Abe Davis

Some Stuff I Do/Have Done

I haven't updated this in a while, but I recently coauthored a paper that was acceptied into Eurographics 2010. The paper is on high dimensional filtering. I will add a more details when I get back from winter break, but here is the paper website:
http://graphics.stanford.edu/papers/permutohedral

Current Research (Stanford):

I am currently working with Marc Levoy on a project involving Computational Photography on mobile phones (mostly the N95, though I have experience on other platforms). The project has been the focus of my research for this and last summer, and I have implemented components of it as two different course projects. I am going to hold off on including this work here for now, simply because I want to publish the work an it's not quite finished yet. Hopefully I will have something to add to the site soon.

At JHU:
In high school I did research at JHU. My main project involved the design of a structure for collision detection and LOD rendering on the GPU. The project won me seventh place at the Intel Science Talent Search. Here is the
• paper
  . Keep in mind that GPU programming was a lot different in 2005, so this may seem dated to some people. Also I wrote it when I was 17, so it's not quite on the same level of the work I do today :-p.




Course Projects:


Glacier Cave: (part of the CS348b Rendering Competition)


This was my team's entry into the CS348b Rendering competition. We modeled snowfall, seasonal changes, and erosion to simulate the formation of glaciers and moulins in matlab. The simulation produced a volume, which we then opened in what we called the 'Glacier Viewer.' The Glacier Viewer allowed us to extract different isosurfaces for visualization, perform quick convolutions on the volume, arrange cameras and lighting, etc. Once we found an image that we wanted to capture, we printed out a scene file for PBRT. We then rendered the image using the extracted isosurface as geometry and the underlying volume as a participating medium. We did modified Volumetric Photon Mapping to simulate different scattering in different wavelengths (this gives the cave its blue color). Finally, we added an adventuring professor to the picture. Perhaps Pat Hanrahan on sabbatical?

Good news! We won the competition! This means I will be going to SIGGRAPH this year in New Orleans! Go team ADJB!



Animelee:


This was a game that I wrote as part of CS248, the graduate level intro Computer Graphics Course at Stanford. It is a 3D fighting game (Think Super Smash Bros meets Powerstone but written by a couple of students). The game has a lot of features in it, but what I was most proud of was fast conservative collision detection on dynamic objects. I basically fit Oriented Bounding boxes to joints in the character model then build a bottom up BVH. The BVH was generated automatically and certain leaf nodes in the hierarchy were tagged as 'dangerous' for each move (the fist for punching moves, feet for kicks, etc). It seems simpler now, but when I was a sophomore this was an intense class project. Last year I became a TA for the course.



Object Detection:


These are just some photos that I grabbed from some object detection work that I did as part of an AI course. I will try to remember to put up the writeup from the project at some point.



Ping:


This was a game that I wrote with a partner for the iPhone programming course. It is sort of a flashy, fast paced asteroids featuring lost of glowing enemies and explosions. We were asked to show it off at the WWDC in San Francisco in 2009. We are hoping to put the game on the iTunes store at some point, but need to get a new developer licence and make sure it works on all of the different ipod and iphone versions first. Thanks to the folks at Apple for offering the course!