Multi resolution character setups

Introduction

Character rigs in production can become very complex these days, muscle, cloth, hair, hard body and soft body dynamics are used in all levels of 3D productions. The expectations of what can be achieved is being raised for ever block buster movie that comes out. Just have a look at King Kong or Lord of the Rings and you can see just how far it can be pushed. For most of us this is only a dream to be able to work on movies like this but the expectation is still there for what ever project you might be on. Even the game market that was once considered to be limited in what could be achieved has broken new ground with the next generation game consoles and PC video cards.

We now have to be able to do it all without excuses about there not being enough money or time, it is just expected.

Multi Level Rigs

With all that is expected rigs can get very complex when handling the needs and wants of the directors. One of the ways of reducing complication at any one point in the production pipe line is to use multi resolution rigs. What this means is you have a set up for animators, one for dynamics and yet another for rendering. Each one leads into the next and keeps any unnecessary clutter out of the way of the current level of production.

Below I describe a couple of possible ways to handle multi resolution rigs in 3DS Max 8 but these methods are the same in most animation packages just the names of the tools may vary.

All In One
Lets have a look at typical broadcast level characters for say a cartoon look that doesn't have supper high poly counts of complexity first. I'm assuming that we don't have major dynamics simulations to take care of like clothing, maybe just some simple hair or loose bits.

In the case above I would most likely have three levels of character all following one rig. I would have a cut up version of the mesh linked to the animation rig so that the animators could have a really fast update in the view ports. This low res might even be optimized a bit just so it was as fast as possible. The second level would be the skinned mesh with TurboSmooth, Skin Morph and other tools that are being used for deformations turned off in view port so they are not being calculated. The third level is the render level character that would have all the deformation modifiers turned on pre-render and then tessellated up using TurboSmooth or other means.

The reason that I would keep them all in one rig is to just simplify the production work flow and not have to switch the high and low res out and have to manage the extra assets. Simpler is always better and you shouldn't jump to complicate things for no reason.
Multiple rigs

At the other end of the spectrum where you have feature film characters that are very complex with lots of high detail, clothing, hair and other bits that would just kill the interactivity for the animators I would go with multiple rigs.

Three levels of rigs would be generated from the previous, animation rig, deformation rig and dynamics and render rig.

Animation Rig:
The animation rig would not have all the muscle bones, twist bones and other deformation rig parts and only have a chopped up mesh linked to it. Animators can now work with this very quickly without slow downs. The animation would be saved off this rig using the save animation feature and then loaded on the high res rig.

Deformation Rig:
The high res rig could be done in one of two ways, either the same rig as low res but with all the extra deformation rig bits added or just an FK bone system with all the deformation bits added. The second choice removes complexity but takes some of the flexibility away as you really can't edit the animation on this rig as well. The high res rig would have all the high res skinned, skin wrapped and tessellation modifiers on it.

Another possible way of dealing with the deformation rig is to have it set up on an FK system without controls but instead of loading animation onto it you merge the rig into the animation file and use constraints to constrain the rig to the animated rig. This gives you the ability to see the high res set up on the deforming mesh and still be able to tweak the animation if need be.

Dynamics / Render Rig:
Once deformation rig is set up and it is approved you need to start dealing with sims for cloth, hair, hard bodies and others. The best way to handle this is to remove the complexity of the rig by Using the Point Cache modifier or point cache 2 by John Burrnett so that the animation is baked into the mesh. It is also best if you have a low poly cage baked as well for the collision object for the cloth and other sims, this will reduce the simulation times a bit. With the rig removed you can now do all the sims on a fast deforming mesh that doesn't have the over head of calculating deformations, the result is then sent to render.

Sounds like a lot of work but most of the processes can be automated with Max Script so that each step is just a button click away. Obviously any high end feature production is going to have some high end TD's on board that would be setting up the pipe line for the artists.

Motion Capture Rigs

Motion capture is used in many productions these days, Andy Sirkis, a Shakespearian actor dawned a motion capture suite for Gullum in Load of the Rings and King Kong. I think that it is safe to say we will be seeing a lot more motion capture used in coming years. Video game titles are using motion capture as it speeds up the animation process and allows for more realism when playing the games.

Character Studio and Motion Builder are two of the primary tools for dealing with motion capture but many of us still prefer to handle it with custom rigs based either on need or preference. Converting the raw motion capture data might still be best done in one of the above packages and then imported into Max, Maya or XSI. Importing them into a package will give you an FK bone driven system but this is next to impossible to animated so the animation needs to be baked onto an animation rig. This might require some simple scripting and a bit of forethought on how best to achieve this in any given package but it is possible to bake rotational motion capture data onto an IK driven rig as well as the FK driven rig in a blended system. The IK and FK portions of the animation rig then drive and FK bones system that the character is skinned to and that is what gets exported to the game engine or Save/Load animation to a high res character for feature films.

Disclaimer
There are other work flows and by no means is this the only root that can be taken but they do give you a few things to think about. For each piece of software there are optional tools that might not exist in others or work flows that will be better or worse depending on how well the tools work in any given package. For instance Maya has a reasonably robust file referencing system where Max doesn't at this point so you could use that work flow in Maya where saving and loading animation would be better in Max.