× Vision in a 3D room One model combines light, adaptation, available colour and directional motion.
Geometry. A closed kitchen with walls, floor, ceiling, a light fixture, counter, window ledge, table and solid objects. Each ray finds the nearest surface; objects occlude the background. Textures come from the kitchen photograph and current window view. Dimensions, hidden surfaces and the light fixture are approximate, rather than measured reconstructions.
Eyes and optics. 857 and 852 measured viewing directions from a female fruit fly. Rear and upward views move with the fly; dark edges mark the eye-map boundaries. The 8.23° FWHM angular profile comes from nearly dark-adapted R1–R6. This is a fixed empirical aperture, not a complete six-receptor neural-superposition model.
Processing. Linear RGB provides a relative light signal. Fast filtering and slow adaptation preserve static structure. Delayed correlation between neighbouring directions retains motion direction; ON/OFF branches are calculated separately. Brightness and motion feed the existing brain's two visual channels. Scene depth is not supplied to the brain.
Colour. A fixed pale/yellow mosaic of approximately 30:70 provides one RGB-based R8 proxy per column, either blue or green. R7 UV, polarization and absolute photon counts are unavailable from RGB. These signals are not synthesized; Poisson photon noise is disabled.
Based on the supplied “Drosophila vision research and simulator” review, pages 21–28. Profile: closed-room-relative-vision-v2. Room dimensions, delays and gains are engineering parameters, not fitted physiology. Neural substeps up to 1 ms and display refresh up to 30 Hz describe software, not a fly's biological FPS. The image is a human-readable visualization of a sensory model.
Measured eye map · Gonzalez-Bellido et al., 2011 .