1. How satisfied are you with Project 3? What grade would you give yourself?

For the most part, I am satisfied with Project 3.  Although I was able to implement a basic motion graph and extract some interesting motion clips from it, I would have liked to do more in evaluating better threshold-finding techniques.  My results for detecting and selecting transition points, creating the transitions between motions, and pruning the graph are all well to my satisfaction, and so I would probably give myself an AB.



2. What went right with your Project 3?

I was very excited to see that I was correctly finding good transition points and also generating good blend transitions between motion clips at these points.  I'm actually very impressed by how well some of my results turned out.  I was a little skeptical at first that creating good streams of motion from multiple motion clips would really work, so it is very rewarding for me to see for myself that it actually does.



3. What could have been better with your Project 3?

I feel I wasted a lot of time in the beginning trying to do a lot of the computations and linear transformations by hand instead of using the existing data structures and functions in the MotionTestBed suite.  Moreover, it is difficult to say whether my approach of starting with a 2-motion based graph and later extending it to an n-motion graph was a good decision, at least in the capacity that I went about doing it.  Having the small-scale graph was very benficial for testing each step of the construction phase and verifying results, but in the future I feel it would be good to approach something like this with more of an eye for scale.  The other part I felt I wasted a bit more time on than I should have was in building the motion graph data structure for an arbitrary number of motion clips.

I am disappointed that I didn't have more time to explore different threshold-finding techniques.  In the end, I just didn't have enough time or a well-enough developed idea of what I wanted to do.  Having a correct and *working* linear blending model for my transitions helped remove a lot of artifacts that I was seeing, which made me focus more on getting my actual motion graph working.



4. What advise would you give to someone doing your project next year?

In regards to building the motion graph, this was a very well-structured and feasible project to take on.  In addition, I also felt it was an invaluable learning experience that ties in with a lot of the motion topics covered in class.  For me, it greatly helped my understanding of several underlying concepts, such as blending motions, spherical linear interpolation, and how to analyze raw motion data.

The second piece to this project dealing with the threshold values was less feasible given my limited understanding of how to numerically evaluate the "context" of a motion.  In hindsight, however, I believe this is something that could be done if one starts with a very good plan of attack.



5. What did you learn from your project?
      Do you feel like you learned enough from it?

I feel I learned a great deal from this project.  The greatest gain, however, was how this project helped me really solidify my understanding of motion graphs and how to linearly blend motions together.  With neither of these topics did I feel all that strongly in my understanding when they were first covered in class.  Moreover, understanding something in concept is much different than knowing how to implement it correctly.  I also feel that doing this project helped lay the foundation for if I ever do future work modifying motion data (e.g. registration curves).

There is always room to learn more, however, and I very much would have liked to study the problem of generating a good threshold model.  Unfortunately, the lack of good resources that address this problem make it difficult.



6. How do you think you did with the other class projects (projects 1 and 2)?

My first project didn't go as well as projects 2 and 3.  Quite often I felt that I was fighting with my own inexperience to overcome some of the problems I encountered.  In addition, there were a few things I didn't fully understand by the time that I was trying to implement them.  I was satisfied by the work I was able to do, but I certainly wish I could have done more.  Since Project 2 and 3 sort of blend together, I will group my feelings of it in with Project 3.



7. How do you think you did with the other aspects of the class?

The Blender assignment and lamp commercial assignment I think went very well.  I put a lot of effort into both of them and found them to be very enjoyable learning experiences.  For these I'd probably give myself a grade of A.  The reading summaries are a bit more difficult.  After reading the reviews of my peers once I had written my own, there were numerous times when I felt I was missing the "bigger picture" of the paper.  This wasn't always the case, but because almost everything in this course was new to me, I had a tendency to feel like I was always "catching up."  Consequently, my reading summaries might not have been as in-depth or critical as they should have been, but I still feel that I got as much out of them as I was capable of.  Seeing as how there were a few weeks when I was late getting to a paper or forgot to submit a formal summary, I would give myself a B for this piece.  Class participation is another difficult criteria to judge.  Like the reading summaries, I always felt a bit behind all of my peers, which in turn led to me being less forthcoming with answers to questions or contributions.  There were also several times when I wasn't sure my failure to understand a topic or concept was due to my lack of experience, or if I had failed to pick up on something.  Being attentive and doing my best to pick up on everything discussed in lecture is the best plan I had.  Regrettably, there were numerous days when I missed lecture and, despite circumstances, I see that as the one area that I should have had better control over.  For that reason, I would give myself at most a grade of B.