Final project discussion & more rigid bodies.

Thoughts

• Final project ideas all look very cool. I will probably implement my own fluid simulator and implement some new paper on it (or maybe MPM?). But also implementing the paper I’m presenting about position-based dynamics for rigid bodies sounds interesting… Or maybe fluids on curved simplicial surfaces, who knows.
• Already went over integrating rigid bodies without contact, but now getting to contract. Real objects are always not 100% rigid, but we don’t really want to simulate that here. We can use impulses to represent discontinuities instead.
• Finding point collisions via binary search is pretty easy. Separating axis tests are rather more involved and slow, but also make sense. (Something something GJK.) Also closest point queries & BVHs for accelerating triangle overlap tests.
• Impulse based methods are pretty straightforward, but finding time of collision is not ideal. Penalty and constraint based ones are quite cool; penalty is essentially simulating the normal force with springs. Constraints seem the most physically accurate, though–the normal force is truly derived via the gradient of collision constraints.
• Morgan McGuire took this course at Brown, huh