Thoughts

Talk about tough times. Computational Geometry is simultaneously the most difficult and most interesting class I’ve taken during my stay at Stony Brook.

Things have been so hectic lately. Between my class projects, my Master’s project, my job, and my home situations, I feel like I barely have time to breath. I guess it will stay like this for another year ( as I should be graduating around December, 2008). If you’re reading this and you’re a friend of mine from home, sorry I haven’t seen you. Just one more year to go…

Though I’ve thankfully started my main Master’s work now. Klaus Mueller is my adviser, and he’s put me with a team working to create an automatic technique for building facade characterization and generation in an urban setting. The project wants to find a way to take a bunch of pictures of city buildings and extract something like a “city building template,” having options which can be adjusted to produce new building faces with characteristics of the original buildings from the photos. For example, say you have ten pictures of the fronts of generic buildings in Manhattan. Some of those buildings can five windows, some seven, the rest maybe nine. Some can have windows partially open, others closed. The first few may be composed of new and uniformly-colored brick, while the others may be degrading to discolored, crumbling stone. This project seeks to capture “parameters” out of these images: things like “window count,” “window openness,” and “wall texture,” automatically. Given a large number of images along with a semantic description of the kinds of things that typically appear on the face of a building, it seeks to collect a set of options which can be applied to new building faces. With the software, someone can take pictures of a few buildings that they find interesting, then generate an entire city that follows the same look-and-feel. They can snap photos of a building that interests them, then see what those buildings look like with two open windows, with four closed windows, with new bricking, or with old stone in just a few clicks. It can be a powerful generative ability, and if generalized to include objects beyond building faces, I can see it being put to use by those in the business of virtual world construction.

It’s a mix of computer vision, feature identification, semantic representation, and urban 3d reconstruction (city-modeling), so it’s interesting and I’m happy with it =)

Now to work on time management =(  Days like these really make you appreciate the gift of free time.