Watch ‘em go! This simulation is part of the ‘Tube’ project.
snot nose writes:
Crowd simulation from the open source film project ‘tube’. I wrote the AI using Python and Blender’s game engine (and also had to do some manual editing).
The bone animation was done using autowalk script, which was re-written and rigged by Bassam Kurdali (Director and Technical Director of Tube), based on the 2.49 ant rig by MCHammond. The Frontal and background animation by Karen Webb.
Still Wip, some roaches will be deleted for the sake of composition, some adjusted. We are going to export their movement into curves and then apply it as a modifier over their main body. The antennas will probably use the new noise modifer in blender’s curve editor. And some roaches will be stepped and crunched…
I used logic bricks for the AI and python mostly for conditions. I will post later the final script and also a small tut that explains it. But as I finished to work on it, blender was just releasing new features in the game engine – the steering actuator, navigation mesh, obstacle avoidance and basically most of the stuff i was already working on. so now it should be actually much easier to setup the AI in blender, and probably would be good to combine some options from my script into it.
Unfortunately, the main problem with using the game engine as a crowd simulator, is that the recorder can’t record bone armatures, just primitves (that were used for the AI). But bassam did great work on the autowalk script and created also an interface that give a lot of control over the roaches.