Archtop guitar, real world units and import / export

* Well, this (my sample above) was a :sick: bit simple minded.
* I actually wandered into the workshop of a local luthier (no guitars, but classical violins) and looked at samples of German Baroque string fiddles and an Armati / Guarneri violin in pieces.
* I still have no knowledge of the construction of Gibson / Gretsch et al archtops, but I can assume the geometry is related.

* A simple loft of linear splines seems useless.
* Splines will need to be closed and at the perimeter angular curves. This will allow a precise control of the curvature and thickness of the archtop generated by the loft creator. This is a quite laborious and extremely fiddly fine tuning of splines.

* You will need to acquire significant skill and experience to produce a 3D model which is useful as input for numerically controlled machinery.
 
Last edited:
* To the best of my evaluation. Cheetah 3D can achieve the results you require for running a numerically controlled production system.
* You will need to fiddle with dimensions / grids / rasters if your primary plans (Gibson / Gretsch or whatever) use inches / fractions thereof. I also suspect that such plans are copyright, but that is your problem.
* However, this is quite complicated stuff and requires some experiments.
 
*
* However, this is quite complicated stuff and requires some experiments.

Thanks for the heads up, Helmut. I’m going to keep wrestling with it for a while, but at the back of mind there is always the temptation just to physically carve the mould for the wood laminate.

FYI, it’s my own design, but I don’t think you broach any copyright laws making Gibson or Gretch or D’Angelico copies unless you try to sell them.
 
Helmut is right, so far I can't find a way to do this without a lot of work.
But it is possible with Cheetah 3D and if you need help don't hesitate to ask.

EXL-1.jpg
 
:unsure: It really depends which production method you implement for the archtop:
1 Using an initially flat top which is deformed by braces (X-braces, A-braces, V-braces, ladder braces, Gretsch´s trestle braces, sound posts or whatever) will be fairly simple. After all, the soundboard is just a flat component prior to assembly.
2 Starting with a thick slab of wood which is then chiseled / planed (via NC machinery) is a bit more complex. The soundboard is a 3D shell. Basically a very flat dome with the odd outline of an obese hourglass.
* Which is why some Gibsons cost €/$ 20 000+ to cover the labour-intensive manual processes. Also note that 3D carving is quite complex for the NC machinery. My (limited) research implies that the basic software of some of then less expensive NC tools severely limits operations in 3D carving.

* For method 1 you already have a spline which is the outline of the soundboard. However, your wording would imply that this is not what you want to model.
* For method 2 - by the postings above - there are two alternatives:
2.1 As suggested by you and detailled by Zoohead, you can generate contours by a series of inner extrudes. You can then select these looped edges and transpose them at a right angle by useful steps.
2.2 Contours are also possible by a series of concentric splines getting smaller. Dump those splines (they need not be pseudo-concentric) into a loft creator and define relevant offsets as per your plans. You can control each of these splines quite precisely by editing primary parameters or by shifting control points / tangents. I think this is a quite efficient and economic method of generating geometry.

Screenshot 2020-09-16 at 16.20.28.png


:poop: I suggest you ignore my suggested solution in posting #20. This is laborious and time-consuming rubbish.

:sick: Sorry about the questionable shape of my guitar. I think it may be the extramarital offspring of a balalaika and a banjo who were locked in a dark attic.
 
Hi James
:unsure: If of any use, upload the primary spline depicting the outline of the body. You need to compress the source .jas. I can add a few contours plus offsets in a loft and return it for your further experimentation.
 
I’m toying with the lofting and splines technique at the moment. I’ll get back to you when I’ve refined it a little more ( as long as you don’t leave my files in a darkened attic with a balalaika). Meanwhile, to clarify the context:

The top of a quality arch top guitar is carved out of solid wood, much like a violin or cello. Gibson started using laminates heated and bent over a mould to build less expensive guitars (eg the Gibson 335). But it’s still a skilled, time-consuming process. A guy in the UK [bagpress.com] has refined a technique of gluing a lamination of veneers, wrapping them over a mould while still soft and then sticking them in a vacuum bag until the glue cures. He uses a CNC machine to build/carve the moulds and templates and takes commissions for one-off projects.
 
For something like this I'm always a bit unsure ... Of course, I would box-model the thing and end up with a nice clean render where it doesn't matter if the thing is 0,5 mm off or not (actually it would probably be more). On the one hand, you'd have to learn a lot, even if you do it with splines it's a tricky business to get a good model. On the other there is this accuracy of which I'm not so sure of (not only with Cheetah, it's with any 3d app I know).

