Getting to the Root Cause of My Problems With Primary Surfaces

[Carl_Mesrobian]

I have been spending too much time on a basic concept that I don’t get, and would really like to hear from others about planning workflow. I’ve tried to “make it easy” by using someone else’s mesh input, but I have specific profile shapes that I want to use from vintage guitars that I own. Here is a previous post that I started

Is "Slicing" a Mesh Possible

but then realized I’m avoiding what I really intend to do, and subsequently caused more work for myself.

Here is the underlying problem. Venetian cutaway in an instrument (see pic).
The lower bout, waste, and upper bout are duck soup, BUT the area in the cutaway/neck area is the killer for me. @pascal has helped me with a design he posted, but I keep getting stuck with the neck/cutaway area on the picture. Also there is also a flat margin about 1/2" wide or so all around the edge.

I watch Sky’s videos over and over, but either I’m trying to put too many CV’s in a given primary surface, or should I just make rectangular surfaces, shape them and call it ready to post. Then the question is how many verts per surface, and how large the surfaces? And where to put a trimmed corner to get 4 edges for NURBS, or, Yes I’m ready to try SubD @theoutside :-). Saw a good video on it by Phil Cook.

My models are only for production, not for presentation or concept. It will post to G code for my CNC machines, which are a small mill, and router. I don’t care about edge treatment, as a profiling program will do a trimmed surface for me. I can project profile vectors to the surface in my CAM software.

I’m making this more complicated than it has to be! No I do not own a 1960 ES-335 like Tom’s - I wish!

[pascal (Pascal Golay)]

Hi Carl - if you can post a model with the surfaces in the area you’re having trouble with, somone can take a look and maybe have something useful to offer.

-Pascal

[Carl_Mesrobian]

CM35V5.3dm (328.0 KB) Cutaway area

[theoutside (Kyle Houchens)]

if you want to have a go at subd - this may help.

Rhino 7 Windows - SubD guitar full build

[jim]

I don’t know much about the design of guitars but I would try to make this as one surface using point editing to create the shape.
one_surface.3dm (398.3 KB)

[theoutside (Kyle Houchens)]

patch actually works nicely here… (sky’s gonna kill me…but result is result…)

patch test.3dm (277.4 KB)

[Carl_Mesrobian]

Can you save as Rhino 7 ? Pretty please?

[Carl_Mesrobian]

Did you patch and then trim. or patch the profile?

[theoutside (Kyle Houchens)]

patch test.3dm (277.4 KB)

here you go

simply patched all 3 curves.

( I used 20x20 as the point count)

if you do it with history, you can adjust it a bit-

[Carl_Mesrobian]

Thanks - I’ll try it.

[Carl_Mesrobian]

This is what I’ve come up with, as a starting point (no pun intended). These are all just lines and curves, split at the intersections. I used EdgeSrf but the isocurves are funky, my point counts are all over the place (degree 2 to degree 5), so I’ll clean those up and try EdgeSrf again.

I’m guessing this will be a more controllable way to get that gentle mound like in the picture.

I can resort to pouring a mold over my 60’s guitar, praying that the exothermic reaction doesn’t destroy the finish, then casting a model, then pouring over that to get the final mold for vacuum pressing.

Or use Rhino and get a good model made, then use the “negative space” (underside) for the mold I need. Another thought I have in mind is to trace cross section curves from the original, which I probably will do, and scan those to trace in Rhino.

My vote is for Rhino

[Carl_Mesrobian]

Anyone?

[davidcockey]

How critical is adherence to the doctrine of single span surfaces only?

How much will the finished surface vary from the milled surface after smoothing and sanding?

[Carl_Mesrobian]

I would like to minimalize removal after machining. My machine is within .005, so that’s probably a good metric to go by. I usually stepover between .03125" and .0625".

Single span with minimal verts seems easier to edit while developing the shape I need. I tried Sweep2 and got a dense point map that would be tedious to edit, so the process becomes
1)draw a curve
2)Sweep2
3) Look ok? If yes, move on, if no delete the surface, edit the curve in 1, and got to 2
So I tried a new method (see below)

I’ve actually begun another angle of attack. Draw a rectangular surface, say 5 x 5 degree, or similar. pin the outer verts and others to z=0, then pick verts in right and front views to change z on selected verts. The profile will be trimmed by a 2D program on the CNC, so modeling the edge is not necessary - a 19x9 rectangle is plenty, mirror it, post the CAM, then post a 2D profile of the body shape

Then I will create and adjacent 5x5 degree and shape it, and match , Sky would be proud, I hope! I learned to draw period furniture full scale with a parallel rule and triangle, so this method doesn’t bother me

My only concern is the points so scrunched together where I want a quick change , but I guess all is fair to reach an end? I am trying to copy the 1960 guitar in the picture in the post. I doubt G2, etc were even thought about 65 years ago. (Maybe they were?)

