Skin Deformation:

There are many beneficial reasons for studying the effects of skin deformation and many different ways to approach the problem. I think that there should be several different methods combined in order to give a final, useful, reproducible end result.

What have I done as of Dec. 19, 2003?

I've done some very rough and preliminary work with skin deformation motion capture using a small mesh of motion capture markers (46 3mm diameter markers) attached to the area of my left elbow. Some of the images are fuzzy, but they serve to show what I was doing. Marker placement was fairly arbitrary, and it's pretty obvious a few trips to the gym would do me a world of good.

I tried to place the markers on areas that I felt would move more than the surrounding areas (by flexing and moving my arm before attaching the markers in order to see what areas of my armed displaced the most), and some I used as anchor points (such as the elbow). Further study in this area would be very beneficial to getting a better marker placement scheme.

I was using a Vicon 8 motion capture system running build 121 of Vicon's Workstation software for collecting and reconstructing the data. I used an 8 camera (each camera has 1 million megapixel resolution) set up (we have 14 cameras at our disposal, but I felt 8 was enough for a test run). The cameras were setup up in a ring, four above my arm and four below in order to capture all the markers all the way around my arm.

It took a few different adjustments to fine tune the system to work correctly with the data. There were some technical issues that had to be addressed before I was getting satisfactory results. Also, once the markers were tracked, I had to spend a few hours helping the system to attach the correct marker name to the correct marker since the markers would become "lost" once in the fold of my arm and not visible to any camera. This is also a fairly bad way of deciding on which marker is which...sometimes I had to just guess and use my best judgment.

When my arm was straight out, the system did a good job of tracking the markers, even when flexing, etc. When I bent my forearm all the way, that's when the real problems start. Markers were occluded by the fold of my arm and therefore couldn't be reliably reconstructed or tracked. Because of this, I decided to make the data in those sections up with a direct linear interpolation (quick, but definitely not the best method). I used Kaydara's Motion Builder 5 software to examine and interpolate data. I then used Maya5 to build a 3D skin based off of the motion capture markers, attached the model's vertex points to the motion capture marker data, and created a few movies.

WARNING: The movies are long. The hash marks are where a motion capture marker was located.

As you can see from the movies, it's quite evident where the optical motion capture breaks down. Because of this, I'd like to use MRI (or some sort of useful imaging solution) data to recreate the sections that motion capture can not create. I'd like to do this with the same marker set for capture and MRI so as to have a 1 to 1 correlation on marker positions. The model I created was quite crude. Therefore, I'd also like to take the MRI data and use it to create a better 3D model of that section of the arm instead of just connecting the marker locations together in order to form a rudimentary skin.

In some of the motions I was trying to quickly move/shake my arm to see how the system would react, so in some places there are very jerky reactions, but this is from my own actions (although I'm not positive the system did a good job of tracking the markers in this situation). You can also see where some of the marker data "drifted" away from where it should have been...this is a combination of missing data as well as one of the limits of the systems ability to track non-rigid systems (doesn't happen in all cases, just every now and then).

From imaging information, we not only get helpful data in regards to what happens to the markers when occluded, but we can also extract bio mechanical data on the location and rotation of the radius and ulna in order to give us a better understanding of how those segments are moving.

We also gain a good deal of visualization data on where the muscle mass is, how's the connective tissue reacting, what happens to the skin and fatty layers when they are are in close contact, how do ligaments behave in this situation, etc. All of these can be used to help improve out understanding of this area of the body which can be used in bio mechanics as well as computer graphics in order to better visualize and animate this area of the anatomy.