Attending SIGGRAPH 2005

This essay originally appeared here on the Silicon Valley ACM SIGGRAPH site. I submitted it on Thursday August 4 2005, during my last day at the conference. After some minor editing by Jessica Fernandes of ACM SIGGRAPH ("mostly a few changes to break up run-on sentences") it was made publicly available in September 2005.

SIGGRAPH in Los Angeles is becoming quite familiar: 1999, 2004, and now 2005. I'm finally beginning to get a system in place: attend the courses the first couple of days, attend the papers next, try to catch everything else, and then finish off with more papers. And keep an eye out for old friends. There's no way to experience everything, since there are way too many things happening in parallel. This essay will be a description of how I navigated through the many choices this time around.

One of the big changes now is that there's a Conference Presentations DVD ROM. In 2003 it didn't cover much, but in 2004 it covered most of the courses and papers. I expect that the 2005 version will really make it possible to catch up on the courses and papers I wanted to attend but couldn't. At less than $300 it's a bargain, and I pre-ordered mine before the conference.

I bought the 2004 DVD ROM last year, and made a point of listening to Mike Bailey's 10 minute session on "How to Attend SIGGRAPH 2004" just before this conference. He basically said that you should attend things that might interest you and won't be recorded, like the panels. Courses with animations and real time demos are also good to attend. He also mentioned that the 4 digit exhibitor codes are really encoded Cartesian coordinates. The last two digits are number of feet from the front and the first two digits are the aisle number. I didn't know that about the last two digits.

I started off on Sunday by trying out the all day Discrete Differential Geometry Course. I had a plan B in place: the Modern Techniques for Implicit Modeling. That's one of the advantages of having multiple sessions lined up: if something isn't what you'd expected, you can quickly switch to something else. I never needed plan B. I would have to say that the Differential Geometry course was, for me, the best SIGGRAPH course yet. When I was a mathematics student I attended several differential geometry courses, and ended up specializing in the subject. As such, I was very much in a position to appreciate how much effort had gone into pulling together different sources and giving intuitive explanations of the concepts. My very high expectations were surpassed. That doesn't often happen.

The most important event to attend, in my opinion, is the fast forward papers review, given on Sunday from 6 to 8. I was really surprised by how many people, some of them previous attendees, didn't know that this existed. As I explained repeatedly, this is where the presenters (of the almost one hundred papers being presented this year) each give a roughly one minute ad for the results they'll be discussing. It's very fast paced, and at the end of it you have a pretty good idea of what to investigate further. The main things I noticed are that there's lots more inspiration from advanced mathematics and physics these days, there's a strong international presence, and Jim Blinn is still going strong.

That was the end of day one. As is usual at SIGGRAPH, I came away a satisfied customer. I also picked up a detailed schedule of the upcoming week, and used it to prepare my schedule for Monday.

Monday was the first day where I had a hard choice to make. Do I attend the Manifolds course, which promises to be a lot of fun, but is probably not going to help me much in my day-to-day work, or do I attend the Performance OpenGL course? I finally chose Performance OpenGL. I skipped it last year, and this really is something I need to be knowledgeable about (if I'm going to keep a straight face when I claim that I know something about 3D graphics and OpenGL). And who knows, something might go wrong with the recording and I might not be able to listen to it later. I sneaked into the Manifolds course during the break and met up with an old friend. We agreed to do lunch, and then I went back to the performance course.

Lunch was good, but we returned a little late for the awards and keynote. We caught the lifetime award presentation. The winner, Tomoyuki Nishita, discussed his work, reminisced about the days before graphics cards and monitors, and showed an old line printer printout of a scene demonstrating hidden line removal. Ah, the good old days! Then it was time for an hour of George Lucas. I had no idea he was involved in the early development of the nonlinear editing system that evolved into Avid, and the early development of Photoshop (long before Adobe took over the product).

After that, it was time for a fun course: Visualizing Quaternions. It was a big hit, and people were turned away once the room filled up. That also happened with a course I attended on Tuesday. It pays to be early! And then it was time for the Electronic Theater.

The next few days were much less structured for me. On Tuesday the Exhibition Floor opened, and I did my first walk through. I did this again on Wednesday and Thursday. It seemed to me that there was less activity than last year. Most of the courses occurred on Sunday and Monday, and there were only a few left for Tuesday and Wednesday. I popped in and out of some of them, especially the GPU courses, and finished off on Thursday with papers all day. When I get home I'll be comparing notes with three of the other people from the Silicon Valley Chapter who attended.

I'm looking forward to doing it all again in Boston in 2006.

As it turns out, for various reasons I didn't attend the Boston conference in 2006. That was too bad, as I have fond memories of Boston, and know many people there. I really liked Cambridge, and I used to visit every week. The weather is a lot better here (Silicon Valley) though.

My first non academic full time job was in Waltham Massachusetts, which is essentially outer Boston. That was with Parametric Technology Corporation, where I worked on the core geometry code for the CAD program Pro/ENGINEER from 1990 to 1994. If you've ever used Pro/ENGINEER to create "variable section sweeps" (aka "multi trajectory sweeps" and several other names), that's my creation. I'm still amazed by the performance we achieved with the ancient hardware available back then.

Feedback

If you have corrections, additions, modifications, etc please let me know mailto:walterv@gbbservices.com

Back to top of page