Fish (cyan = highly aligned) Vision cone (highlighted fish)
⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.

Vision-Based Alignment in Fish Schools

Most schooling simulations give every fish an omnidirectional sensor: average whatever is inside a radius and steer toward it. Real fish align using eyes, and eyes have a limited field of view with a blind sector directly behind, plus the mundane fact that a nearby fish can block the line of sight to one further away. This simulator builds a school where alignment and cohesion are driven purely by that visual pipeline — field-of-view culling, distance-ordered occlusion, and visual-angle weighting so closer, larger-looking neighbors matter more — while short-range collision avoidance is modeled separately as omnidirectional lateral-line sensing, matching how these two senses actually differ in real fish. Widen the blind zone, shrink the school, or startle it and watch the live polarization order parameter track how well-organized the group's motion really is.