> Are there any model traces that give the actual point of > collision, instead of just the 0.0f-1.0f value in a trace_t? > IPhysicsCollision::TraceCollide() seems like a good > candidate, but not quite, as it doesn't give me the point of > intersection.
If you're talking about a ray trace, then the point of collision is trace.endpos. If you're talking about a swept collide or bbox, then there isn't necessarily a single point of contact. You can have face vs. edge and face vs. face contacts. So really it's a point, line segment, or polygon of contact. You can get a vertex that is nearest the plane of contact by taking the support map of the swept object in the opposite direction of the contact normal (trace.plane.normal). For a bbox, the support map is the corner of the box that has the same component-wise signs as the normal. For a collide, you can call physcollide->CollideGetExtent(). so something like: Vector contactPoint = physcollide->CollideGetExtent( pSweptPhysics->GetCollide(), trace.endpos, pSweptEntity->GetAbsAngles(), -trace.plane.normal ); would work for swept collides and // mins, maxs are the bbox passed to TraceRay() Vector contactPoint = trace.endpos; contactPoint.x += ( trace.plane.normal.x > 0 ) ? mins.x : maxs.x; // note > 0 because we want the -normal contactPoint.y += ( trace.plane.normal.y > 0 ) ? mins.y : maxs.y; contactPoint.z += ( trace.plane.normal.z > 0 ) ? mins.z : maxs.z; for a bbox. But remember, this will give you a vertex on the box even when it's a face of the box in contact with the other object (even if some sub-section of the face that doesn't include the vertex is the only contact). Also in case you can use this, if you have physics objects that are actually in contact (i.e. not just traces), you can use CreateFrictionSnapshot() to get a list of contacts and those can return a point actually on the contact manifold. Jay _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlcoders
