Three body collisions

[Darthfett]

One cool thing I saw was a PDF talking about how you can increase AI counts with Spatial Hasing, here Ill send the link: http://www.cs.ucf.edu/~jmesit/publications/scsc 2005.pdf

Spatial hashing/partitioning is also good for reducing that O(N^2) complexity of checking all objects against all other objects, and this document provides an excellent description of it, as well as very nice images, thanks for posting that!!

 

[D.V.D]

Four scores and seven moons ago, this resource has helped me since last year. I think you've probably seen this by now. (Especially the PDF articles)

Hmm that looks interesting. I have seen a similar tutorial but not exactly that and my tutorial dealt with 2D only, not sure yet about yours but still, I'll check it out

Here's the one I read: http://www.yaldex.com/games-programming/0672323699_ch13lev1sec6.html

Spatial hashing/partitioning is also good for reducing that O(N^2) complexity of checking all objects against all other objects, and this document provides an excellent description of it, as well as very nice images, thanks for posting that!!

Haha no problem, I thought it was an interesting paper just because it took Spatial Hashing and applied it to AI and Object Occlusion. I haven't read it yet but I just finished my exams so I will have some time to play around with this paper

------

Also, is the normal vector and tangent vector for collisions a standard across all rigid body simulations? Even ones like Havok and PhysX? Or is it just one of many efficient options?

 

[tom_mai78101]

Also, is the normal vector and tangent vector for collisions a standard across all rigid body simulations? Even ones like Havok and PhysX? Or is it just one of many efficient options?

I think it's one of many options. Efficient? Not sure about it.

 

[Darthfett]

Also, is the normal vector and tangent vector for collisions a standard across all rigid body simulations? Even ones like Havok and PhysX? Or is it just one of many efficient options?

I'm not quite sure what you're asking about here, but the normal vector (or collision vector) and point of impact define the line upon which the colliding objects apply collision force against each other.
For example, when two circles head straight at one another and collide, the normal vector is the vector connecting the two centers, and force is applied in this direction to both objects. There is no change in velocity along the tangent vector (the line perpendicular to the normal vector).

 

[D.V.D]

I think it's one of many options. Efficient? Not sure about it.

Oh okay, well I assumed it was a "more efficient" option just because you don't need to calculate the velocity along the tangent vector.

I'm not quite sure what you're asking about here, but the normal vector (or collision vector) and point of impact define the line upon which the colliding objects apply collision force against each other.
For example, when two circles head straight at one another and collide, the normal vector is the vector connecting the two centers, and force is applied in this direction to both objects. There is no change in velocity along the tangent vector (the line perpendicular to the normal vector).

Yea I understand that, but for calculating velocities of objects after collision, are there quicker ways of doing so or is this one of the fastest ways to do it? Assuming they're rigid bodies of course, soft bodies have a lot more to simulate.