Help Which Obi To Use?

[josemendez]

I'm posting this here in the ObiRope section, because my organization has already purchased ObiRope, and we've been attempting to use it to achieve our desired results with no luck. I'm trying to determine whether or not ObiRope is appropriate for other "strand" uses, that are not rope-like. For example, a very long piece of hair attached to a comb would drag behind the comb as it's moving, much like ropes do with ObiRope when they're hanging from something. However, that same strand of hair would also blow about in any breeze. More specifically to what my team is trying to achieve, is with long sheets of paper. Say in game we're trying to simulate the unraveling of a paper towel roll that was thrown. The issue we're running into is the effect of the end of the roll snapping back towards the roll each time it extends away from the roll. This sort of "snap back"/"bounciness"/"spring" effect isn't accurate to paper as it is to a rope. Is there any way we can disable this affect?


If not, would ObiCloth be better suited to our needs? Meaning, could the simulation of paper unraveling from, and flowing behind an airborne roll of paper towel being achieved?

Thanks in advance!

Hi there,

This "bounciness" is inherent to all iterative physics solvers, which means pretty much all physics solvers used for interactive applications will suffer from it. This encompasses all Obi products, PhysX, Havok, ODE...

This happens because iterative solvers generally don't reach a perfect solution to the underlying system of equations. They only get arbitrarily close to it (the more iterations you allow, the closer they get). Some more technical reading: http://blog.virtualmethodstudio.com/2017...3-solvers/.

There's many ways to alleviate it, that you can combine:

- Use more solver substeps.
- Use more distance constraint iterations.
- Crank up solver's "damping" parameter.
- Use tether constraints, if possible.

Now, specific to your use case: Obi Rope does not model torsion, which is not very important for radially symmetric objects (like a rope), but is essential to objects like a flat piece of paper. Obi Cloth would be a much better match, as it does model torsion, as well as much more accurate aerodynamics.
http://obi.virtualmethodstudio.com/tutor...amics.html

Still, the mechanics of airborne rolls of paper are quite complex. I'd reccomend not relying on the default collision detection for the unrolling effect (as the amount of self-contacts in a rolled up piece of paper is huge, and would bring performance down), but rather creating a strand of cloth programmatically behind the roll, keeping the roll itself as a regular mesh.