So, today I had my first meeting of the semester with Matt Bett. It was pretty productive, though my brain is currently in the thrall of a nasty cold, and I gained a new sense of focus.
These past two or so weeks have been rather unproductive I must admit, mainly due to being ill and to me panicking slightly when I should have been focusing. My original plan of action for the project was to first create a demo application - containing lights, shadows, physics etc, which I could then implement the ambient occlusion into. But I think I was placing too much importance on that aspect, and my talk with Matt made me realise that I should focus my efforts on the Ambient Occlusion, on actually anwering my research question and not on having an amazing 3d scene. If I have time at the end of the project, then I may go back and improve my framework but all I really need is a reasonably simple directx framework with the ability to use models and shaders.
I already have a basic framework to work with - I stripped back the framework I used for the maths coursework - so my tasks for the next fortnight is to get a simple 3d scene running with the basic SSAO shader. If I can get this done before the fortnight is over, then I will next think about the blending (or blur) that will be applied to the SSAO to make it look softer/less noisy.
Matt already suggested to different blending methods during the meeting - Gaussian blur (which is quite expensive) and a simple Bilinear blur (vertical blur then horizental blur) and I'm sure that some research will turn up even more variations.
Onwards!