Implementing bounding spheres
Bounding spheres as abstractions of higher detailed models can be used in different ways. One possible way is shown in Figure 8.9, where the sphere encloses the entire model. But as you can see in Figure 8.9, we would have a lot of empty space around the model, resulting in more false positives. Plus, we already have a method to do a broader check for possible collisions: the bounding boxes.
Instead, we will use the spheres to simplify the instances at a more detailed level by adding configurable bounding spheres to the model’s nodes. Even though we would have to now check several dozens of spheres against the spheres of another instance, we are still way below the computational power we would need to check every single triangle of both models against each other.
The number of false positives will also remain at an acceptable level. As stated broadly in the Simplifying the instances for faster collision checks section, choosing between...
