[Bf-committers] Softbody revisited :)

bjornmose bjornmose at gmx.net
Fri Apr 1 19:13:32 CEST 2005

Ton Roosendaal wrote:
> Hi,
> This is all going to be too complex... what I try to find is the  
> simplest access possible, with the least settings & variables involved.
> Currently I'm halfway coding work for softbody (like moving data out of  
> object into a persistant SoftBody struct), so will play a bit with all  
> possibilities to find something optimal.
> My prediction is that we need 2 different things;
> 1- a simulation stage for finetuning all settings, for all frames, with  
> instant updates
> 2- a method to bake the result, and edit that (like allow layered in  NLA)
> What I was talking about is only stage 1. Below some questions still;

define "the above solution doesn't work"
humm.. think of a object IPO defined as bezier and you set (random) 
positions every frame, then the resulting bezier will shoot over in 
between the frames.In this case the simulation only 'sees' the position 
of the keys  and not the overshooting in the movement of the goal. So 
the worst thing that will happen is that the reaction of the SB object 
becomes shallower than it would be physically correct. There is no need 
for resetting  or such.
This is a very theortical situation and i think we can savly ignore it.


> If you want initial velocities you just create an initial animation  
> that cause such velocties. Then - after baking - you should be simply  
> allowed to to cut off that part.
Sounds simple and good.


More information about the Bf-committers mailing list