TGEA 1.70 Vehicle Collision Physics
by Ronald J Nelson · in Torque Game Engine Advanced · 03/28/2008 (2:44 pm) · 5 replies
I wanted to see if there where going to be any enhancements to the vehicle collions physics code in a future release?
#2
03/31/2008 (9:29 pm)
Wel thank you Matt for a very clear answer. My concern has only been with the often unreliable collisions with vehicles. Sure you can tweak them to get them fairly close, but it still fails to recognize collisions every now and then. 
#3
04/01/2008 (12:41 am)
Was the collision fixes posted a while back by Tom Spilman and team added to TGEA 1.70? 
#4
I have not pulled those over as of yet. Ultimately, if you were reading through that thread the conclusion at the end was that those "fixes" weren't actually making that much of a difference. I spoke to Ross at GDC and he agreed with this assesment so I didn't put a very high priority on it.
Whenever I get a chance to look at the vehicle physics in detail I will loop back on it and pull out anything useful from that thread.
04/01/2008 (10:53 am)
James,I have not pulled those over as of yet. Ultimately, if you were reading through that thread the conclusion at the end was that those "fixes" weren't actually making that much of a difference. I spoke to Ross at GDC and he agreed with this assesment so I didn't put a very high priority on it.
Whenever I get a chance to look at the vehicle physics in detail I will loop back on it and pull out anything useful from that thread.
#5
One thing I do recommend if you can't wait and still need to use the current physics is to completely ditch the while() loop outside the for loop that goes through each of the gathered hits. The outer while loop there is responsible for the occasional freeze (basically, it generates too small an impulse to sufficiently resolve the collision, and keeps looping, with ever smaller impulses) and hiccups. In general, it's no worse than with the loop in terms of failures. The thread we posted has some insight from the guys that did Mini 37, if you're okay with a patched solution.
One thing that will work much better with the new TGEA is integrating a physics library. If you really want to ensure things are rock solid, and you don't have time to patch or rewrite the current stock physics, I'd go that route.
04/04/2008 (10:16 pm)
As Matt says, the vehicle collision *response* (detection is great) basically needs a revamp. Baraff's method wasn't really ever meant to be used with high speed objects of any sort. Some form of continuous collision detection would probably be best.One thing I do recommend if you can't wait and still need to use the current physics is to completely ditch the while() loop outside the for loop that goes through each of the gathered hits. The outer while loop there is responsible for the occasional freeze (basically, it generates too small an impulse to sufficiently resolve the collision, and keeps looping, with ever smaller impulses) and hiccups. In general, it's no worse than with the loop in terms of failures. The thread we posted has some insight from the guys that did Mini 37, if you're okay with a patched solution.
One thing that will work much better with the new TGEA is integrating a physics library. If you really want to ensure things are rock solid, and you don't have time to patch or rewrite the current stock physics, I'd go that route.
Associate Matt Fairfax
PopCap
I do have some really tentative plans to beef up the vehicles but it is a lower priority than some of the other tasks I have to get done so I can't make any promises.