december 28th, 2013
box kite version 1
One of the non-coding hobbies I'm looking at trying is kite photography, which is using kites to take aerial photographs. So far I'm only doing research/testing of kites, but eventually I'd like to be able to use my Raspberry Pi board to build a controllable camera system for a high performance kite. Over the last few days I tried my hand at a building a simple box kite...
december 24th, 2013
I've been spending a bit of time working with my 2D codebase over the last few weeks. I added quadtrees to help accelerate ray collisions, and decided to stress test the code a bit by making a simple block world with dynamic snow...
november 24th, 2013
I mentioned earlier that I was looking into MLT, an alternative sampling method for ray tracers. It's based on the same underlying pathtracing/bidirectional path tracing logic, but attempts to make statistically advantageous sample choices to produce better results...
november 11th, 2013
realtime 2d light tracing
My family visited me for a while last week, so I didn't get as much work done. I'm back into the swing of things again now with work on Ignis, but I do have one video from my 2D light tracer that seemed worthwhile to share. While refactoring the 2D code I decided to add threading support and see how well it ran in real time...
october 27th, 2013
2d path tracing
One of the challenges of working with new ray tracing techniques is that the algorithms tend to be hard to debug and test in stages. This is compounded by the fact that application is in 3D and non-realtime. I've found it's easier to implement logic in 2D first if possible...
october 26th, 2013
metropolis light transport
Over the last week, I've been spending my evenings reading through papers on Metropolis Light Transport, a modified ray tracing algorithm that often converges faster than plain path tracing. MLT is based on a sampling algorithm called Metropolis-Hastings, which was originally developed for use in statistical physics.