On Wednesday, March 19, 2003, at 09:05 AM, Geoff Canyon wrote:
But before you even do that, you should come up with a rough, simple calculation that will tell you there is no collision, or that you need to do the more exact test. So for example, if polygon A has no point more than 30 pixels away from its loc, and polygon B has no point more than 25 pixels away from its loc, then your first calculation would be the distance between the loc of A and the loc of B; if it's more than 55 pixels, there is no intersection. If it's less, then you do the more exact calculation.
A rough test is a very good idea, especially if there are lots of objects that might collide.
An alternate rough test is the bounding rectangle overlap.
If you're looking for collision detection, I think it would be faster to check the line segments of polygon A against the line segments of polygon B to see if any of them intersect. If they do, there's a collision. There may be a faster way than that, but I think at least that it would be faster than computing the areas.
Ah, I just mailed off the same idea. There is a hole. That is the case of one being completely inside the other. I covered that in my mailing, but it may not be needed.
Dar Scott
_______________________________________________ use-revolution mailing list [EMAIL PROTECTED] http://lists.runrev.com/mailman/listinfo/use-revolution