For stuff like this I'd consider to use a CAD program; maybe even a free one is good enough. Whatever machinery is used needs anyway exactly those files and those programs should be accurate because they are made for exactly such purposes (and there are many free CAD models you could use as a starting point).

3d apps like Cheetah, 3dsmax, Maya, etc. are essentially made for something else, for example product design or anything else where you need a beautiful picture, game content, 3d printed art, animation, whatever - all those things where nobody in his right mind wouldn't even think of really using a CAD program or where you have models coming from one of them have to recreate (retopology) to get what's needed. My logic here is, if it seems reasonable to use CAD, then it probably would be the right thing. And last time I looked (many, many years ago) they where somewhat easier to handle.

Just my 2 cents.
 
* Minor research implies that the .stl format is compatible with some CNC routers. That 3D printing is primarily an additive process whilst CNC processes are subtractive should make no difference.
* After some further experimentation I am increasingly confident that a solution via lofted splines is suitable for modelling the soundboard. If it is of uniform thickness, a simple shell modifier (about 0.1"; :oops: Wow, decimal inches, what next?) will do. I find no references on soundboards of varying thickness, but that can rather easily be achieved.
* The soundboard needs to be tooled from 2 sides, anyway. If the manifolds slightly differ that is no problem.

Screenshot 2020-09-17 at 15.25.20.png
 
Last edited:
* In my own 3D projects I am mainly dealing with organic / biomorph topology so I very rarely employ box modelling as a generator. I therefore can not comment on any advantages this approach may provide.

* However, I do think that the lofted spline method allows superior freedom and flexibility in editing / fine-tuning the soundboard.
* Splines need not be flat, they can be angled / 3D curves
* Adding control points to splines is trivial
* Equally, adding a spline in-between is quite simple if you require greater accuracy of the local curvature of the archtop
* You can define the u / v sections to the granularity of 500 max
* As mentioned above, you can move and edit any of the control points of any of the creator splines, basically as required by the preexisting plans from luthier X

* By setting the raster grid to the decimal equivalent of fractions of an inch (1/8th, 1/16th, 1/32nd, assuming that imperial measurements are used), you can achieve precise adherence to the plans of a given designer, Gibson / Gretsch et al. Use the ruler to verify points and the transform panel will precisely reflect any translocations. You can set points manually by dragging the transformation widget or by entering the parameters in the panel.

Screenshot 2020-09-18 at 07.19.49.png


:unsure: I believe that Rev(?) James B is a highly experienced person in this area of plucky strings who precisely knows what devine result he wants to achieve.
 
* In my own 3D projects I am mainly dealing with organic / biomorph topology so I very rarely employ box modelling as a generator. I therefore can not comment on any advantages this approach may provide.

Actually the advantage is highly individual ... For me it's the natural way to model something which I do highly prefer over splines. And organic models are very often box modeled with subdiv, not only by me.
 
Lofting splines does seem to be the way forward, but I’ll need to step back and get a better grasp of how all this stuff works. Thanks for all the help on this so far. :)


Screenshot 2020-09-21 at 16.52.13.png
 
Hi James B
* Your C3D screen shots implies that you currently deploy lofted splines from the bottom to the top, ie you generate a single loft for the entire body of the instrument.

* You may consider splitting this into 3 individual objects:
1 BACK board, which currently is a loft with 3 splines (back, b-safe and b-peak)
2 SIDE sweeps; these may be a simple extruded spline with a shell
3 TOP sound board, which currently is a loft with 4 splines displaced vertically (front, f-safe, f-arch and f-peak).
* the splines front and f-safe (and. presumably back and b-safe) seem to be on the same plane, thence f-arch and f-peak produce the cambered archtop of the soundboard.

* In the elements Back and Front you then can increase the number of V-sections. Currently there is but a single section which results in an angled / linear curvature which needs to be planed manually (?) in post processing.
* By having separate objects for Bottom and Top you can increase the V-sections and achieve a smooth surface.

* Apex of the arch: You currently use a covered loft which generates a flat element at the very centre of the soundboard (and, presumably, the back). In my experiment I have added a pseudo-linear spline parallel to the Z-axis. By fiddling the shape of this spline into a very flat arc you can generate additional polys so that the centre part of the soundboard attains a slight curvature.

