Just checking if using PointDeform is the best way to tetrahedralize a deforming mesh for FEM? See attached.
Thanks!
Deforming mesh for FEM.
4009 8 1
-
- anon_user_37409885
- Member
- 4189 posts
- Joined: 6月 2012
- オフライン
-
- tamte
- Member
- 9418 posts
- Joined: 7月 2007
- オフライン
-
- anon_user_37409885
- Member
- 4189 posts
- Joined: 6月 2012
- オフライン
-
- tamte
- Member
- 9418 posts
- Joined: 7月 2007
- オフライン
Maybe in a simple and very isolated case, but yours seems to be tetrahedralizing deforming geo every frame which can result in different amount of tetrahedrons which will break your pointdeform and on top of that it's much slower, maybe try with heavier geo and more extreme deformation
Tomas Slancik
CG Supervisor
Framestore, NY
CG Supervisor
Framestore, NY
-
- anon_user_37409885
- Member
- 4189 posts
- Joined: 6月 2012
- オフライン
-
- tamte
- Member
- 9418 posts
- Joined: 7月 2007
- オフライン
-
- anon_user_37409885
- Member
- 4189 posts
- Joined: 6月 2012
- オフライン
-
- johm
- Member
- 138 posts
- Joined: 12月 2010
- オフライン
hi artye
as Tomas pointed out, you don't want to tetrahedralize a deforming mesh, rather, you want to deform a tetrahedralized mesh. This is the only way to guarantee temporal consistency with the tet generating utilities.
Deforming a tet mesh via the PointDeform sop is perfectly valid, as is any other type of deformer, as long as the shape change isn't so extreme that it causes tets to invert or deviate so far from the original shape & volume that it becomes nearly impossible for the solver to give you back something reasonable.
*By “any other deformer”, I literally mean any sop that is capable of pushing points around in houdini. Twist, SoftTransform, Bone Capture/Deform, whatever.
By the way, in your original setup, if all you really wanted to achieve was the twist, you could conceivably simplify your network to be Torus > SolidEmbed > Twist.
as Tomas pointed out, you don't want to tetrahedralize a deforming mesh, rather, you want to deform a tetrahedralized mesh. This is the only way to guarantee temporal consistency with the tet generating utilities.
Deforming a tet mesh via the PointDeform sop is perfectly valid, as is any other type of deformer, as long as the shape change isn't so extreme that it causes tets to invert or deviate so far from the original shape & volume that it becomes nearly impossible for the solver to give you back something reasonable.
*By “any other deformer”, I literally mean any sop that is capable of pushing points around in houdini. Twist, SoftTransform, Bone Capture/Deform, whatever.
By the way, in your original setup, if all you really wanted to achieve was the twist, you could conceivably simplify your network to be Torus > SolidEmbed > Twist.
-
- anon_user_37409885
- Member
- 4189 posts
- Joined: 6月 2012
- オフライン
-
- Quick Links