Layered geometry and animation
polycounter lvl 11
So im really curious and confused, do animators know how to handle lots of overlapping strap geometry/belts/bags/pads/armor? It seems like itd be very annoying because wont the meshes intersect during animation?
Or do they expect you to retopo it into one giant lump of play-doe?
Or do they expect you to retopo it into one giant lump of play-doe?
Replies
Can a sleeve overlap the beginning of the arm, or should they be topo'd over?
It's not always easy to understand what will work and what won't without some experience of skinning. I'd recommend showing someone who can skin your geometry as you go to allow them to suggest changes if you plan on doing this.
When deciding whether those things (e.g. straps) should be layered geometry or cut into the base mesh -- I'd say it depends mainly on a combination of what looks best for that particular object (e.g. how tight it is to the body) and how much geometry you have to spare. Also, it's obviously not worth cutting stuff in if it is going to destroy the edge flow.
What I do know, is how they did it in the cinematic trailer for Assassins creed Revelations, because a couple of days ago there was a 3D & Animation conference here in Vienna and Digic (the ones who made the trailer) held a lecture there and explained the process.
All clothes of the characters (except the armor, obviously) were done with cloth simulations (they used Syflex for that). They had a cut-up mesh underneath the clothes that served as a collision mesh.. The animators had to "only" animate said collision mesh, and the rest was simulated.. The clothes were layed out in a way that would keep intersection problems to a minimum. There still were some intersections, even in the close-ups, but noone can notice them, if they dont watch it frame by frame, since there is a LOT of action going on...
A funny sidenote: If you watch the long version trailer frame by frame, in the fght scene, there is a sword and a javelin floating in the air with noone actually holding them, but you cant see it in normal speed, its pretty funny
For most accessories you shouldn't retopo it into one lump. However keeping the geometry a similar density to the geometry below is very important. If your accessories are lower poly then it'll be hard to avoid intersections.
Rigid objects on the hips and back are easy. Anything that needs to flex is always a little harder. Robes and skirts are always hard to work with.
It gets messier and more chaotic when the straps run diagonally...
(you don't even want to see an example of how ugly that can get).
"Ohh... I can remove the lower loop on the green box, I save 8 tris!"
"Ohh... The animator just kicked the model back to me and ask that I to add a loop, which will probably cause me to re-bake the normals, fun times... WHAT 4 U CURSE ME POLYGODS!"
As Monster said. much appreciated. I highly recommend that any character modeler have at least a bare minimum working understanding of rigging skin weighting. It REALLY helps everyone to be thinking about those things well in advance instead of after the fact.
I've bee wondering, what if the concept artist came up with an outfit that looks awesome, the modeller modelled it according to the concept and then when its time to start rigging it you find out that its really hard to rig and skin it in a way that looks natural, without skinning "artifacts" or intersections..
For instance a thick shirt strap that goes diagonally from the chest to the upper arm, over the arm pit and around the shoulder (if that makes any sense), then having the character raise its arm and strech it out chest-outward in an animation.
Second try to describe it :poly136:: A strap that is placed, sort of, over the area that gets bent, between the upper arm and the chest.
Do you, the riggers and/or animators amongst you, really just send it back, saying "nah, this ain't workin' ", or do you just deal with it, the best way you can? How does it work, where you are working at, or how does it normally work, respectively?
EDIT:
Like you said, basically:
Planning and consultation with the team before blazing forward wins in most cases.
In my experience the rigger can end up being asked just to skin something even if it's unlikely to work in order to prove it doesn't. It then may need to be sent back for geometry changes and then have the skinning fixed again. So basically bad planning can end up wasting everyone's time.
Nothing a quick conversation can't fix
We are more less going with the "deal with it" kind of thing with most rigging related topics. The good thing is that I'm always learning something with that
What works, what doesn't work, how to deal with certain things etc.