Skip to content
blendshelf
Geometry Nodes
Physarum - Slime Mold Simulation (Blender Geometry-Nodes)
Geometry Nodes

Physarum - Slime Mold Simulation (Blender Geometry-Nodes)

by Alex Martinelli

Geometry Nodes setup for Blender 5.0 by Alex Martinelli.

Free / pay what you want

Get it on Gumroad

Sold by Alex Martinelli. Check current pricing, license and Blender compatibility on Gumroad before purchasing.

Type
Geometry Nodes setup
Blender version
5.0
Last checked
14 Nov 2026

From the creator’s description

A Blender file with Geometry-Nodes (GN) modifiers to run and visualize Physarum Transport Networks (aka slime mold). Requires Blender >=5.0. Have questions or want to share your results? Reach out on Twitter/X or Threads. The setup allows simulation of Physarum Transport Networks (more details). Physarum polycephalum is a slime mold that presents complex emergent behavior, like formation of intricate patterns via simple rules of perception and reaction. We provide a model were a "population" of entities perceive signals from the environment, change direction based on such signals and in turn leave trails to affect the environment. This feedback loop creates the emergent behavior, which can be controlled via the exposed sim parameters. Even small changes in such parameters can lead to a wide variety of complex organic patterns and behaviors. Summary of the Setup In the provided file you can find a "physarum" object with two modifiers slots: a "Run Sim" for the base simulation and a "render" for rendering the results. You can start playing with these and get results right away! Notice that some functionalities are available only for the premium version; check the table below for all details. The simulation is run using points ("population") as entities that sense and leave traces in the trail, modeled via a Volume Grid. For the theory behind it, see here. The simulation is controlled via multiple parameters, the most important of which have been grouped in a bundle. For the premium version, multiple presets are also provided. Otherwise you can simply play with each param individually and run the simulation to see the results. While the "Run Sim" modifier will return the raw results of the simulation, in term of points and volume, we also provide some utilities to render the results. "Physarum Renderer" allow better control on how to visualize the results, focusing on the various components, like points, volume and paths. The sim resolution depends on the resolution of the target Volume Grid, controlled via the "Num Voxels" param. A higher number of voxels (aka higher resolution) implies more details but requires more computation. Similarly, a bigger population (controlled via "Points Density") allows for richer results but will be slower to compute. "Sim Substeps" controls how many simulation steps are run in a frame. The higher, and the faster the simulation will run (without losing on accuracy), while requiring more time to be computed. Results vary significantly even with minor parameter adjustments (e.g. +/-0.1). Many configurations may not produce compelling results, so experimentation through trial and error is essential. Also it might take some time for the initial random distribution of points to converge to something interesting, so experiment first with low resolution and higher substeps. Here a short summary of the sim params: Horizon Walk Fact: how far an entity can move Angle Sense: the angle that an entity can span from the current direction to sense the trail Direction Sense Scale: how far an entity can sense Angle Walk Fact: how much an entity can rotate when changing direction Trace Strength: the strength of the signal left on the trail by each entity Extra content for Premium version The free version allows to run the simulation only on a 2D plane, while in the premium version you will be able to run it in 3D inside any arbitrary mesh. Some presets for the initial mesh are also provided. Additionally in the premium version you will find multiple simulation params presets, option to extrude the entities movements along a path, texture/image conditioning examples, plus a richer set of control for the 3D rendering. You can also control when to add or delete points to the population. Technical Considerations We implemented the model via a "population" of points that read (sense) and write (deposit) data to a Volume Grid (trail). Additionally the trail is also subject to a diffuse and a decay step. The core logic is that each entity senses the trail based on the sim params and shift its direction towards the trail point where the accumulated signal is stronger. For now we had to implement this step from the pov of the trail, where each voxel looks at the closest entry in the population and increases the value if close enough. It would make more sense to do the other way around, where each entity leaves a trace in the closest voxel. This way, multiple entities resting in the same spot would all contribute to the trail, while the current implementation gets at most one contribution. However Geometry-Nodes currently poses a limitation for the latter implementation option.

More geometry nodes