Skip to content

SIGGRAPH sketching and coding

SiggraphSo, on Friday, Abhijeet and I submitted a sketch to SIGGRAPH (the premier computer graphics conference, though sketches aren’t held to anything like the esteem that papers are). And hey, it’s not blind review, like my last paper, so I can tell you about this. Though blind review or no, I can’t post the work until it’s published somewhere, or else bad people might steal my brilliant ideas.

The nice thing about a SIGGRAPH sketch is that it’s only a single page, so it doesn’t take long. The bad thing is that it’s only a single page, which means that I have to compress all of my work, including math, into a few paragraphs.

However, I’m actually pretty happy with it. It’s a method of helping computer artists find the right parameters for surfaces by showing them examples and asking them to indicate which ones are closest to what they have in mind. The idea is that instead of the user twiddling all kind of parameters they don’t understand, the computer handles that part, trying different parameters and asking the user to say which final result they prefer. Abhi’s work is in providing the model of material properties.

Funny story: I wrote the code for some of the math during my latest bout of insomnia. I noticed it wasn’t really doing what I wanted, so I looked at the equations, which were not only wrong, but so wrong that I have no idea what I was thinking about or trying to do. Apparently in my sleepless state, I based the first part on one equation, got distracted, and then based the rest on something completely different. It was the coding equivalent of saying “it was the best of times is an illusion and lunchtime doubly so”.

I also commented some of my code “## FIXME: I can’t think straight and have no idea what I’m doing ##“. But that part works.