🚀 go-pugleaf

> Could someone please give me some simple collison/detection routines?

This discussion was quite good, but it missed a couple of points for reducing  
the overhead of collision detection (that aren't applicable to all  
applications...)

If things move in a predictable fashion, it may be cheapest to precalculate  
_when_ a collision will happen, rather than doing it on every iteration (where  
an iteration may or may not correspond to a single animation frame).  This has  
the additional advantage that you can't iterate "over" a collision (things  
going through each other between iterations but never colliding at iteration  
time.  Another advantage is that since you know exactly when things collide,  
you know how they collided rather than trying to figure this out  
post-collision.

As an example, imagine two objects, A and B, moving in straight lines.  Rather  
than comparing positions every frame, you can project A's vector until it  
collides with B's vector at time T.  Knowing this, you can happily animate up  
to time T without checking for collisions.  (In the multi object case, T is the  
minimum of all collision times.)  Note that you must recalculate T any time  
something interesting happens, like any object's vector changing.   
Recalculating projected collision times isn't necessarily more expensive than  
simply calculating current collision statuses, however.


Some kinds of games lend themselves to an active collsion list.  Objects that  
are outside they realm of interest (like villains two screens behind the hero)  
can be removed from the active collision checking list to save cycles, and  
things entering the realm of interest can be added as needed as part of  
bringing them to life.

Hope these thougts are useful.  I might add that the cost of collision  
detection is insignificant compared to the cost of moving graphics around in  
virtually every application I have ever seen.  If performance is important,  
your first concern should usually be to minimize the amount of drawing you do.

-sam