Our Approach

Our tentative schedule became interrupted when one of my math professors announced a delayed due date for an assignment because he thought it was too long. He made it due last Friday but I managed to finish it on Tuesday night.

My first meeting with Ian was the day after the proposal was due. I gave him an overview of my program, and we uploaded our initial project into CVS. Our second meet was two days later, and it quickly became apparent to me that I had forgotten to schedule in time for us to just communicate with each other and for us to familiarize ourselves with the existing project. I sat with Ian for a few hours and together we made regular file reading work.

When I started working on my part, I realized that it would be much better if I produced my demos first. On one hand, it will produce an interface for visual testing and retro-testing. On the other, it will provide a simple, expandable paradigm for adding more demos.
We found this interface immensely helpful for debugging purposes. Several times, we made some error coding and the errors were reflected upon retro-testing. I made an especially obscure error in declaring one of my tetrahedra in the 6-fold subdivision, and had I not added in the demos it would have taken me a lot longer to find the problem.

The implementation of the 6-fold subdivision was otherwise fairly straight forward. Besides salvaging and integrating demos from my old code, I devoted a good portion of my time helping Ian implement his parts, because I figured that way, we could get much more done as a team.

Ian decided to read in vuVolume transfer functions to go along with the data that we read in even though that was not part of our original proposal. He built a color table to store the transfer functions, and that speeds up color calculations which amounts to table lookups.
Because it was desirable to display larger data sets, we needed to find a way to render them in a reasonable amount of time. We discussed two alternatives. One was to "mipmap" the data set so that it only ever displays up to a sixty-four cubed data set (which renders about 5 million polygons per frame and takes about 15 seconds to show up). The other way was to display only portions of the data set at once. We went with the second way and the resulting controls allow the user to specify the size of the portion to be displayed and the starting location of that portion.