Sense, turn, deposit
Three virtual sensors link each agent to the shared field. Local reinforcement builds bright pathways; diffusion softens their edges and evaporation removes abandoned routes.
How the network forms
Physarum polycephalum is a slime mould whose transport networks adapt as it searches for food. This explorer borrows that idea for an artistic agent model: each particle samples a trail ahead, left and right, steers toward the strongest signal, moves, and deposits a little more trail.
Change the behaviour
Long sensors connect distant strands. Narrow angles encourage sweeping streams. Wider turns can close loops and form fine meshes. Try a starting condition, then change one control and let it settle.
An artistic model
Colour maps trail density to a palette. These particles are a simulation device, not cells, and the result is not a biological prediction. The field wraps at its edges, joining paths across opposite sides.
Explore different behaviours
For background on material computation and Physarum-inspired network formation, see the Physarum research paper on arXiv. The live model here uses a simplified sensing and deposition rule, with visual parameters chosen for exploration.
Agents update in parallel with WebGL 2 transform feedback. Trail deposition, diffusion, decay and colour rendering also run on the GPU. No floating-point render-target extension is required.