Motion Capture - The Basics

Motion Capture has been a really interesting and efficient way to get realistic animation as long as you know how to manage and clean up the results.

The program Cortex is a very efficient way to clean up all those markers from the mocap data. Most likely all the cameras in the capture volume haven't caught all of the markers on the suit 100% of the time. That's where Cortex comes in to fill in the blanks. This is also where you want to name, organize, and link your markers. A unique feature in Cortex is that there are two different save files you work with, project files and track files. Projects are the environment or capture volume, This will save the number of cameras, volume size, and things of that nature. Track files are the marker data that you captured and edited, this is where you will see the motion (overall, you will probably have more track files than project files). Another important aspect to know about motion capture is that you want to get a t-pose of your capture subject to have a default pose to reference.

After finishing in Cortex it's time to go to MotionBuilder (Mobu). When first opening Mobu it'll ask you for a preferred control style. I chose Maya because that is the control style I'm used to. A lot of Mobu's interface is already reminiscent of Maya, both being Autodesk programs, in fact if you have used the human IK in Maya, a bit of Mobu may be familiar to you. The reason Mobu is a part of the motion capture pipeline is that you use it to further clean up the motion, assign geometry to the your capture markers, and export one simple file for use in Maya. Below is a moving car that was animated using motion capture markers on an RC car. Using a solid object is a good way to learn the programs and controls because you don't have to worry about the complexity that goes into, say a human, but you still use many of the techniques that are necessary for all types of capture subjects.

Speaking of a more complex capture subject, a human takes just a bit more extra steps than the car. When you create the marker links in Cortex for the car, there are only 5 markers to link to one another, which goes really fast, but with the human there are 49 markers, so you need to manage the links to create and efficient and easy to manage skeleton.

Another big difference is that when capturing the motions of the animation in the capture studio, you need to make an additional take or the range of motion (ROM) that covers all the range and possibilities that your character will make. This ROM also needs to be touched up in Cortex, just like the t-pose and final movement.

In Mobu the geometry takes a little bit more effort to bind than the car because certain sections of markers go to certain parts of the geometry. This was the final result in Maya.

The animation is far from perfect but it is a good base that can be further improved with a few strategies that I will cover in the next mocap update.