## Point of Collision using OBBs

### Point of Collision using OBBs

I need to find the point of collision between two OBBs and a OBB and a
plane.

Does anyone know of a document on the web which would explain how to find
this.

BTW - I already have the intersection tests working, but mine can only
determine if they intersect or not.

Thanks.
--

Live Long and Prosper
ICQ - 45835769

### Point of Collision using OBBs

Quote:> I need to find the point of collision between two OBBs and a OBB and a
> plane.

> Does anyone know of a document on the web which would explain how to find
> this.

For OBB-OBB intersection, see
http://www.magic-software.com/Documentation/dynacoll.pdf

For OBB-plane, project the OBB onto a normal line to the
plane.  The plane projects to a single point.  The OBB is
just touching the plane when the projection interval has
the plane projection point as an end point.  You can easily
determine which vertex of the OBB generated the
corresponding end point of the projection interval.

--
Dave Eberly

http://www.magic-software.com

### Point of Collision using OBBs

Quote:> For OBB-OBB intersection, see
> http://www.magic-software.com/Documentation/dynacoll.pdf

> For OBB-plane, project the OBB onto a normal line to the
> plane.  The plane projects to a single point.  The OBB is
> just touching the plane when the projection interval has
> the plane projection point as an end point.  You can easily
> determine which vertex of the OBB generated the
> corresponding end point of the projection interval.

Thanks Dave. Makes sense.

However that depends on the box not moving very far "into" the plane
(without modification).
But it is processor efficient :-)

--

Live Long and Prosper
ICQ - 45835769

### Point of Collision using OBBs

Quote:> > For OBB-plane, project the OBB onto a normal line to the
> > plane.  The plane projects to a single point.  The OBB is
> > just touching the plane when the projection interval has
> > the plane projection point as an end point.  You can easily
> > determine which vertex of the OBB generated the
> > corresponding end point of the projection interval.

> However that depends on the box not moving very far "into" the plane
> (without modification).

What I described is for a static OBB.  For a moving OBB, you
can compute the projection interval as a function of time.  If
the time interval over which you want to detect collisions is
[0,T], then you can look at the projection interval at t = 0 and
determine on which side of that interval the projected plane
point is.  At time t = T, it is possible that the OBB and plane
are still separated, but you can test if the projected plane
point is now on the *other side* of the projected interval for
the OBB.  If this is the case, then the OBB must have passed
through the plane during the time interval.  If you want first
time of contact, you can determine that by computing the time
when the plane projection point is one of the end points of
the projected interval for OBB.  Just linear equations in t to
solve.

--
Dave Eberly

http://www.magic-software.com

### Point of Collision using OBBs

Thanks Dave.

--

Live Long and Prosper
ICQ - 45835769

> > > For OBB-plane, project the OBB onto a normal line to the
> > > plane.  The plane projects to a single point.  The OBB is
> > > just touching the plane when the projection interval has
> > > the plane projection point as an end point.  You can easily
> > > determine which vertex of the OBB generated the
> > > corresponding end point of the projection interval.

> > However that depends on the box not moving very far "into" the plane
> > (without modification).

> What I described is for a static OBB.  For a moving OBB, you
> can compute the projection interval as a function of time.  If
> the time interval over which you want to detect collisions is
> [0,T], then you can look at the projection interval at t = 0 and
> determine on which side of that interval the projected plane
> point is.  At time t = T, it is possible that the OBB and plane
> are still separated, but you can test if the projected plane
> point is now on the *other side* of the projected interval for
> the OBB.  If this is the case, then the OBB must have passed
> through the plane during the time interval.  If you want first
> time of contact, you can determine that by computing the time
> when the plane projection point is one of the end points of
> the projected interval for OBB.  Just linear equations in t to
> solve.

> --
> Dave Eberly

> http://www.magic-software.com

Hi, I have read an article on the web (I forgot where it is) about using
collision map for pixel-accurate 2D collision detection - the basic idea is
to bitwise AND the maps to find any non-zero bits.

Is it plausible to apply similar techniques for 3D collision detection?
My idea is to flatly project the objects onto the XY, YZ and XZ planes
respectively, and see if all of the 2D maps collide.