I have molding and casting resins and could bang this model out in a couple days, but would not have the capability to use it to derive other models from. Seems like, after using a 3D CAD program , I don’t want to go back to the flat earth way of doing this! And molding and casting would not work for carving wood!

[sgreenawalt (Sky G)]

So what you’re really fighting here - other than guitars being WAY more difficult than people think at first blush (although I’m pretty sure you know that now!) - is the mismatch of a technology that’s inherently quad based - NURBS - trying to deal with something that does not lend itself to a quad patch layout - arched guitar tops. It’s not that I think that Rhino is the wrong tool for this - it’s that I think NURBS are. For many parts of a guitar, NURBS work great, but for a handful of select areas, they make life far more difficult than it needs to be. I once did a whole series of necks from scan data for Santa Cruz guitar - they wanted to take master neck carvings done by their lead carver, and make CNC toolpaths to allow volume production. I very quickly started to respect just how difficult necks can be - the heels and the transition from the neck to the headstock. I was using T-Splines at the time, and for me at that time that was a viable solution. I can certainly do those shapes using classic patch modeling now, but good lord the knowledge base to do them well in NURBS is an order of magnitude more than doing them in SubD. So you’d think I may be recommending you do these in SubD and well…sorta. Let me explain.

Assuming the goal here is to create toolpaths for CNC machining of bodies - you may find it’s easier to work with a hybrid software/technology approach. Find the right tool for each portion of the guitar - approach it piecemeal. For instance, defining your outline curves in top view is perfect for NURBS and I wouldn’t do it any other way. So long as you can get all your pieces together into your CAM package in alignment, life is good. Every CAM packaged I’ve ever used ultimately uses meshes, even when it looks like they are using polysurfaces. You find this out real quick when you cut a model on your CNC and end up with faceting, and realize that it’s pulling the display mesh of the polysurface for the toolpath generation, and that your display mesh needs to be more dense for smooth cutting. So, anything that makes a mesh, that can be put well enough in alignment with your other geometry, I’m saying is a valid approach. You’re dealing with wood, so that mesh only needs to be smooth enough to allow for hand finishing. Bonus - a guitar top doesn’t have that much curvature. You only need a mesh that’s good enough, that gets you to your next step. While this may have you saying “hey maybe V8/Subd” I’m saying…yeahhhhh maybe but maybe not. The toolkit is still so limited that it’s maybe not very efficient to work this way. Lack of falloffs/blend modes, in the year of our lord 2024, I’m looking directly at you with a stink eye. I would suggest two other pieces of software that might be able to handle this better - Blender or Zbrush. I don’t know any Blender power users, but for sure they are out there. Kevin Pasko is a friend of mine and an absolute monster at Zbrush - a true silverback gorilla of mesh modeling. If you want his contact info I’ll PM it to you. The approach I would take is to find a way to export the relevant geometry to Zbrush or Blender, create your arched top mesh for CNC, and then toss it back into Rhino. Say, make sure you send over geometry that shows how much above the edge level of the guitar you want the “peaks” of your features to kiss. Close enough and TLAR is the goal - no one is going to be taking a set of calipers to your guitar and saying that you model showed it was supposed to be arched 1/64" more. This is the sort of thing that once you have the right person and the right workflow should be less than an hour. Well worth your money if you’re spending it. Remember the goal here is any valid object to feed into a CAM package, not a class A surface model for us geeks to drool over. I mean, that would be cool too, but eyes on the prize ya know?

Hopefully this helps!

-Sky

@theoutside

patch actually works nicely here… (sky’s gonna kill me…but result is result…)

NU FONE. WHO DIS?

[David53 (David)]

If you don’t have the need for whatever a Class A surface is supposed to be; does this degenerate into a problem you can optionally buy your way out of using something like Xnurbs; create your base geometry concept style using curves & Xnurbs, add fillets, quad remesh, and smooth accordingly?