1
$\begingroup$

This summer I will be working with a professor on a research topic of my choice, in the field of graphics & simulation. While my professor is knowledgeable in the area he is being very hands off as he does not want to make the student research his research. The intent is to use OpenGL to work on the topic.

We previously had a talk about the idea of taking a scene, specifying a water level, flow direction, and then generating a heat map at that height level of water disturbance to then allow us to render a scene at low cost with "semi-realistic" water flow.

While this idea interests me, I was looking to try to look into more ideas regarding fluid's or possible other area's of the field to attempt to find other possible projects.

In short, what would be a good place to look for entry level research projects regarding computer graphics? Or just idea's in general to help steer a beginner.

$\endgroup$
  • $\begingroup$ If it were me, I'd avoid OpenGL and use Metal, DirectX 12, or Vulkan, as OpenGL is now a legacy library that probably won't be updated in the future. $\endgroup$ – user1118321 Mar 4 '17 at 20:58
  • $\begingroup$ @user1118321 I plan to move to Vulkan after the project, OpenGL is just more covered and accessible right now. Also, my professor doesn't have experience in Vulkan sadly. $\endgroup$ – Zombleh Mar 5 '17 at 1:35
1
$\begingroup$

If you are going for physics. You can do physically based rendering. Physically Based Rendering(PBR for short) is a filed of computer graphics where we take consider of physics(energy conservation, spectrum, etc) in to account and try to not do approximations.

The are two main sub fields of PBR.
(quick intro, layman's term below)

  1. Production Rendering. This is where accuracy is the focus. We do as little or no approximations as possible. There are algorithms in this field where if the wave-like property of light is ignored. Then the result generated by the algorithm is physically accurate. But each frame might take more than a few minutes or even hours to render. There are a lot of different topics actively being researched in this field(fast ray-scene intersection, better Monticello integration, fast/robust acceleration struct building)

  2. Realtime Rendering. In this field, speed is the key concern. Where accuracy is achieved by doing good approximations. But will never be physically accurate. Most research on this field is focusing on doing good and fast approximations.

These are my random ideas for fluids.

  1. Fast Ray-Liquid intersection testing
  2. Fast triangle mesh generation from particles.
  3. Fast acceleration structure building from particles.
  4. Simplification of liquid surface to generate simpler mesh

And some random ideas for general GC.

  1. Improve sampling efficiency for sky maps in ray based rendering algorithm
  2. Robust method of dealing with floating point accuracy problem in shadow maps

I don't know if the above topics are possible or not. Those are just ideas.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you! This helps a lot to get to a good starting point. I really like the idea of Fast Ray-Liquid intersection testing, but might be out of scope. I'm going to look into it. $\endgroup$ – Zombleh Mar 5 '17 at 3:16

Not the answer you're looking for? Browse other questions tagged or ask your own question.