The Latest News

The project report has been completed and handed in to our advisor. The project homepage contains links to all the latest releases, as does our project page on SourceForge. Having published our report at Worcester Polytechnic Institute, we are uncertain if either of us (the authors) will ever look at this code again. However, anyone interested can contact us for information regarding the current status of this project.

Jay finally came down to RI to code up and write some more into our report. We are decimating beautifully with great results. Both computers here are decimating 200+MB meshes for multiple level-of-detail analysis on Matt's Indy. In other news, the report is really taking shape. Much planning is still underway for tackling the results, analysis, and conclusions sections, but the end is finally in sight! You can look at our latest report and reference manual snapshots by linking to Report and Documentation at the left. Here's to graduating in January! - Tim

I met with Matt today and tried out the code. He must have a bad GL implementation (Visual C++ default) or maybe a bad STL implementation, because our results were giving coocky normals and we were looking through to the back of the results. I checked the latest code on the CCC machines and at home, in cygwin, VC, and Linux, and everything looks ok. I did fix a bug that was crashing the program when calculateDistance was called on a Vertex with no neighbors. As long as we can demonstrate that the code decimates properly with yoda.wrl and large meshes, we should be all set. - Tim

I added the new lit reviews and polished up the introduction. Something is still wrong with the bibliography citings and some of the pictures, and we still have a long way to go. Please tackle anything that says FIXME as soon as possible and don't forget to update before you code or edit the report and commit frequently! I can't stress that enough... I keep making major changes and can't update because you keep overriding my changes with old code you haven't committed. Thanks. :) - Tim

The code has been updated, hope you like it. If not, too bad, it's our project not yours. Anyways, I made some more of that infamous side progress today. I wrote a removal algorithm, and made hours of attempts at getting sorting to work. But, I can live without those speed improvements, damn STL! Like I said, code is up. Download it with cvs. Someday I will put zipped packages of the code in a readily accessible place on this website, but until then you will have to use cvs. Am I talking to myself, or does anyone read this. - Jason

The promised code update will be coming tomorrow. We are one step away from freezing our code. The last step is making the vertex removal algorithm a little better. - Jason

This is another big day for progress on the code. Updates include basic VRML 2.0 writing, the code decimates MUCH faster, and now the command line works, so you can decimate without opening the GUI. To be fair there are a few issues that still linger in decimation process, but it's coming along fast. No code update right now (still testing), but I may have time tomorrow. - Jason

New code has been updated in cvs. Update includes more work on memory issues, fixed the Windows port and included a compiled binary, and made some speed improvements to the model loading code. - Jason

I've been moving stuff around in the repository and altering scripts for documentation generation. Please re-checkout, not update, to make sure you get the new directory structure. I've been removing the compile scripts in the report directory and putting the functionality into a Makefile. Just type "make" in there to compile both the mqp writeup and the reference manual in many formats. Output goes to the /products/mqp_writeup and /products/refman directories. - Tim

Fixed the memory issues. Now all vertices and triangles removed from the Model are unallocated in memory. What does this mean? Well, it means the code is near completion.

There are two issues I need to work out. The first is that sometimes we have overlapping polygons, which is an unwanted artifact in the Model. The only possibility I can see for this is that we are not choosing the best triangulation method, since any hole remaining from a vertex removal can be triangulated cleanly.

The other issue is with my computer's implementation of the STL. I have tried many of the examples on the Internet, with no luck. The set container is not sorted based on my overloaded less than operator, which is what it is supposed to do.

What is important now is that I am currently decimating a huge model, and it's going at a good speed. - Jason

The work that was completed last night works perfectly with the existing code (after some tweaking). So it seems that this code is now better than the existing code in cvs. But, please wait for an update if you need to download code, as I need some time to test this code. It should be noted that I am having memory issues. It seems that when I remove a Vertex I am not actually deleting from the Model, so it still has memory allocated. And with triangles, I remove (not delete) and then I create more, so this is even a worse situation. - Jason

I seemed to have some side-progress coding today. That means I wrote a lot of code that I won't use. I'll probably have to get the old code from cvs, so I can re-think my strategy tomorrow. Goodnight! - Jason

The code in cvs has been updated. There have been general bug fixes, corner and edge recognition, and some drastically improved decimation code in the Decimate object. However, work is still needed on the Decimate object. This seems to be the last hurdle in having a robust decimation program. - Jason

Today we completed more of the website. We are now in the process of writing reviews for the previous work we have researched. - Jason