Posted: 21st Jun 2007 20:23
Has anyone used both of these plugins? Which do you recommend?

Or simply, how do they compare?
Posted: 21st Jun 2007 20:43
You can't really compare them Nuclear Glory is a collision plugin, while Dark Physics is a fully-featured physics plugin based upon Ageia.

http://www.ageia.com/
Posted: 21st Jun 2007 21:06
There is no comparison. NGC only does collision, Dark Physics does physics and collision. However, I don't like DP's collision system, so I use Dark Physics+Sparky's Collision DLL. DP+Sparky's is faster than NGC alone.

If you want to compare collision systems, you'd compare Sparky's DLL (free) to NGC. Sparky's is way faster, free, much easier to set up, supports box collisions (which NGC doesn't) and allows you to rotate your collision bodies (which NGC can't).

Even though NGC uses ellipsoid collision and Sparky's uses spherecasting for sliding collision, the quality of the sliding collision is the same, except once again Sparky's is faster. No contest, Sparky's wins.
Posted: 21st Jun 2007 23:59
Wow, I was under the impression that NGC did more precise collision testing. I'll have to test rotation now and try out Sparky's.
Posted: 22nd Jun 2007 18:35
So can I use sparkys for precise collision detection of oddly shaped objects like trapezoids or triangles? I tried out NGC and it seems like it only supports ellipsoid to ellipsoid and ellipsoid to polygon based collisions. I want POLY 2 POLY collision. Has anyone done it or care about it? Not all objects in the world are spheres or rectangles.

Here's two related links in my quest for this capability.



Check out this forum thread on the nuclear glory site.

http://www.nuclearglory.com/php-bin/forums/view_topic.php?id=641&forum_id=2&highlight=poly_2_poly


Also, this:

http://forum.thegamecreators.com/?m=forum_view&t=108648&b=1
Posted: 22nd Jun 2007 18:49
Poly to poly collision is totally impractical. It would be incredibly slow, and wouldn't provide any sliding collision at all. You can do this yourself by placing boxes on each limb and testing for collision. You could even go so far as to elongate them each to their parent limb's length and orient them to that limb's direction. You could get the collision you desire, but sliding collision is more important for a game. In the end this wouldn't help your game.

Has anyone done it or care about it?

You have a lot to learn about making games.
Posted: 22nd Jun 2007 19:25
Well. My particular game idea hinges upon having accurate poly based collisions. I got it to work right with the Sparky DLL. It was really simple to get working and felt more like upgraded dark basic code than an external dll. It's very nice.

If I am careful how and when I test, the game probably won't have to slow down.

Thanks.
Posted: 22nd Jun 2007 19:29
You gave me an idea though. I will also try ellipsoid/spheroid based collisions in sparkys for this to see how well that works. At the very least I could detect what collisions to check more thoroughly for. And there is some value to sliding, yes.
Posted: 23rd Jun 2007 1:07
NGC is really not that bad =) It's a good system, albeit a little slow. For most apps it's not too much of a problem, but if you're running a sub-par system or need all the FPS you can get, there are better options.

I just recently tried sparky's dll and found it exceptional. It doesn't bother me though; six of one, half dozen of the other. I've already dished out my 25$ for NGC years ago, so till I need to switch, it works for me ^^

Cheers,
Bishop
Posted: 23rd Jun 2007 5:14
To get an almost poly to poly collision, I use a combination of vertexdata and intersect object. It's not perfect, but it works for what I need it for. (Irregularly shaped objects rolling across and irregularly shaped ground.)