Agents

The agent group has found that using the DOP network is very slow. I have created a different method of setting up agent simulations that uses only SOP networks and it seems a lot faster (I have not tested the difference in speed yet). The way it works is that at each frame, you store (i.e. save in memory) some attributes for your geometry - usually this will be the position of the points (but may also include some other attributes if needed). These attributes can then be retrieved in the next frame, so that the agents can start from where they stopped in the previous frame.

Here are the nodes:

• phtj agents - has a 'store' and 'retrieve' nodes, for storing and retrieving attributes.

And here is an example file using these node:

• Example agents - some simple agent behaviours