Screenshot 2020-09-22 at 12.41.57.png


* You will have noted by now that the topology generated by the loft creator reacts very finely to any modifications of any of the parameters, from UV sectioning to edits related to the controlling points / tangents / approx angles / etc of the splines in the hierarchy.
* Skill results from experiments, as :unsure: Somebody mumbled on day 6 of Genesis.

* PS: BTW, the archtop in the model is vastly exaggerated,
 

Attachments

  • Guitar 04.jas.zip
    22.9 KB · Views: 203
Last edited:
* Experiment on ƒ-holes:
* Just a simple -ve Boolean operation.

Screenshot 2020-09-22 at 16.41.46.png
 

Attachments

  • Guitar 05 ƒ-Holes.jas.zip
    15.4 KB · Views: 205
Thanks so much for all the help so far. Here’s a model of the top of a Gibson 335 style. I’ve drawn one side so that I can use a symmetry object, and extruded the side separately. So far, so good. This is so close. But there’s a funky kind of distortion on the arch at the waist.
Screenshot 2020-10-19 at 11.47.18.png

Any tips on manipulating this? Is it down to fiddling with the individual splines?
 

Attachments

  • 335 body contours v3 left.zip
    28.9 KB · Views: 196
:devilish: I encountered very much the same problem when experimenting with lofted splines at an earlier date in this thread, hence my comment above
* You will have noted by now that the topology generated by the loft creator reacts very finely to any modifications of any of the parameters, from UV sectioning to edits related to the controlling points / tangents / approx angles / etc of the splines in the hierarchy.

* The mesh being generated is highly sensitive to the topology of the subsequent contour lines of the sound board. As the splines are not similar (congruent but scaled) the resultant mesh gets distorted, locally producing some annoyingly bumpy artifacts.

:unsure: AFAIK:
1 Fine-tune the individual splines, as you have already suggested.
* Bear in mind that individual splines need not be "flat", ie controlling points need not be in the same plane. You can fiddle any individual point with the transform tool to a precision of 4 decimal digits. As you have applied steps of 0.2 in "altitude" that should allow for ample wiggle space.
* Collapse the loft into an editable mesh of polys and fine-tune by applying:
2 Brush tool, a sort of sculpting tool
3 Point mode plus smooth select (scroll with control key depressed to expand or focus)
4 deleting or shifting points or edges may help in tidying up messy areas

* Any of the above modes requires precision and much time. Maybe somebody with advanced lofting experience can add suggestions.
 
Thanks so much for all the help so far. Here’s a model of the top of a Gibson 335 style. I’ve drawn one side so that I can use a symmetry object, and extruded the side separately. So far, so good. This is so close. But there’s a funky kind of distortion on the arch at the waist.
Any tips on manipulating this? Is it down to fiddling with the individual splines?

I think your polygons are skewed too much in places.
That creates the anomaly.

Skewed.jpg
 
* It seems that tidying up 6 or so points in the "ear" of the guitar´s body already produces a major smoothing of the camber.

Screenshot 2020-10-20 at 14.14.10.png

Screenshot 2020-10-20 at 15.52.46.png


* As you will note, I have blown out the vertical scale to make the bumps (and any improvement on editing) more visible. This can be reverted once you are satisfied with the topology. Furthermore I applied a Halton Build to show the individual ribbons more clearly.
* Basically, where you have two points in proximity to define the curve in the spline I deleted one and rotated plus moved the remaining point to achieve a similar result.
* As the fiddles :)rolleyes: groan!) are instantly shown in the editor, this is less time consuming than I had expected.
* Please ignore my comment on moving points in the normal direction to the sound board. This should not be necessary.
 
That’s great. I’ve been editing and redrawing the original SVG imports without much success. This seems like a much more intuitive way to resolve it.

Only one question: how does one apply a ‘Halton Build’ sir?
 
* Build is a modifier and available in the associated menue.
Screenshot 2020-10-20 at 18.56.50.png

* Possible types are Halton and Linear. Basically, it renders invisible a percentage of the faces (tris, quads, ...) in the editor (and also the renderer).
* A useful trick to start from a blank screen and build topology out of nothing or vice versa.
Screenshot 2020-10-20 at 18.47.47.png

:) In any case, it is good to hear that you feel to be on the right way to a satisfactory solution.
 
Back
Top