About a month ago, I finished my master's thesis I've been doing for Nvidia, Image Quality Metrics for Stochastic Rasterization. My degree is practically done, all that's left is just paperwork. From now on I'll be working full-time for the foreseeable future.

It feels pretty odd to finish school - so many fewer duties to fulfill, though fundamentally much hasn't changed. There's always more learning to do, whether it's inside the formal system or not. Hopefully I won't miss all the good things provided by the university too much, most of all the external motivation. Having a clear long-term goal can help one forward a lot, and that's something that seems to be missing from my life right now, to a degree.

I will be continuing work on the research I did for the thesis, though. After working hard on it all winter, I needed to wrap it up for now and take a break, have a bit of personal life for a change, but there's still a lot more mileage in the topic. Expanding the data and analyzing the results more thoroughly could give interesting insights on different multidimensional sampling patterns used in computer graphics. Results I got while working on the thesis indicated that even very subtle regularities in the patterns can result in certain kinds of structured error in the rendered images, but I didn't have enough time to study the exact mechanisms how this happens.

The research also yielded interesting theoretical results, most specifically that masking phenomena in the human visual system can be mostly disregarded when measuring the quality of rendered images. But I'm not here to write up a summary of everything I did - if you're interested, just grab the full text of the thesis (PDF).