Extraordinary vertices problem

Grasshopper
1

Hi! I am trying to turn the following subd into nurbs. The problem is, the topology isn’t good. As a result, I get the following effect:

Screenshot 2023-06-05 at 11.40.10 PM
Screenshot 2023-06-05 at 11.40.10 PM1414×1426 133 KB

This somewhat to be expected, because Rhino’s documentation says this is an (extra-)extraordinary vertex. And it’s caught in a bad spot too. It looks okay (zebra-stripes-wise) in subd but, once I go into nurbs (both the method show in gh and the ‘tonurbs’ in Rhino), the above problem happens.

From doing retopology by hand, I know the above extraordinary vertex could have been an ordinary one with some topology fiddling and therefore not have the squiggly nurbs problem. Is there a way to achieve something similar in gh? (letting the code do the work)

Thanks a lot!

2

The original mesh is the denser, more wrinkly one
The current remesh result is the second mesh

In discussions about triangle remeshing, I saw that some triangle remeshing are better than others because they have less vertices with many edges (less extraordinary, I guess?). It seems there is only one way to quad-remesh (comes with gh) and I can’t find a way to convince it to go easy on the extraordinary vertices :skull:
extraordinary.gh (3.7 MB)

3

  1. have you played with all the options for how to deal with extraordinary vertices in the _tonurbs command?

  2. there are a lot of options for quadremesh, both in rhino and in grasshopper - there’s a separate quadremesh settings component in grasshopper. And playing with those settings and understanding what they do can often help a lot to get better remesh results. However, if you feed quadremesh your bad topology and you don’t set it up to use significantly fewer faces than you did, it could still show a glitch in that bad spot because it’s trying to mimic your bad topology.

  3. if you were retopologizing manually, and you got an 8-pointed star vertex, you can probably do better manually. Why not a 4 point? just take out half the edges. Also, read this: https://topologyguides.com/ and read any other links on topology that other people post in reply (I’m hoping they post some good ones). Sometimes pentagonal faces convert to nurbs better than quads that meet at a vertex with too many edges.

1 Like
4

Thanks for your reply and insights into topology, Max!

1 this made me lost continuity but is better than having a weird bump
2 I played around with the quadremesh settings but, it looks like the program doesn’t want to move the extraordinary vertex
in particular, whenever there is a y-joint (one pipe breaking into two pipe), an extraordinary vertex appears, and doesn’t go away with different settings
having good starting topology makes a lot of sense. But in this case my mesh is quite dense and messy–perhaps the quad remesh code picks up some ‘hints’ on topology that the human don’t see
3 that pentagon faces fare better is interesting. I wonder what happens if quad remesh can make use of it.