Help The easiest way to create complex rope?

[josemendez]

Hi josemendez,
Thank you so much for the quick response! 
This image shows how our idea operates:

The ropes first go through each holes on the frames, which then separates to create a 3-D structure.
The length of the rope is not fixed because we leave one end still on spool so it can length.
It is a idea of transformation from 2-D pattern to 3-D structure.
The rope's self-collisions are dependent on whether one thread is on top of the other before pulled apart. 


And the apple was a playful thing, not a part of the project. 
What really matters to us is the unpredictable nature of transformation from 2-D to 3-D.
we aim to study the cause-and-effect relation.

Could you say more about using spline to make Obi Rope?
Which documentation or link should I look at?
I currently only know how to use Obi Rope(fully setup).

If "one continuous thread simulation" won't work: 
Because the thread is continuous, theoretically it has the same tension distribution along its entire length if disregard frictions. 
Other than simulating it as one thread, could we make each thread connecting two holes an individual thread, but with constant tension? 
I imagine they can automatically lengthen when reach a predefined tension when pulled apart.


Thank you so much again for your response!
Best regards,
Po

Hi,

The problem with your setup is that high-tension ropes and self-collisions do not mix well. Obi uses an iterative position-based solver at its core, and represents ropes as chains of particles joined by constraints. This means that the longer a rope is, and the higher the tension it must endure, the costlier the simulation. If you add accurate self-collisions as a requirement under these circumstances, the cost skyrockets.

It could be said that your project is a worst-case scenario for any traditional realtime physics engine: thin, long rope under high tension with lots of fine self-collisions. My advice would be to ditch traditional iterative solvers (which are the most common in games, and which Obi uses) in favor of other approaches: direct solvers & capsule based collision detection, cable simulators (http://matthias-mueller-fischer.ch/publi...Joints.pdf, http://www.physics.umu.se/digitalAssets/...hybrid.pdf), or a custom approach.

It might even be possible to derive a fully analytical model of your cables, that calculates the final shape as it is only determined by the order in which cables cross each other in their rest state (which ones pass on top of others). This would be the fastest/simplest approach since if does not involve simulation at all.