BEPUphysics

What is the mass in your engine computed in? Newtons? Pounds?

It is unitless; you can consider it to be anything you'd like (though newtons (and usually pounds) are measures of weight/force, not mass ).

The important thing is the numerical values. Always remember that you're dealing with a finite precision representation of numbers, so setting values like 10e-10 or 10e30 won't work so well.

Also, the solver has a hard time when a heavy object 'depends' on a light object. For example, a 5000 ton truck smashing a tiny box would result in the box squirming around and popping out. Note that the reverse is not true: a tiny box on top of a heavy truck will work perfectly fine, because the tiny box is 'dependent' on something massive, not the other way around.

If you're seeing undesired instability in articulated structures or piles of objects with differing masses, examining the mass ratios involved is a good place to start. The allowable difference varies, but in general, going above a 10:1 ratio between an object and its dependency will cause problems.

Norbo wrote:It is unitless; you can consider it to be anything you'd like (though newtons (and usually pounds) are measures of weight/force, not mass ).

Actually, I just wanted to know so I could convert it to pounds and get something that would weigh around 10 lbs. =p

I think you can just use 10 for the mass, and then base everything around that. Norbo will correct me if I'm wrong but I believe everything is relative.

If you want a 10 pound box to be carried by a 200 pound wagon, make the mass of the box 10 and the wagon 200. As long as you are consistent between objects (and follow the rules in Norbo's reply) it doesn't make a difference.

Now watch Norbo come in and say I'm totally mistaken ha.

I think you can just use 10 for the mass, and then base everything around that. Norbo will correct me if I'm wrong but I believe everything is relative.

If you want a 10 pound box to be carried by a 200 pound wagon, make the mass of the box 10 and the wagon 200. As long as you are consistent between objects (and follow the rules in Norbo's reply) it doesn't make a difference.

That is correct!

There is one more small detail that I left out that makes the absolute value of the masses involved matter, in addition to floating point issues. The solver will early-out if the impulses are tiny enough. Giant masses will prevent the engine from reaching those tiny values sometimes which can make computations take longer while extremely small masses may let the solver early-out too much. The minimum impulse that constraints start with is SolverSettings.DefaultMinimumImpulse. This effect is rarely noticeable, though.

All of my other objects are kinematic, so the only one that ever matters is the one I set. =p I'll play around with it, thanks for your help.