I wrote this article explaining algorithms for physarum-style simulations bleuje.com/physarum-exp...

My impl basically always used constant step size/velocity - adding momentum is a cool extension, I need to try that

please try stuff and report results :)

forgot how much fun this was - I tried using larger values for the sense distance and got some cool results... I want to try moving the parameters onto the agents so that there can be distributions of parameter values in play

nice, how are you doing the deposit? drawn dots, or atomic counters like me? Also that's a large resolution you've got!

Atomic counters.. I liked that you just wrote 1’s during the deposit phase and took a square root, I think some remapping like this might help emphasize contrast - I’m actually doing everything on the uint buffers, I wonder if there’s significant benefit to doing the blur step with floats