Methods of Animating Cloth?

#1 Personally, I'd model the poncho with "strips" of polys all the way around... by that I mean that I'd create vertical strips of interconnected faces and build a rather complex skeletal structure attaching joints to every intersection. (Think in quads here, not tri's).

Once that's done, it's a matter of animating the poncho to mimic the character animations.

Is it the best method? Probably not, but then I don't work with cloth type materials much.

#2 I would do something similair... model the poncho with strips of verts and tris... then create a node half way, and all the way down each strip... Then animate the strips with the character movement.. It will take a damn fine animator.

#3 Nah - just a really patient one. =P

#4 Hmmm. One of the issues I have with doing the "pinetree" (strips of bones originating at about the neck, going down the length of the cloak) of bones is the overwhelming # of bones & joints. I mean, I'm thinking I'd be lucky to get away with using 20 bones just for the cloth!

This is the most obvious method, but I was wondering if someone had any other ideas.

#5 Well, you dont really have ,much of a choice if you want it to look right...

You could always just leave it static... About the only way I think you will limit the number of bones.

To put it into perspecitve... I am making a game, and the rig for my character alone is 30 bones... not to mention the 46 bones that will be mounted to the rig.

#6 The actual number of joints need not be overwhelming, at least not unless you're throwing too many poly's into the mesh in the first place, which I assume you aren't planning to do based on the statement that 100's could be running around.

Let's say that you model the poncho from a simply cylinder... let's say 12 sides and 6 stacks (random guesses for arguments sake).

If you originate you skeletal tree from the neck, you're only really looking at 60 joints (assuming 1 stack gets used for the "layered" effect in the image.

Is that alot? Yes it is. Is it "overwhelming"? I suppose that's a matter of opinion, but I've worked with more. \C:

#7 How much of an impact will all those joints have on the speed of the game? lets say we have 200 of those guys running around. Does it create a huge amount of slowdown?

Here is another way I was thinking:
What if I only had 1 chain of bones going down the center, and simply moved and scaled that? I could create diferent shapes & proportions with the scaling (on different axis) of each joint and move them as well. won't be as precise, but it might work. has anyone tried that?

#8 @Anthony... You would have to try it and see... you'll have a bigger problem with poly counts... You won't have very many joints, because you cant have that many polys.

20 bones is NOT a lot of bones at all...

#9 Cant you just fake it. No bones. All animation?


Surge

#10 It is generally faster (performance wise) and easier to animate bones then it is to animate every vertex, if I am not mistaken.

#11 My programmer seems to think the opposite= frame-by-fram vertex animation is faster. just makes the file size bigger.

#12 AFAIK it's not the sheer number of joints and bones that you need to worry about, it's polys.

Also, Chris is right - it's way more costly in terms of performance to animate by vertex than it is to use a skeletal system.

The only other way I can think fo the get the effect you're looking for would be to build a cloth simulation algorithm and add it to the engine - you think bones would hit your performance!

#13 Vertex meaning frame-by-frame or tweening vertex? saved frames seem like they would be faster.

Anyways, here is what I was describing with the single chain:


but you all seem to think many bones is not a problem, so I might as well go with that.

#14 Well, when I say per vertex, I'm talking keyframe (frame by frame).

The thing is, it may execute faster, but eats much more memory... and that get's to be problematic pretty fast.

The idea you've laid out is an interesting one. I'd be curious to see the results of this if you decided to rig it up and test it out...