I am having difficulty figuring out how to architect a solution to my problem. I
am creating a program that simulates agents of various types -- these agents all
have small bitmaps associated with them for rendering purposes.

Currently, the agent base class handles all the image loading itself -- but this
causes a problem when the system is implemented as an applet, due to latency
etc. The solution for this of course is to use MediaTracker, but this causes
more problems...since I need a component to pass to the MediaTracker.

I prefer to keep the actual GUI code separate to the implementation, although
the render function is part of the agent class...I'm just having a hard time
piecing together the most efficient way of doing this.

Anybody done something like this recently, or have a good idea how it could be
done?