Hyper‑Náttura Beta ← All benches
Bench 02 · Models

BeeVision

A set of custom models trained toward pollinator visual perception. They take environmental imagery and put it through a position built from pollinator research: ultraviolet‑informed colour translation, and a star‑field standing where the compound eye would be.

Details · BeeVision series in preparation
The work

BeeVision began in the antique slide cabinet. New Observations on the History of Bees, started in 2021, worked through nineteenth‑century microscope preparations of pollinator anatomy, and the models grew out of that looking.

Most attempts to picture compound vision reach for a pixel mosaic, a grid of repeated tiles standing in for the ommatidia. BeeVision takes another route and uses a star‑field, which treats the eye as a field of separate points of attention held in one dark continuum. The result is docu‑fictional, honest about being a construction, and closer to how a distributed sensorium might actually hold a scene together.

What the models process is environmental imagery, photographs of flowers and foliage made in the field. The research on pollinator perception and neonicotinoid effects sits underneath as the position from which the translation is made.

The position

Three decisions carried by every image the models produce.

Ultraviolet

Colour translated toward the range a pollinator reads, which brings out the nectar guides and markings that stay invisible to us.

Star‑field

Ridge‑traced vein star‑chains and a categorical dot palette, blue‑white through cyan and violet with rare warm points, set against the pixel‑mosaic convention.

Ground

A continuous near‑black field holding sharp foliage as dim constellations, with pointillist density and quiet capped anchors.

Lineage

2021 New Observations on the History of Bees begins, working from antique microscope slides
Apr–May 2025 BioViscous and BioGlass trained with Alias Studio
Sept–Oct 2025 Alias Studio × Serpentine Future Art Ecosystems. Lead series across Serpentine's channels, three works acquired, presented across two Arts & Technologies community sessions
2026 Models in retraining, with the perceptual grammar settled at version four
[DATE] [NEXT ENTRY]

Record

Published material, open to read.