A developer extended the cellular Potts model with directional bias and volume-loss rules to simulate cells competing like gladiators, then made it playable.
Key Takeaways
The cellular Potts model builds on the Ising model: grid sites flip stochastically based on energy terms enforcing neighbor similarity and target cell volume.
Combat emerges from a third energy term biasing a cell’s expansion toward its nearest neighbor, with target volume shrinking on site loss as a hit-point mechanic.
No top-down behavioral rules are used; purposeful motion emerges entirely from low-level probabilistic updates, mirroring biological emergence arguments.
Player control replaces the nearest-neighbor bias with keyboard-input bias, keeping interaction at the rule level rather than issuing direct movement commands.
The author extended the demo to a two-player game with bullets, and source code is publicly available.
Hacker News Comment Review
No substantive HN discussion yet.
Notable Comments
@chychiu: Points to alien, an open-source project focused on emergent artificial-cell dynamics, as a related prior art reference.
