2026, three-part generative video installation
variable size
custom LED screens, LED curtains, flatscreen, used server hardware, webcams, power cables, network cables, metal mesh, power supplies, Raspberry Pi, computer, custom image generators, custom text generators, custom audio generators
_Pitch
This three-part installation — made of screens, cables, and computers — traces the quiet ways we move through algorithmic worlds. These networks don’t just carry information, they carry us: our ideas, our assumptions, our biases.
Live images from a webcam generate moving visuals in real time. Each screen runs its own AI model, not just to record what it sees, but to interpret and hallucinate. The work touches on something we’re all living through: the idea of a “programmed self.” Our identities are increasingly shaped by the algorithms we scroll, click, reply to and engage with every day. What you see and hear is raw, unpolished, full of the noise and glitches that are inherent to this kind of technology.
Human hubris — the belief that we can fully understand, control, or predict our reality — is not served here. Instead, the project gently questions it and offers a different path: open source, rooted in sharing, not just in code but in spirit. The generated images are made with non-commercial tools that are built to be open, not locked away.
Marlene Wenger
_Operating principle
For each of the three parts of the installations, a real time video stream of a webcam is feed into unprompted open source AI model, generating images that mimic the input images.
_Project description
The piece offers a glimpse into the many mirror-like manifestations of our digital webs. Composed of the artifacts of the hyper-scaled technological abundance and obsolescence, three distinct screen sculptures form a speculative infrastructure. They serve as portals into the nets that span in these spaces, materializing the mesh-like textures that compose our algorithmic environments.
The installation probes the diverse aesthetics and biases of our digital environments, shaped by repetitive engagement with computational entities. It is meandering through the statistical noise of unsupervised machine learning models by real-time interaction with the audience and the self-mirroring of the installation. This process shapes fleeting appearances out of the noise that make up the model’s compressed archives. The resulting images are rather dull, gray, and flat, standing in stark contrast to the bright, overly sharp images typically produced by this technology. It is an attempt to grasp the true aesthetics of our networks.
_Exhibitions
06/2026 Nuit Blanche – Le Consulat – Paris
03/2026 Art foundation PAX art award exhibition – HEK (Haus der Elektronischen Künste) – Basel
_Support
Developed thanks to the prize funds of the Art Foundation Pax.
