- It will be really good idea if we could combine ray tracing and rasterization in an innovative way. Prof. Subodh gave an example of figuring out ambient occlusion at any point by putting a camera at that point looking out, rasterizing the scene from there and somehow ( say, by occlusion culling) find out how much of the rendered depths are more than some "radius" depth. By radius I mean radius of the sphere of influence, occluders inside of which only will matter. Similar to AO, can we make use of rasterization within ray tracing for much more than just using it for primary ray casting.
- There were two projects he talked about. These two projects are being worked upon in coordination with some German university. On account of which, he mentioned, there would be good chances of doing a summer internship at Germany, to work on these projects. They are,
- Capturing and rendering fine surface details: He talked about capturing surface details from real-life photographs and using them. He was talking of achieving effect something similar to bump mapping.
- Microsoft Kinect Device: Making novel use of it to do various things: Doing things like ,one, creating point clouds or, two, extracting 3D models from Kinect captures. The device, as he said, basically captures RGB+depth info of what it "sees", -and-here-it-comes- in real-time. That makes it a really useful device.
These lines underline the importance of modeling and surface detailing. The same motivation is , in my opinion, behind the first project that Subodh mentioned.