Hi,
I am trying to stick / align geometry over time to a deforming geometry. I am using the polyframe sop to extract a transform, however as the underlying geometry deforms, the output of the polyframe can flip. I have attached a simple example - you can see the undesired behaviour around fr 130. Has anyone encountered this behavior before or found a good workaround? The fur sop must be solving the issue internally, but unfortunately I can't pry it open…
Thanks for any suggestions you might have!
Cheers,
Carsten
sops: stable polyframe over time?
2313 4 0- carstenk
- Member
- 84 posts
- Joined: 1月 2013
- Offline
- circusmonkey
- Member
- 2624 posts
- Joined: 8月 2006
- Offline
- carstenk
- Member
- 84 posts
- Joined: 1月 2013
- Offline
Thanks for the reply!
I realize that my initial assumption was wrong and the polyframe is actually very stable - so my error must be in the way that I am constructing the rotation quaternion from the polyframe attribute I want to use the align the geometry. Which brings me to the second thing that was wrong, which was my test scene (I forgot to add the step where I am converting the tangent and normal information into a rotation quaternion).
Rob - in your example you used the normal to build an up vector instead (and then it is up to the copy sop to contruct the orientation matrix from the up and N) - I basically tried to do what the copy sop does internally.
I've attached my updated example that highlights the actual problem (I presume in the polyframe to quaternion vopsop). Notice how the red cubes change their orientation at certain frames.
Sorry about the initial confusion.
I realize that my initial assumption was wrong and the polyframe is actually very stable - so my error must be in the way that I am constructing the rotation quaternion from the polyframe attribute I want to use the align the geometry. Which brings me to the second thing that was wrong, which was my test scene (I forgot to add the step where I am converting the tangent and normal information into a rotation quaternion).
Rob - in your example you used the normal to build an up vector instead (and then it is up to the copy sop to contruct the orientation matrix from the up and N) - I basically tried to do what the copy sop does internally.
I've attached my updated example that highlights the actual problem (I presume in the polyframe to quaternion vopsop). Notice how the red cubes change their orientation at certain frames.
Sorry about the initial confusion.
carsten kolve - ds @ image engine
- carstenk
- Member
- 84 posts
- Joined: 1月 2013
- Offline
- circusmonkey
- Member
- 2624 posts
- Joined: 8月 2006
- Offline
-
- Quick Links