Was wondering if there is a way to sudivide a mesh (so that it gest smoother), directly Rhino or Grasshopper, without changing the positions of the initial vertices. Sort of like interpolate curve works, that makes sure the curve passes through the input points.
I want to make a topography using points measured on site and smooth it so that it looks more realistic and has more detail, but making sure that the resulted surface passes through the points that were measured (if I use something like weaverbird subdivision, that doesn’t happen).
Thank you for your time!
The classic approach on subdivision keeps the OEM Vertices intact. Imagine a triangle from a given Mesh Face Vertices triad/quad … and then mini triangles/quads due to some subdivision Loop (1: 4 mini faces, 2: 16 mini faces etc) meaning that the start vertices are kept the same. So … for obvious reasons … I would recommend to combine subdivision and some Laplacian smoothing (but that would alter your survey nodes) for your terrain.
There’s components that attempt to do that … but if you want some C#'s that do things like these (Relax, Subdivide, Laplace … blah, blah) notify.
BTW: For the record in case that you can read code the Laplacian algo is this:
BTW: For the “general” terrain case (whatever this means) … I have reservations for the Butterfly way > would yield somehow “odd” results. Is maybe more realistic an GC/L SD while measuring the initial survey vertices deviation > if this is within some acceptable Interval > why bother?
Thanks, Daniel! Unfortunately the component doesn’t return any result (tried with different settings on different meshes). The plugin is probably too old…
Thank you for te detailed reply! Not sure I understood correctly, but the last two screenshots depict a version that uses a custom made C# component that actually does what I was looking for? And if so, could you share it?
But that C# keeps (as an option) only the naked Vertices intact/fixed. You said that you want all the Vertices fixed. That said if I was you I would rather search/exploit ways to convert the Mesh into a Nurbs Surface/Solid (I hate working with Meshes for AEC purposes not to mention that you can’t see a thing or two when you are in Wireframe mode).
On the other hand implementing in a C# the Butterfly approach is a 15 minutes job (the only thing required is a knowledge on Mesh Connectivity: FE/EV etc … plus some elementary List (of type int) Intersections - see Stanford d 5/6/7/8 Vertices).
