccessible. Presented is a methodology that encourages a détournement of automation, exploring the agency of technology through errors in models of transparent, plain, reflective, and ephemeral objects and environments. The works probe at the limitations of algorithmic understanding and force the technology to visualise uncertainty. The research promotes self-reflexivity and critical reflection in users to deliberately side-step, avoid, or actively ignore prescribed workflows of digital tools through the use of a dynamic methodology.