Description
The AABB surrounding_box() function assumes that the min point is strictly less than the max point (i.e. x_min < x_max, y_min < y_max, z_min < z_max). However, the hollow glass sphere example from Book 1 Section 10.5 breaks this assumption by having a negative radius that flips the bounding box coordinates. This is a result of the bounding_box() calculation for spheres.
Consider a 1-D example with two bounding ranges (min, max) of (0, 10) and (-5, 5). If you flip one so that it is (10, 0) and (-5, 5), the surrounding_box() calculation gives the incorrect result (-5, 5). It should be (-5, 10).
A reasonable fix is to make the ordering assumption clear for surrounding_box() and update bounding_box() for Sphere and MovingSphere.
bool sphere::bounding_box(double time0, double time1, aabb& output_box) const {
output_box = aabb(
center - vec3(radius, radius, radius),
center + vec3(radius, radius, radius));
return true;
}
should probably be something like:
bool sphere::bounding_box(double time0, double time1, aabb& output_box) const {
auto abs_r = abs(radius);
auto vec_r = vec3(abs_r, abs_r, abs_r);
output_box = aabb(
center - vec_r
center + vec_r);
return true;
}
A similar approach would be taken for MovingSphere.