Exploring a Dark Museum

My project simulates walking through a dark room in a museum, full of statues. There is a spotlight to light the world that is the users "flashlight" (or it looks more like a headlamp). The program is runs a bit slow with screen capture on, but you can watch a demo below:

Some graphics technologies I used were: My camera control and texture mapping are fairly simple and similar to past assignments, so I only delineate below the collision detection and spotlight work that I did.

Collision Detection

Collision Detextion

The image above is an example of the camera colliding with one of the statues in the room. To implement collision detection, I calculated bounding spheres for each of the statues in the program's init() function. Because the meshes I used had a lot of vertices, it took a long time to calculate their normals and bounding spheres. To speed this up, I kept track of the minX, maxX, minY, maxY, minZ, and maxZ when loading the meshes in Shape.cpp, and I used that to calculate the bounding spheres. Instead of summing all the vertices and taking their average, I used the min and max values to calculate a center. Also, because the camera was anchored to a set y-value, I did not use the meshes' y values to calculate their bounding sphere's radius, which made the bounding sphere a more realistic size.

To avoid colliding with the walls in the room, I had previously attempted to calculate bounding boxes for the walls, but I eventually found it easier just to set a maxX, minX, maxZ, minZ restriction on the camera's location.

I didn't use any tutorials or websites to learn about collision detection because Dr. Wood briefly explained bounding spheres, and I managed to play around with it to make it work.



As you can see, in the image above, the lighting attenuates as it reaches the edges of the spotlight circle. It's actually not a circle; it's a cone (which is more evident when the light source's position is not the same as your eyepoint). In the glsl shaders, I calculated the vector (cone_vector) from the eyepoint to the "center" of the cone, and I calculated the vector from the eyepoint to the current vertex (view_vector). Then I calculated the angle between the cone_vector and the view_vector. The bigger the angle, the less light I put into the reflected color.

The hardest part was getting the spotlight to follow the camera. I had a bad bug where the spotlight would stay in the center of the room, as I walked around, so if you implement a moving spotlight, be sure to update the view_vector I just mentioned.

If you're interested in implementing directional lights, spotlights, or multiple lights, I found this tutorial to be useful.