Vincent, Another method, is keeping the fcurves the way they are, would be to use an ice tree or script to define new user motion on the meshes. So the motion blur is calculated properly.
On Wednesday, May 14, 2014, Vincent Ullmann <vincent.ullm...@googlemail.com> wrote: > Hi again, > > here is another try, using ICE. > Not totally sure if it works in all cases, but it could at least handle my > Animation-Skills. ;-) > > Am 15.05.2014 06:54, schrieb Vincent Ullmann: > > Hi Morten, > > i dont think its possible to solve this in ICE, becouse you cant change > the AnimationCurves.... > (While tipping this i actually have a Idea....) > anyway.... > > I found a little snipped in the Web to get the Objects COG in Global Space > (https://gist.github.com/yamahigashi/9309426) > Now I just wrapped this in a little Loop, and created some PoseConstrains. > > After executing the Script, you just have to > - Plot the Nulls Transforms > - Remove the RBDs Transforms > - Parent or Constrain or Envelope the RBD's to the Nulls > > Hope this works. > Could'nt test this.... just some thoughts. ;-) > Vincent > > > > > Am 14.05.2014 16:22, schrieb Morten Bartholdy: > > I need to find a way to fix fcurves and pivots on a large number of > objects (tumbling boulders) RBD simulated in Realflow (don't ask). When > they come in they move correctly, but their pivots are way off from the > geometry and fcurves suffer from gimbal flipping, so motionblur obviously > doesn't render properly. > > > > I was thinking it would be relatively simple to get good motiondata by > piping them through ICE, calculate motionvectors for each vertex and set > the data again. I guess there is a way to move the pivot of each object to > its centre too, while maintaining the same motion. > > > It is however way beyond my ICE capabilities, so I am looking for > tutorials, pointers, scripts or possibly other tools that might do the > trick. > > > > Thanks > > > Morten > > > > > >