You are viewing the course site for a past offering of this course. The current offering may be found here.
Lecture 9: Raytracing (46)
julialuo

If we optimize to cache data about intersecting a particular object that spans multiple grids, wouldn't this scene be okay? We could pick a grid size that errs on the small side (matching the detail on the table) and then for the larger objects rely on caching to avoid having to do unnecessary intersections.

jgforsberg

If the uniform grid was small and the image big, then I would worry about running out of room in the cache. The majority of this scene should cover multiple grid squares. I'm confused on how you could cache the ray intersections to save time as each ray has a different (origin, direction) pair.

moridin22

@julialuo Another important probem with using a small grid size is that for all of the empty space, we are traversing through tons of empty grid squares that don't have any objects in them, which I think is the main cost of having grids that are too fine relative to the image.

dtseng

What do we do in games, when the scene is generally changing all the time? For example, what about a transition when a character is halfway between two rooms, or if there's a window that shows outside world? How would a cache work in that case?

You must be enrolled in the course to comment