I have just submitted the slides for this week’s workshop and am pretty confident with what I intend to do for my honours project. My current working title is “An evaluation of deep G-Buffer use to enhance screen space rendering effects in games“. A bit wordy but put simply it means by storing additional depth information in our G-buffer during a deferred rendering pass how can we improve screen space rendering effects. I originally wanted to look at ways of producing real-time interactive Global Illumination in games, when looking around I found this paper titled “Deep G-Buffer for Stable Global Illumination Approximation“. This was the first paper that I found that didn’t require large amounts of pre-computation( e.g. Voxelization or Rendering baked light maps) but still produced visually impressive results in real time. In the paper, the authors talk about other effects that can be improved using their deep G-Buffer method so I thought that this was worth investigating which allowed me to open up the scope of the project. There are papers from as far back as 1998 that describe similar techniques called “Layered Depth Images” so there is plenty of background however there isn’t that much recent work except for the one paper linked above. I am slightly worried that I will be taking most of my information from this latest paper which feels like I won’t be carrying out any of my own work since they have provided such a thorough explanation of their research. I don’t want to feel like I have copied their paper and just moved some of the words around.
I am currently working on building a graphics framework in which I can create a prototype to demonstrate the researched techniques. So far this is going well. The lighting is currently very basic but the underlying framework is quite solid and includes plenty of helper classes and functions to improve the speed of iteration. It also includes a basic scene graph to make scene management much simpler and It is also able to run both in a Windows 10 universal application and an old school Win32 application. The reasoning behind this is that it is far easier to develop using the Universal Windows APIs as it makes file input and output far easier as well as simplifying UI changes. Yet I feel that I will need a Win32 application for demos as the university is yet to upgrade to Windows 10. I will continue to update the code which I manage through this public GitHub repository.
Over the next week, I plan to do a simple literature review and create a plan detailing what I will need to have ready for the feasibility demo in week 14.
By the end of the week, I plan to have a list of white papers, books, presentations, blog posts and videos which I can use as a reference for my dissertation.
I also plan to have asset importing working inside the graphics framework.