Efficient Rendering

There are many aspects to efficient real-time rendering of interesting scenes. We’ll be discussing just two of them: level of detail control and coherent mesh representation.

Required reading

First lecture

Of interest to us today is adapting the resolution of a mesh so that we don’t use more polygons than necessary to paint the pixels that it covers. It turns out that the work we’ve already covered on simplification and progressive meshes provides a very convenient way of addressing this problem. Iterative edge contraction induces a hierarchical structure on triangulated surfaces. This paper shows how to use this structure to extract view-dependent approximations of mesh geometry.

Second lecture

Recommended reading

There’s actually a whole book on level of detail control: Level of Detail for 3D Graphics. You can find out about it at http://www.lodbook.com/.

There’s also an entire book on Real-Time Rendering — see http://www.realtimerendering.com/.

Several other view-dependent refinement methods have been published. Here are a couple other examples that were published contemporaneously with Hoppe’s paper above.

More recently, the emphasis in this kind of work has been on massive models. Here’s one recent example.

The work on ordering meshes is closely related to the problem of finding good triangle strip decompositions of a mesh. Here’s some work on that problem: