NURBS to .obj export failure

Serengeti (Rhino WIP)
1

I thought I’d better extract the .obj failure from my thoughts on meshing issues in the PaperCuts thread.

Standard mesh exports.3dm (1.1 MB)

Check the .obj imported mesh in WIP6 - I created a simple shape, exported it with standard .obj settings from Rhino and imported it back into the Rhino file - it’s a bad mesh.

Cheers

Andy

2

No one?

Seemed like a simple ask from the program… The NURBS model was quite basic but the .obj export created a load of non manifold edges.

Andy

3

Hi…maybe…
I think it’s because the default ngons options is checked. Uncheck it by default.
Do not create ngons.
ngons were evil at a time, and it can still be (some 3d render engines do not accept ngons).
You have to get the full control of meshes each time it’s possible. Using the advanced preview is recommended.

I I would recommend to disable this option by default to the V6 developers.

4

Yeah, that sorts it - it punts out the same mesh as the standard Mesh command within the program this time.

5

Hi Andy - I see that, thanks. The orientation is correct if both Import and Export use the same setting for mapping obj’s Y to Rhino’s Z

-Pascal

6

Yeah, I used the basic setting for each mesh/export/import but that one was different from the rest as standard, I had to change it manually to get the same result. It’s not a biggy but only realised after I’d posted the first time.

Andy

7

Hi Andy,

Thanks for the model. This will work better in the next release of the WIP. The problem was actually with OBJ import. The OBJ file was fine and should have worked with any app downstream app that handled more than 4 vertices on a face. The importer was getting the same answer for the triangulation of the ngon as V5 but was screwing up the indexes when it made the faces. If you open the file in the next wip you’ll only see 4 degenerate faces. Technically they’re valid because the vertex location and indexes are unique for the faces. But the vertexes are also colinear so the faces have no area, normals can’t be properly calculated, etc. I’ll look into that sometime in near future. Here’s a YouTrack issue for this if you’d like to track the progress.

https://mcneel.myjetbrains.com/youtrack/issue/RH-35613

Cheers,
Tim