Hi there!

I'm enjoying MoI a lot and have started to get the hang of coaxing nurbs surfaces into the shapes I want. I often run into weird boolean situations where I have to try a couple of the different tools, then possibly have to fix things up a bit afterwards (delete extra faces, do another trim, etc.). That's totally okay, and rarely breaks my flow.

I just ran into a situation that I can't explain and can't seem to work around. I have a solid that's become somewhat complex. Now I want to cleave it into two solids with a plane. Initially this appears to work, and I can make two halves with open sides, select the edge loops and close them with Planar. However, the results are not Solids, but Joined Srfs. There appear to be four naked edges (two pairs), but the naked edge script finds 6 on the top half and 9 on the bottom (the extras must be so small I cannot see them, as if I do random window selects, I'll eventually grab them.

Original object with trimming plane:


Bottom after trimming and closing (naked edges selected):


Top after trimming and closing (naked edges selected):


Is there a way to repair this, or maybe general guidelines to avoid creating geometry that results in this problem? I've attached the 3dm file if it helps.

Thank you for any suggestions!

Chase

Hi Chase, I think the problematic stuff are a couple of surfaces with squashed together control points near their ends, like this one here:





Instead of that you'd want something more like a larger trimmed plane for this whole area here:



The surface/surface intersector does not seem to be getting a very good intersection on that one, it's stopping just a little bit before it hits the edge:





But it's kind of an odd surface because it's flat yet it has a lot of internal twisting within it.

it can be better for stuff like this to be a trimmed out portion of a larger more relaxed surface rather than one patched in by being directly constructed to an non 4 sided boundary.

- Michael

Hi Chase, ok I guess I see what makes that area kind of difficult is that it's slightly tapered and the tapering starts right at this spot here:



So that area is not all one single plane.

- Michael

Or sorry that wasn't right the whole line there is tapered it's just at such a slight amount I didn't realize it.

Anyway the part that I'd probably try is to get that area filled in with one larger trimmed plane instead of having that twisted and pinched piece.

- Michael

Thanks, Michael.

That entire area there should definitely have been planar. I went back through the construction, and it turns out my initial top profile was definitely not optimal; composed of multiple collinear segments, when a single segment would do. I think a similar cause might be leading to the naked edges at the nut (opposite, narrow end of the neck), as well.

I'm simplifying those curves and focusing on doing larger planar sections where applicable. I see it's definitely easier to get lax about the underlying topology compared to something like subsurf or direct spline patches where you live and breathe the grid :D

Hi Chase, another problem area was that there were a few micro edges in this area here:


I've attached a .3dm model here where that one coplanar area has been replaced with a single larger trimmed plane and the micro edges removed by a process of untrimming the surrounding faces and retrimming them.

This version should get a clean split by the plane now.

- Michael

Hi Chase, yes definitely it's good to make the initial curves to be as simplified as you can, any excess segmentation or stuff like that will inherit down into later processes pretty easily.

One thing to watch out for in general is building stuff only by filling in a 3D wireframe in a kind of "patch by patch" method. Although it's an easy to follow strategy you can end up with bad quality surfaces that are stressed or bunched up awkwardly in certain areas. That then makes some operations like calculating surface/surface intersections more difficult.

So when possibly you want to have more relaxed surfaces that look more like a net of quads. It can have a bendy overall shape but if it's pinched or twisted that's not so good. Then the boundaries for those shapes come from cutting them with either profile curves or other surrounding shapes (that extend past each other a bit) instead of always doing a loft or sweep to make a surface directly hugging the boundaries. Hope that makes sense.

A guitar neck is actually a pretty difficult model shape too.

But basically an important element to good quality NURBS models is having simple underlying surfaces that have more detailed boundaries on them coming from intersections rather than all boundaries being directly part of a surfacing operation.

- Michael

Hi Michael,

Definitely a lightbulb moment, there! My takeaway is commit to the booleans, rather than patch modeling. Now I realize that should've been more obvious, as "network" is just one tool, while there are half a dozen devoted to smooth lofting, trimming, etc. It's really more intuitive, anyway. I'm getting the hang of trimming, retrimming, etc. and it's quite different than anything I've done in solid and polygon modeling previously. MoI is the first NURBS tool I've tried that doesn't get in the way of itself so I can actually focus on the model enough for things to begin to make sense.

Thank you for your thoughtful reply--every sentence is full of insights. I think I'll reapproach the same model from scratch after absorbing some more videos, etc. (though of course your repaired model is working).

Hi Chase, you're welcome!

There are some tips here for people coming from a poly modeling background which may help some too:
http://moi3d.com/forum/index.php?webtag=MOI&msg=4865.2

But yes the main area where NURBS modeling really shines is when you're mostly using 2D profile curves where some profiles generate extended base shapes and others as cutting objects for booleans. This strategy fits particularly well with mechanical parts.

With a guitar neck you're getting a little bit out of the zone because of some swoopy and blending shaping will need to draw from the 3D surfacing toolset for some areas but you want to limit that when it is possible. You'll be hitting a somewhat steeper learning curve.

Also another recent discussion about switching a particular model from a "patch by patch" fill in method to a "build extended shape then boolean" method:
http://moi3d.com/forum/index.php?webtag=MOI&msg=10210.16

When your model is more melty/blendy shaped all over that's when it becomes difficult to drive the design using 2D profile curves and then that's better to do in a poly modeling program with sub-d modeling instead of with NURBS.

I know it can be difficult at first because it's such a different strategy than what you're used to with poly modeling. But the payoff is that for the right type of model it's super fast. You form more of your model by 2D drawing instead of managing a ton of vertices in 3D space.

- Michael