Fractal Benchmarks

I blame whoever printed out the Sierpinski triangle wikipedia page on Friday, but I’ve always been interested geometry, so I had to have a go at building one. The title of this post is a reference to the fact that this type of procedural geometry is often used in benchmarking 3D graphics systems as it quickly becomes computationally explosive. I’ve been writing geometry classes to create physical networks from graph topologies anyway, so this was easy by comparison.


The movies of depth 4 and 5 pyramids are on YouTube at the following links:

As a sneak preview of what else I’ve been working on, you can see a 3D GIS engine to visualise all the real-time London data that we’ve been collecting. If you look very closely, you can just see the tube network inside the pyramid. The general idea is to make it easy to generate some of the 3D tube and bus movies that I’ve previously used 3DS Max for. What was missing was a geospatial data aware 3D system, which was lacking in 3DS Max or Three.js. I’ve taken the Javascript Three.js visualisations about as far as they will go, so a higher end visualisation system in native C++ using OpenGL was a natural progression.