Hi there,

I am in need of some ideas for my final year project.

I know c++ and OpenGL pretty well and feel pretty confident with both of them and ideally want something around these.

My current idea would be some kind of GPGPU programming using OpenCL and CUDA to implement a rigid body physic system and a Cloth simulation running "mostly" on the GPU.

My main problem is scope. I think the word count is to be around the 12,000 mark and im not sure that I would be able to write that amount about it.

Do you guys think that topic would be suitable? (and) or can you suggest others?

Thanks

My current idea would be some kind of GPGPU programming using OpenCL and CUDA to implement a rigid body physic system and a Cloth simulation running "mostly" on the GPU.

Do you want to handle constraints? If yes it will be hard and will take some time, specially to defend the project (assuming that is a thesis). You will need to check all Erwin Coumans GDC articles about GPU parallelization also.

I would pick cloth simulation because is not so generalized as a rigid body simulator and explore the acceleration algorithms such described on these milions of papers.

Changing the topic to the Math and Physics section you will have a lot of good advices if you pretend to keep with physics.

My main problem is scope. I think the word count is to be around the 12,000 mark and im not sure that I would be able to write that amount about it.

I faced a similar problem when I had to write my thesis. Depending on what you need to cover in your paper, you will likely cover a big amount of your word count by covering the research you did and implementation details and defending why you chose implementation X over Y and details around those decisions (pros and cons etc).

If you're keeping the topic you are having right now, you have 3 main topics to cover: rigid body physics, cloth physics (and likely enough is to be written about the interaction between the 2) and GPGPU (why GPGPU? What's the (dis)advantage etc) and those are 3 pretty big topics in itself.

I don't think you will (or rather should) have any issues getting towards your 12.000 mark, there should be plenty to cover even if you only chose to cover one of the topics!

Thanks for your reply.

I am thinking about changing my idea to be something more along the lines of:

Creating an Engine with a Deferred Renderer, and then creating the Cloth simulation to be integrated into said engine.

I have prior experience with engines and understand how to create them etc etc, the only problem now would be that is the idea too simplistic for a "final year project"?

With deferred rendering I feel that there is a lot more for me to talk about and discuss the reasoning for than if I were to stick with my first idea..

What do you think?

P.S I was just looking around at shadow mapping stuff and found this:

At 27 seconds there is a scene - This is something similar to what I would be creating..(Sponza with flowing cloth banner)

Thanks!