Design 950 Autumn 2003 with Wayne Carlson

This quarter's two 950 classes focused on my thesis work. Since I covered my work on cleaning the dog motion capture data within Brian Stone's 950 documentation, I'll describe the 3-d modeling in this one.

Ideally, my goal was to have completed the dogs' bones, muscles, and outer skins. I wanted to show the muscles and other structures underlying the skin because I felt it would provide more information to the students on whom the learning module will be tested. I've looked at some books on gait analysis from the College of Veterinary Medicine's library. The best ones show underlying muscle structures and describe which muscles are doing what at various times in the gait cycle. Two books, I found impressive were, "The horse in motion" by Sarah Pilliner, Samantha Elmhurst, and Zoe Davis, and "The Dynamics of Canine Gait" by Leon Hollenbeck. I wanted to take this concept into the 3 dimensional world. Another book I've been using a lot is "Dog Locomotion and Gait Analysis" by Curtis M. Brown.

I motion captured 6 dogs and I have 18 takes of them walking, trotting, and climbing stairs. It has taken me hundreds of hours now to clean this motion capture data. Though, I think it would be wonderful to show all of the takes with perfectly bound 3-d models of the dogs' bones, key muscle groups contracting, and their outer skin, realistically, this amount of work is beyond what I can do by myself within the course of a Master's degree. Even if I completed all the bones, muscles, and outer skins for all six dogs, perfectly scaled, I would have to solve binding and rigging problems. So, I'm going to have to look at alternative means of showing this data, e.g. binding simple geometric shapes to the motion capture skeleton. I can use cylinders for bones, and spheres at joints. I can color code them for the left vs. the right side of the body.

I've made a committment to create a learning module of mocapped canine gait both to Dr. Jonathan Dyce, and TELR. This is part of my thesis work. However, since I am testing how the students would like to view / study motion, I'm thinking it may be possible to create one 3-d dog rig that shows muscles contracting, and follows the motion capture data. I'm going to compare this to video, and sequential photographs borrowed from either Edward Muybridge's "Animals in Motion," or Curtis M. Brown's "Dog Locomotion and Gait Analysis." I'm thinking I can use one gait pattern - for example walking or trotting. Since it would be only one animation, instead of 18, it may be possible.

Outer Dog Skins

This quarter, I've worked on taking a Nurbs dog model, which had a lot of separate geometric shapes making up a dog's body, into a complete seamless Polygon model. I not only needed the joints to be seamless, I needed a polygon model to take into Motion Builder 5, the program I learned and used this quarter for cleaning my dog mocap data. View the polygon modeling details.

Dog Skeleton

I also worked on skeletal canine bones. View skeletal bone creation.

The Animation Skeleton for the 3-D Models

I worked on creating an animation skeleton to go under the 3-d models (to bind them to). I used a cat skeleton in Motion Builder as a guide. Motion Builder requires you to follow their naming convention for the joints. This animation skeleton was never able to follow the mocap marker data, but I still have it to use in Maya for binding the dog models. Kaydara, the makers of Motion Builder, sent me e-mail saying that I was trying to do something with their software it wasn't meant to do. Though they say they have quadruped support in version 5, this is not for people trying to get imported animation skeletons to follow quadruped marker data. It is only for people who have created a quadruped model in Maya and are importing it into Motion Builder to rig and animate. Why anyone would want to use Motion Builder instead of Maya for animating, I do not know. View the animation skeleton. Note that the skeleton's joints are extended because this is what Motion Builder documentation recommended for import.