Detect circular flow pattern?
860 2 1- Forests
- Member
- 24 posts
- Joined: June 2019
- Offline
Hi guys,
I would like to detect a circular flow pattern. This hand-painted demo should make it clear:
The measure SOP doesn't do this well. I guess it's because it only looks to the immediate neighbour points to determine the curvature:
Is it possible to do this?
Cheers,
I would like to detect a circular flow pattern. This hand-painted demo should make it clear:
The measure SOP doesn't do this well. I guess it's because it only looks to the immediate neighbour points to determine the curvature:
Is it possible to do this?
Cheers,
Edited by Forests - Feb. 24, 2020 17:32:02
- Aizatulin
- Member
- 475 posts
- Joined: July 2005
- Offline
Hi,
you can try to calculate something like a local winding number for example.
For each vertex get the neighbour vertices +- a number, like (…,vtx - 2, vtx - 1, vtx, vtx + 1, vtx + 2,…). Next step is to calculate the local center of these neighbours.
After this you can use the center to loop over the vertices and accumulate over the angles. The angles should be big near the curls.
Here is an example (using some wrangles). The result attribute is a vertex attribute called @w. This can be used to determine the local curl. But the results will strongly depend on the input (like local vertex density, planarity etc …).
you can try to calculate something like a local winding number for example.
For each vertex get the neighbour vertices +- a number, like (…,vtx - 2, vtx - 1, vtx, vtx + 1, vtx + 2,…). Next step is to calculate the local center of these neighbours.
After this you can use the center to loop over the vertices and accumulate over the angles. The angles should be big near the curls.
Here is an example (using some wrangles). The result attribute is a vertex attribute called @w. This can be used to determine the local curl. But the results will strongly depend on the input (like local vertex density, planarity etc …).
- Forests
- Member
- 24 posts
- Joined: June 2019
- Offline
-
- Quick Links