Optimizing a 3D Model For Animation

When you are building high-detail geometry, chances are you will be viewing it from different angles as well as for different lengths of time in your animation. Because high-detail models can get a bit unwieldy during complex sequences, you might try to evaluate what is absolutely critical for the scene to pass without the viewers…

When you are building high-detail geometry, chances are you will be viewing it from different angles as well as for different lengths of time in your animation. Because high-detail models can get a bit unwieldy during complex sequences, you might try to evaluate what is absolutely critical for the scene to pass without the viewers noticing.

For instance, There is no need to model people in a stadium if you are not going to get close to every one of them. Even if you are, you can swap low-detail versus high-detail as you get closer or farther away. These are not really cheats, but rather optimizations.

There is no need to overload the computer with a highly detailed model if you are viewing only one section of it for a particular shot. Studios do not build entire buildings just so they can shoot one room on one floor. To do such a thing would be just plain impractical. Even though a computer can theoretically handle an entire high-detail model, there has no practical reason to do it. When you are working with an animated shot, you will find yourself being able to take advantage of some of these optimizations.

The first step in working with high-detail animated scenes is determining which parts of it are actually animated. For instance, is there an asteroid somewhere in the scene that comes hurtling past the camera? Is there a character that is in the scene for a second or two?

Run through the shot to see which parts of it are animated in the camera view. This will help you figure out which areas of the scene you can optimize. Consider some possibilities for optimizing your scene:

To go back to the asteroid example, think about the asteroids themselves. Are they all individual objects? If so, think about using a particle system instead. Not only can you control an asteroid field more easily this way, you can also randomize the animation much better. Less individual objects means less transform data that your 3D Program has to keep track of and, consequently, less resource usage.

If you are animating several of the same pieces of geometry, are you using the Scatter compound object? This is a much more efficient way to model andimate if your scene contains several of the same object. The Scatter compound object even has the capability to display proxy geometry instead of the actual mesh.

This can be a huge time saver.