NEED HELP Complex Lofting for Stream Surface Generation

Hey,
I am working on my thesis project in which for one part we are trying to convert streamlines (generated in simscale then exported to paraview, then x3d to lines in blender then imported the lines into rhino and joined to get each streamline as one joined open curve). The complexity now lies in creating a srf that describes the crvs generated. I tried grasshopper construction where i created perp frames, then found the intersection points and tried to interpolate to create a series of sections that i could then loft but the interpolatations missed alot of the points and didnt generate much. I tried unify dir then rebuilt and fitcrv and tried series of lofts and patches but still not retain the gesture of the lines within the end state of the surface. Any idea of how I could make this a surface? Any help would be appreciated. This is a video ref: https://www.youtube.com/watch?v=52KosiinmGk

here are the raw crvs
crvs_help.3dm (311.0 KB)

screenshot of the curves in question:

… well this beautiful ball of wool does look like a extraordinary modelling challenge.
for sure not a click once standard nurbs command.

did you read the paper that is linked under the video you posted ?
Matthew Edumunds, Tony McLoughlin, Robert S. Laramee, Guoning Chen, Nelson Max, and Eugene Zhang, Automatic Stream Surface Seeding, EUROGRAPHICS 2011 Short Papers, pages 53-56, 11-15 April 2011, Llandudno, Wales

If you re lucky @DanielPiker has some ideas

1 Like

Hey, thanks for the reply, yea I checked it out however that is more of a c+ and open gl approach and many of the processes dicussed seemed to be the process that a modelling software undergoes when it creates a loft. I would specify that potentially the wool ball could just be the general shape and the others would be the focus— or if there was some way to apply a stream ribbon approach.

I could also lower the amount of resolution of streamlines (I.e) less crvs or more crvs I just had 100 to get a general idea of the form

If you want results like from that paper (https://web.engr.oregonstate.edu/~zhange/images/seedingpaper.pdf), I think you’d need to implement something at least roughly along the lines they describe (independent of the particular language), and from a glance it looks significantly more complex than simply taking the curves you have and lofting.

It looks like the seeding is an important part - you probably want to get collections of streamlines starting at points along a given curve, but it sounds like you have that part happening upstream of Rhino, so you’d need to figure that out there.
You might be able to just import a very dense collection of streamlines then select all those that pass close to a given curve in Rhino, but I imagine that would get extremely heavy to produce reasonable resolution.

Once you had a collection of curves starting along a common seed curve, they won’t be simply structured, so probably the easiest way to get a surface would be something like Shrinkwrap or Dendro.

1 Like