Re: [Algorithms] Simple player collision response problem

[Jason Z. <zi...@n-...>]

I guess I phrased that line poorly.  It should have said, each frame I keep
sliding the player recursively until he comes to a stop then flip the frame
and do it again.  Problem is, when the player is in a valley he will often
come to a stop on alternate sides of the valley each frame, and thus kind of
bounce back and forth.  I can reduce this by including friction (which I
intend to do anyway), but unless I put a LOT of friction he will oscillate.
I'm assuming the oscillation will occur in corners of buildings too.

Maybe I'm doing something simple wrong in my implementation?  From people's
replies it sounds like what I'm doing should work.  Here is my code, maybe
someone can spot something.  It is fairly simple.  Note it isn't actually
recursive, but it should be the same effect.  It originally was recursive
but I found this easier to debug.

-----------------------------
// do the initial collision with world, sphereCenter is player's location,
// vel is his velocity, cInfo is a struct that contains collision info
bool collided = terrain->CollideSphere(sphereCenter,3.0f,vel,&cInfo);

// if no collision just move to destination
if(!collided) sphereCenter += vel;

// do collision response if we hit something
while(collided)
{
  // remember what our destination point would have been
  zVector3 dest = sphereCenter + vel;

  // Move to the nearest collision
  zVector3 V = vel;
  V.normalize();
  V = V * (cInfo.dist - 0.01f);    // cInfo.dist is distance to collision
  sphereCenter += V;

  // Determine the sliding plane
  zPlane sp;
  sp.normal = cInfo.point - sphereCenter;// cInfo.point is collision point
  sp.normal.normalize();
  sp.dist = Dot(sp.normal,cInfo.point);
  // create ray starting at where we should have stopped and going in
  // direction of plane normal
  zPickRay r;
  r.origin = dest;
  r.dir = sp.normal;

  // intersect with plane to find out distance to new position on that plane
  sp.Intersect(r);
  zVector3 slidePlaneNormal = sp.normal * r.dist
  zVector3 newDestinationPoint = dest + slidePlaneNormal;

  // Generate the slide vector, which will become our new velocity vector
  // for the next iteration
  vel = newDestinationPoint - cInfo.point;

  // make sure this new mini-velocity doesn't collide with anything else
  collided = wState->Terrain()->CollideSphere(sphereCenter,3.0f,vel,&cInfo);

  // if no collision add slide velocity and break out
  if(!collided) sphereCenter += vel;
}
-----------------------------
And thats it.  I then set the player's location to sphereCenter and its
done.  It works wonderfully for the most part, until you get stuck in a
crease.  Is there something I'm overlooking?

- Jason Zisk
- nFusion Interactive LLC


----- Original Message -----
From: Steve Wood <Ste...@im...>
To: <gda...@li...>
Sent: Friday, September 08, 2000 7:44 PM
Subject: RE: [Algorithms] Simple player collision response problem


> > -----Original Message-----
> > From: Jason Zisk [mailto:zi...@n-...]
> > Sent: Thursday, September 07, 2000 9:42 AM
> > To: Algorithms List
> > Subject: [Algorithms] Simple player collision response problem
> >
> > What I'm doing now is simply calculating a sliding plane
> > based on the poly the player is hitting and sliding him
> > along that with the remaining velocity after the initial
> > collision.  I do this recursively each frame against the
> > environment until the velocity is so small I can just set
> > it to zero.
> >
>
> Don't wait until the next frame to recurse the rebound.  In my collision
> detection I let the movement continue as far as is possible during the
frame
> that the collision takes place so the frame displays only the results of
the
> bouncing.  What I see at creases is a smooth migration toward the center
of
> the crease until the player is directly on the crease.
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list