Rotational Physics advice needed

Game Mechanics
#1

Does anyone have any advice on rot physics? I’ve got various objects which have multiple moments acting on them dynamically from different points, and need to accurately rotate the objects under the action of the moments.

Main problems which spring to mind:

  • What’s the centre-of-rotation when acted on by multiple moments?
  • What’s the rotational acceleration about that centre? (I suspect I can work this out by getting the A at a point, and assuming it is rotating the edge of a circle whose radius is the distance of that point from the centre, and then calculate the angular/rotational acceleration from there).

Unfortunately, it has been a very very long time since I covered circular/spherical physics, and I can’t seem to find any good resources (my books are long since vanished, and there appear to be no good resources online for this particular branch. Several days of google searches currently fruitless)

#2

Rotations are usually about the center of mass… I think you would be able to merge things to rotations in a few planes centered on that point.

#3

Check out:

http://www.d6.com/users/checker/dynamics.htm

There are some pretty good articles there on rotational phsyics.

The biggest trouble (if you are gonna do 3D) is picking a representation for your orientation. Hecker uses 3x3 matrices, but I quite like quarternions.

#4

[quote]Does anyone have any advice on rot physics? I’ve got various objects which have multiple moments acting on them dynamically from different points, and need to accurately rotate the objects under the action of the moments.

Main problems which spring to mind:

  • What’s the centre-of-rotation when acted on by multiple moments?
    [/quote]
    Well, duh! After thinking about this a bit more, I realised that by definition the centre-of-rot has to be the centre-of-mass. I’m dealing with multiple differently-defined centres-of-mass for the same body (e.g. centre-of-buoyancy for floating objects) and was being very thick, and worrying about something I shouldn’t have bothered with!

Sadly, I’ve also discovered that Java2D is horrifically slow once you start using complex CAG operations :((. It wasn’t my physics engine that was slow, it was Java2D that was painting slowly!

#5

[quote]Check out:

http://www.d6.com/users/checker/dynamics.htm

There are some pretty good articles there on rotational phsyics.

The biggest trouble (if you are gonna do 3D) is picking a representation for your orientation. Hecker uses 3x3 matrices, but I quite like quarternions.
[/quote]
Hmm. I’ve looked at Chris’s site for physics info before, and although I just checked again, I couldn’t find anything particular to rotational physics - although I didn’t bother looking closely at the books (I don’t have access to a library - online online resources, sadly).

Sorry if I’m being thick (again) but which resources in particular did you spot on Chris’s pages?

P.S. Yeees, 3x3 matrices are rather crap if you’re doing anything interesting. I’m using J2D as a zero-maintenance display technology (very useful to not have to debug graphics AND physics in parallel), so happy to use the actually-rather-well-designed AffineTransform class for all matrices.

#6

There are some PDF articles there (written for Game Developer) that are a very good starting point. Plus there is some sample C code too:

http://www.d6.com/users/checker/dynamics.htm#articles

http://www.d6.com/users/checker/dynamics.htm#samples