1 / 6
‘Lens’ is an interactive audiovisual installation that organises the Earth's surface according to an algorithmic logic of visual similarity rather than geographic location. Satellite data covering the full range of our planet’s terrain has been scraped and rendered into imagery directly from Landsat’s numerical remote sensing archives, a novel form of photography where the vast apparatus of planetary computation is the aperture, and DIY homemade python scripts are the shutter.
The resulting corpus of 50k+ images was then analysed using a custom visual embedding pipeline, producing a 379-dimensional vector encoding each image’s visual characteristics. These vectors were projected into two dimensions and sorted into a spatially coherent space of visual proximity which is freely navigable via motion capture hand gestures.
A large central ‘portal’ displays the current Landsat tile at high resolution, producing a stroboscopic flicker as the surrounding vector space of satellite images is moved through. When left dormant, the system engages a drift mode, navigating autonomously.
‘Lens’ asks in a straightforward, sensuous, and embodied way, what it means to move through and perceive the Earth as an algorithm. It hacks the extractive, all-seeing apparatus of global capitalism’s entanglement with science and the military-industrial complex, and turns it on itself. Untethered from the geographic grid, it surfaces connectivities, flows and symbioses in place of the ground level divisions, limits and boundaries these technologies are built to enforce.
But the gaze is not innocent. The same apparatus of Earth observation that discloses these connections is thermodynamically bound to their exhaustion. As the climate crisis advances, so does the drive for more monitoring, which means more material extraction, more energy use, & more infrastructure, all directly contributing the acceleration of the collapse it observes.
The work offers that the systems and structures of big data can be briefly inhabited in unintended ways, and made to disclose their structural complicities instead of hiding them.
Damien Roach
Damien Roach is a London-based artist, designer, musician, and lecturer. His projects span art, design/creative direction, publishing, sound/music and audiovisual. His RCA PhD ‘Acid Realism’ interrogates AI, machine vision, and non-human planetarity. In 2023 as patten, he released the first album made from text-to-audio AI samples. Recent projects include ‘Grounding’ at SXSW London, ‘SEED’ commissioned by Daata, immersive AV performances at ICA and Tate Modern, curating the cross-platform exhibition ‘Real Time’, and design from Caribou to Disney. He has exhibited internationally, including the 51st Venice Biennale, ‘Learn to Read’ at Tate Modern, and ‘Art Now’ at Tate Britain.
