BodyRope physics object

[ulf.schroeter]

Proposal

 

Addition of BodyRope physics object similar to BodyCloth object for efficient simulation and artifact-free rendering of small-diameter ropes

 

Reason

 

Simulation of small-diameter ropes based on BodyCloth leads to uncontrollable rendering artifacts (e.g. rope mesh twisting see attached screenshot) and numerical instabilities/jittering especially in case of stronger rope bending (e.g. at pulley). Also realistic modelling of inductile and/or stiff ropes (e.g. steel cables) is not well supported.

 

Simulation of ropes with high number of small capsule shapes connected via ball joints is cumbersom, most probably not efficent and dynamic mesh creation/update for rendering has to be handled by the user itself.

 

Features

 

• simulation of rope physics compareable to BodyCloth but based on simplified particle-chain model
  
• BodyRope object attachment via JointPin like for BodyCloth
  
• rope mesh cylinder geometry based on spline-interpolation of simulated element particles with smooth, curvature-adaptive tesselation and avoidance of mesh twisting
  
• support for rope breakage in case of joint overload
  

 

[frustum]

I will try to implement BodyRope in september.

 

Spline will be used as reference geometry and mesh will be generated in runtime by specified radii and tessellation levels.

[ulf.schroeter]

I will try to implement BodyRope in september.

 

Spline will be used as reference geometry and mesh will be generated in runtime by specified radii and tessellation levels.

 

Great !

[frustum]

http://unigine.blogspot.com/2010/09/bodyrope.html

 

Rope is ready.

[ulf.schroeter]

This looks MUCH better than BodyCloth ! Implementation speed is incredible !!!

 

Does the implementation also properly support nearly inductile ropes (e.g. steel cabels in a crane simulation), where even heavy attached weights causes nearly no rope stretch ? In the videos the ropes seems to be little bit like rubber bands.

[frustum]

This looks MUCH better than BodyCloth ! Implementation speed is incredible !!!

 

Does the implementation also properly support nearly inductile ropes (e.g. steel cabels in a crane simulation), where even heavy attached weights causes nearly no rope stretch ? In the videos the ropes seems to be little bit like rubber bands.

 

Stretching of rope depends on number of iterations and rope/body mass proportion. Joints in rope is very simple and we can setup 32-64 of internal iterations.

[ulf.schroeter]

Stretching of rope depends ... rope/body mass proportion.

 

hm, getting the right effect might be very hard if it's modelled as some kind of mass-spring-system like BodyCloth. I think that most use-cases for ropes (exception might be bungy jumping simulation) would like to prevent rope stretch even for very heavy loads. Therefore I would expect that a physical rope model is more like a large number of rigid capsules connected via fixed ball joints (= no/nearly no stretch) than a mass-spring-system (=stretchable, good for cloth and rubber band effect, but bad for ropes)

 

But maybe I am too pessimistic :-) I will just try it and thanks a lot for the super-fast implementation !

[ulf.schroeter]

Hi UNIGINE,

 

the new BodyRope object is a GREAT improvement ! Only remaining problem at the moment is missing damping of vertical rope swinging amplitude after movement of pinned rope attachment points. Is there any way to force much stronger damping, so rope stops swinging much quicker ?

 

[frustum]

I will add linear velocity damping into the Rope and Cloth bodies.