I have an interesting question regarding the use of joints:
I want to allow the player to connect boxes together using a sort of “rope” that is made up of joints. Now my problem is this: some of the time, the boxes being “tied” together will be at rest, but at times, they may start to fall, and in such cases, the rope should hold them together in a chain. The problem arises from the fact that for the “rope” (series of character joints) to be stable, each rigidbody in the chain must be of similar mass or else the links in the rope will begin to “freak out” (I’m sure you’ve seen this before and know what I’m referring to).
Okay, so the solution would seem to be to make both the boxes I intend to allow the user to “tie up” and the links in the “rope” have similar masses. But that leads to another problem: if my “rope” has, say, 10 links between each box I’m letting the player tie together (this is so it looks nice and saggy like a rope should when not under tension), then if the mass of each box is 1, and the mass of each rope “link” is 1, then I’m exerting a force of 10 units of weight to every 1 box, meaning the rope itself will pull heavily on each box it is “tied” to.
I’d rather the rope be like a real rope and be far lighter than the objects it is tied to so that it doesn’t exert any significant force on it. But when the boxes begin to fall, the rope will hold them together without becoming unstable.
try raising the solver iteration count under Edit>Project Settings>Physics to something around 20.
and lower the fixed time step to 0.01 under Edit>Project Settings>Time .
Check if the whole thing become more stable.
then take the values a bit back if this gives you performance trouble.
The next thing i would check to keep stability is that you dont have any code affecting transforms of rigidbodies in Update() functions , if you do move them all to FixedUpdate() functions.
lastly play with the damping values to stable the system some more.
Tell me how it goes for you.
Well, that certainly increased the stability. Though if I tried to allow bodies on the “rope” with masses, say, 100 times greater than the individual links in the “rope”, it freaked out again. About the best I could do with those settings was to make it so that the rope was stable with a 10:1 ratio. Meaning a box of weight 10 would be equaled in weight by a length of rope attached to it with 10 links in it (not even really enough to make it look like real rope).
I know there’s got to be a better workaround for this as I’ve seen several games pull this off before (like Armadillo Run). I don’t think they were using PhysX, but surely that isn’t the limiting factor.