Newer movies (still not finished...now I need to smooth out the mocap a little better to make it less shaky, and put the little creature into an environment, but here are the links...movie1 and movie2.)
Using some various materials lying around the mocap studio, I decided to create a creature that could utilize mocap, but not in the traditional sense of mapping human motion to a human character. I believe that more of a puppeteering approach works best for what I'm working on. I'm not a trained animator, so using mocap to drive my character is easier for me and provides believable results, at the same time being fairly fast and easy to change if I need to.
Here are the most recent results using a diferent model and a few more markers. Performance animation was myself and Eric Camper. Below these images are some information on the process and a look at the attachments that were being worn/controlled as well as an early movie of just trying to get everything to work. All of the mocap is raw out of MotionBuilder and hasn't been filtered. All are still works in process.
There are quitre a few things to be finished. The binding is quick and dirty. I plan to change the bars coming out of its mouth to look more intricate, the swinging cones will definitely change, clusters are going to be added around the mouth so it moves with the cylinders sticking out of the mouth, the legs have to be redisgned because there isn't enough travel in them, I may enlarge the front, I have to add the blink controls (captured, but not applied yet), and then massage the mocap data to get rid of shakiness and popping.
1st, Figure out what you want to do. I had an original thought of a bird that ran around with a pole across it's back and water jugs at each end of the pole. It would just run around a town, giving water to people. I thought the weight of the water and the running would give some very nice motion.
2nd, Figure out how you're going to move your chracter. Now I had an idea, but I usually do things in three or four stages instead of doing everything all at once. I try to make small leaps to see how things will work out instead of jumping head long into a project without trying to figure out what I'm going to do. So the first part was the easiest. Make the pole with the water jugs on it and then move around to see what type of weight transfer I get. This was easily done and the results are below. Sometimes the bottles ran into my leg, so you can observe how they cause the bottles to suddenly change course and causes the motion to look jerky. There is little to know cleaning of the mocap data, so the character over all may have some jitter to it.
3rd. Develop the character and create a story. Now that that was done, I thought about the character a little more. I thought it would be more interesting to have some type of creature and not just a bird, so I sketched out a blobby character. After looking at it, I realized that he had no legs, and I didn't want to give him traditional legs. I thought of ow I could get floor impact and motion without a complicated leg rig, so I thought the easiest thing to do would be to go with the simple approach. One point on each foot and then use this to drive a sort of "shock absorber" or pogo stick effect for the legs.
Now I need a story(I'd normally say create a story first, but the creatures and people in a story seem to pop into my head first and then a story develops out of the characters.). Since the little blob is carrying water around, but has no arms, he tries to swing the water to himself, but can't make it. He's eventually helped be another little creature in an act of kindness. The other creature has arms, but no legs...maybe? Could be interesting to see one trying to get on top of the other, maybe the get some sort of weird saddle? I'm obviously developing the story as I go along....need to put more thought into this. Setting in a western town or a middle eastern bizarre. Either one would possible have saddles for sale and have them hanging around outside.
4th. Put all the pieces together. I'd also drawn the character with eyes on stalks and wished to make these much larger and control these myself. I took a short trip to the Dollar Tree and Big Lots and found the inexpensive things I was looking for. I picked up a few flashlights for eyes (the ones with the rotating head) and a few small clamp on reading lights for eye blinking. If I really had a budget for this, I could have done much better for eyes, but since I'm trying to do this with spending as little as possible, this will have to do for now.
5th. Final performance. This is the performance that fleshes out the story I want to tell. This is the majority of the moves I want to mocap with this character, but there may be more characters involved in the story. Since the character has no arms, some of the story rideas around the character not being able to fet
6th. Computer Graphics. Now it's time to put all the pieces together. There has to be the character, but how do the different motion capture parts actually animated the character? In the previous stages, the majority of this was figured out, doing tests on locomotion, bending, blending, etc. Now it's fairly easy to move the character and have that character dropped into some sort of CG setting.
I've found that when taking this approach, that it's much easier to write a master script in Maya's MEL language that rigs the entire chracter each time. Since there are several different takes of the same character, but no clear pipeline on how to apply the points to the character and in what order, I think scripting is the best way. It gives a reliable and consistent way to attach the markers to my character.
1. create rig
2. test rig and collect data
3. clean mocap with Workstation, bring into MotionBuilder for further cleaning, import point data into Maya
4. create models so they're the same size and fit the motion data (I know this is rather backwards in regards to most production, but it's easier from my frame of reference although doing it the other way around is also possible, it's just harder to get the right results)
5. use mocap data to drive the character through sets of constraints and other methods such as driving blendshapes, etc. At the end of the day, mocap is just a bunch of input points that can correctly or incorrectly drive componenets of a CG character.
6. look at the results and see what needs to be changed.
8. if the results are good, start creating a script to automatically rig your character. Since it's mostly a set of constriants, etc., the scripts can be fairly simples and straingt forward, setting up aim, position, and possibly orientation constraints. If you're using mocap to drive blend shapes or change variables, these are also easily scripted out.
9. look atthe results again. If you like it great, if not, recapture and re-attach points. I find that I usually need 3 captures. The first to get everything started, the second to get a good idea of what i want and work the kinks out, and the third to fine tune perfromance, timing, etc.