Quadruped Study: Mocapping a Dog

How I got to work with dogs over the Summer

At Drexel University, students usually don't have a free Summer after starting Sophomore year. This being the case, I still wanted to do something fun and memorable aside from the mandatory courses. I convinced a professor in my department to advise an independent study that looked at researching and documenting quadruped motion capture; specifically mocapping a dog.

Image of a study of canine anatomy compared to Cat Soto's marker placement

I started out with searching if anyone had done this before, as well as reached out to a few alumni that may have encounter similar challenges. It's easy enough to find pictures of dogs adorably in custom mocap suits, but it's a bit harder to track down the methods and theory behind it. The closest I got to documentation of such experiments were from a SIGGRAPH talk about creating a labeling and solving set up in Blade for a horse and from Cat Soto, who did a similar project to mine.

(Top) Labeling setup, (Bottom) Solving setup

By watching the tutorial and replicating the bone structure of a canine, I was able to create a functioning template which to test on.

By this point I had contacted the handler for the Drexel Therapy Dogs. The two dogs I worked with, Chai and Expresso, are Cane Corsos. I conducted two shoots with them: the first to test the template, the second to work out the kinks



Here's a video from the midpoint of my research:

Here's a video from the end of the Summer


Ultimately, I don't feel that I completed the project. The marker data I believe is correct, but I don't feel the solved work was successful in achieving the quality of data I was hoping for.

What I aim to do is to use the knowledge I learned to develop my problem solving skills better, and once I've learned more about motion capture, revisit this project to create a virtual god boy.

My notes can be view at the link below.

Want more pictures of the Therapy Dogs? Here's their Instagram: