In class, Z discussed several risks we might encounter with our physics engine, including risks in collision detection, epsilon issues, rolling, etc. Since she has already explained these risks, and essentially told us that we will need to find ways around them, we are not documenting them here. In other words, the following risks are specific to our particular mini-golf game:
Acquiring models of complex shapes
Risk: We want to have some obstacles on our course that use fairly complex shapes, such as a golf cart. We need to find an existing model -- it will probably be too difficult to design our own.
Alternative: We won't have these shapes on our course, but will find simpler shapes or other models to use instead.
Solution: We found a good model, although we had to convert it through several programs and reduce its complexity considerably to get it to work.
Rotating obstacles
Risk: We would like the golf cart obstacle to rotate, and don't know how easy this will be to implement.
Alternative: The golf cart won't rotate.
Solution: This was really easy -- OpenGL? has a built-in rotate function.
Implement bounding boxes
Risk: If we incorporate complex models, we will need to implement bounding boxes or else suffer a significant performance hit.
Solution: What we decided to do was create a separate class for shapes, which does not perform collision detection. We then created very simplified transparent models, which do implement collision detection.