if i was mitsufumi i would post something like "i want the 32 and the 20 please."
if was ludwig i would post something like "your brain should adapt to the new reality"
but if u really want an answer beyond the lag explanation i once gave in a thread that got deleted, its like china and taiwan. one car wanna penetrate the other and sometimes forgets borders. why? because when they check borders they already crossed them! then fight begins. whoever reaches the sky wins.
I'm just thinking here.. what if the engine would have a maximum of rebounce speed caused by the collision by two cars being 'inside of eachother'.
I mean, right now when there's a collision, the engine checks the difference of the positions of the colliding polygons of the collision meshes of the cars (how much they are inside of eachother), it checks the speed of both cars and then calculates the rebounce force and throws them outside of eachother based on that information. The more they are inside of eachother and the bigger the speed of one and/or the other car, the bigger the rebounce.
If there was a grade of, let's say, 0 m/s min to 10m/s max rebounce, depending on the severeness of the collision, wouldn't that cancel out rubber rebounce crashes?
(I've never seen cars rebounce more than a few meters after crashing with eachother other than when it still had kinetic engery that had changed its direction by the crash.)
So what I mean is, set a max speed but also spread out the possibilities for collisions to result in 6m/s, 1m/s or 9m/s for example, so not all reasonably heavy crashes will automatically cut to 10m/s because there is a max.
This would especially work well in lag crashes because lag allows cars to be practically totally inside of eachother.
In short, decreasing reaction levels and adding a limit.
@(SaM):
That's not a bad idea at all. And it's not even that much of a physics compromise, because when you see real cars collide they certainly don't bounce off each other that hard. In fact, I imagine the maximum rebound speed would be well below 10m/s, even after a 200mph crash.
Basically it's just the lag equivalent of a crumple zone
The problem is that when the damage modelling is improved I assume things like crumple zones and absorption will be taken into account, so this quick fix would need to be undone again. So maybe that's why it's not been fixed already.
I am pretty sure that it cant be done it a "compatible" patch. Some change has to be done in the contact/deformation model, which will come, sooner or later.
Sounds simple, but it never is that simle in practice...
If the cars react at a slower speed to an overlap, then after the next frame there will still be an overlap... than add to that the posibility of compound overlap - more than 2 cars or additional lag... you end up with so many corner cases and exceptions that all need extra supporting code, and never really work right...
So the only suitable solution is a more accurate physics model that copes better without needing a 'fix' hacked in there.
I would guess that when the damage model is updated, this issue will change from "why did I fly to the sky after that small tap?" to "how come after that small tap my car was completely destroyed?".
:-D
