For real, one of the things I love most in computer animation (especially real-time simulations) is cloth physics.

Dream of the day when a Sims-like gives me all the beauty of properly-rendered capes and other flowing fabric. I live for it.

Point Sampling is used to define physics collisions in games sometimes, and this video shows some of the drawbacks to point sampling.

There's a part where it looks like sharp pieces of an object are piercing a thin plane. But they're not; what's happening is more complex.

Basically, the collision simulation, in order to have to do less work and therefore compute faster, it's only looking at the plane in question as a few dots with space in between, that space only vaguely estimated by the simulation.

When the bits pierce the plane, what's happened is the simulation just kinda lost track of the geometry that got the piercing for a second. Then, when it caught back up and realized what happened, it had two objects intersecting, but its math renders intersecting objects immobile

You see this happen in video games all the time, any time you're playing Fallout 3 and you try really hard to get a book to intersect with your dad and do something weird.

Clothes in games aren't usually separate objects from the character model. Why would they be?

Well, in some instances, and we see this especially with capes, developers and audiences alike love to see a good cloth physics simulation. Capes are fun, in real life too! Try one out, let's bring back capes!

But, because of PS and other methods...

The video at the top of this thread shows off a new, faster, more complex collision simulation method. It blows Point Sampling out of the water.

Man, computer scientists are cool, and way smarter than I'll ever be