To my amazement I could find no CollisionPoint. I now use a tiny CollisionSphere, but that looks like overkill. Am I looking at the wrong place?

a Collision point as a collision solid?
would make no sense. a point is infinite the chance that it will collide with anything is infinite small. (this could only work with other collision solids like plane,tubes,spheres but no actual geometry)
what special case of collision do you have that you need a point?
other question… why would a collision sphere be overkill? they are fast and reliable.

@ThomasEgi: Uh… infinitely small? Do you don’t even have the slightest idea that when you would put a CollisionPoint inside some object it won’t collide?

inside a collision solid with a volume it will collide… a point inside a filled sphere certainly does collide. but like i said. no chance to make it collied with anything else than collision volumina.

Nothing weird, I should think. I have a plane circling around the earth (well… my game has :cry:), and in order to find out when it arrives I have it fly into a CollisionPlane.

hm… well you could use about anyting… i would recomend either sphere or a small ray segment.
so using a sphere which has the size of your spaceship is perfectly fine.
as long as it works it’s all right :slight_smile:

Maybe a CollisionPolygon???

Actually, I currently use a very tiny sphere, just to create an event when the centre of my airplane passes the desired location. But of course computing when a point passes a collisionplane is only half as much work as computing when a sphere collides.

I don’t really care, because I am not short of cycles, and there aren’t a lot of planes flying, but i was just wondering why the simplest collision from object didn’t exist.

Thanks for thinking along, though!

Is that true?

If I were to write a point-plane collision test, I would use the distance-to-plane equation, which is basically a dot product. If that distance is less than zero, the point has intersected. But the sphere-plane collision test is almost exactly the same–you compute the distance-to-plane from the sphere’s center (a dot product), and compare that to the sphere’s radius. If the distance is less than the sphere’s radius, the sphere has intersected.

Is there a way to determine the point-plane intersection faster than a dot product?

But the real answer why we don’t have a CollisionPoint is that it is so rarely needed. And, frankly, the overhead of crawling through the scene graph looking for things to intersect with is usually going to dominate the actual intersection test itself, so even if a CollisionSphere were vastly more expensive, it probably wouldn’t matter much in the big picture.