Can someone explain how can this happen? Same topology, same exact lines, I mesh it with meshFromLines and I get one result and then again I undo, mesh it again and I get a different result. The file is attached. Untitled.3dm (42.3 KB)
I select them with either selCrv or Ctrl+A. Now that I opened the file again, it is consistent. But many times, there is no consistency. I found out the reason is that in different times Rhino creates different sets of extra lines as it thinks best to create the faces, therefore different sets of faces are created as well. When the user is not 100% explicit with the interconnection of the lines, Rhino takes control but there are many times that it is not consistent.
In general, the command “meshFromLines” needs a lot of work to perform well. At the moment it has poor performance
@piac - any ideas about this? I thought perhaps pick order might have something to do with it but with Ctrl-A or SelCrv would that order curves inconsistently?
@Theofanis - are there other examples that you could give us to illustrate your last comment?
Sure, pick order makes a difference in the output from MeshFromLines. Also a single line can give an entirely different result. Please note that Rhino is one of the few programs on the market that can re-create a mesh from 3d lines. We optimized MeshFromLines for cases where there are hundreds of lines. We did not optimize it for twisted, concave shapes with few lines. This is more like an art, because there is no unique solution in general.
Again, if you want to give any sample, this might help us improve in new versions of Rhino. We will add it to the sample set.
Finally, if you are not going to provide any sample, keep these suggestions in mind: always choose the minimum _MaxFaceValence that will work for your case. Make sure lines meet at ends. Do not add cross-intersecting lines. This will help to create better results.
Thanks,
Giulio
–
Giulio Piacentino
for Robert McNeel & Associates giulio@mcneel.com
I guess you already know this is not the first time I am giving you problematic cases for “meshFromLines”. I already provided a sample here, in this post. What concerns me most is that these differences in the final shape when I use the command “meshFromLines” it’s kind of abstract. Sometimes I have it, I close the document, open it next day, and then it works “properly”. Next day it might not work again… I always select the lines using selCrv or SelAll or Ctrl+A.
I know that Rhino is one of the few programs with this functionality, and that’s why I use it, otherwise I wouldn’t. Creating 3D shapes from lines is a very powerful tool for designers, especially for those they really know how the underlying mathematical background of splines work. It’s a very intuitive process… For me, who I happen to have a PhD in splines it’s so straightforward to create 3D shapes from lines. It’s much easier for me this way than with the well-known and widely used techniques people use for 3D design.
In addition, I find ways to make my mesh more specific, and give Rhino very less space to perform randomly.
I will keep posting problematic cases as soon as I find them.
Thanks for the tips! Really appreciated! Didn’t get the last part though… cross-intersecting lines.
That face is very skew/nonplanar, so, at present, it’s more “invisible” to the algorithm currently used by the command and it thinks it’s not wanted. Meshes have faces that are, generally, tending toward planarity.
This face has 5 sides:
In the new example, these faces are highly nonplanar, so the algorithm thinks they are unwanted:
You may be using this code for purposes for which it was not intended. I’m interested what your goals are, here. What type of industry are you serving? Where do these lines come from, and why has your mesh so nonplanar faces? This might help us improve the command in future versions, to serve your needs.
Thanks,
Giulio
–
Giulio Piacentino
for Robert McNeel & Associates giulio@mcneel.com
@piac Hi again. I am sorry for the late reply, I had some internet connection problems.
In general I am using “5 max sides per face”. I am almost sure that if I had developed a model with only quadrilateral faces, the algorithm would perform better, not sure if it would perform perfectly though.
Not having the ability to deal with non-planar faces, it’s a huge limitation for an algorithm. If I can help you in any way to work on this, please let me know.
I can talk to you for my goals, industry, where these lines come from, and many more things in private. Let me know if you are interested, I am.
Before we go in private, I would like to ask publicly and let anyone know if the algorithm takes into consideration the direction of the lines (starting and ending points).