Jump to content

Affordable 3D scanner?


Michael Richwine
 Share

Recommended Posts

A few months ago, I took 15 fiddles down to a music town to be evaluated by three of my favorite clients, full-time performers who are friends and generous with their time. These ranged from inexpensive Chinese to pretty expensive German Kunstgeigenbauer instruments, all selected because I thought they might be interesting to these players, and I wanted some feedback to help me serve my market better. To my surprise, all these players ranked the instruments nearly identically in order of preference. Price seemed to have little relationship to how they were ranked.  I got to looking at them carefully, and the most preferred instruments all seem to have very similar arching. 

I'd like to run some experiments, duplicating the arching on the most successful examples that were tested, then assembling those plates to existing ribsets that I have available, and having a listen at how they compare. I have access to inexpensive instruments damaged in transit, from which I can scrounge virtually identical ribsets, plus some decent tonewood sets for plates which should work. I also have access to an industrial CNC router that is already set up for violin parts. The only problem is that the router was always been programmed manually to a specific original design.  I need to come up with a scan of the shapes I want to duplicate and test.  I've contacted some scanning services, but am not getting anywhere very fast, yet.

I've seen some inexpensive 3D scanners, but am not impressed by anything I have seen yet. If I had my own scanner, it might pay off on this project, and also on some other future projects. All production oriented. 

Does anyone have a recommendation for an affordable scanner capable of 3D scanning violin plates and producing a point cloud in an IGS file format? 

 

Link to comment
Share on other sites

51 minutes ago, Advocatus Diaboli said:

Haven’t seen results from most of the cheap scanners around.  It might be cheaper/easier to find someone with a scanner who would let you use it. 

I've initiated contact with a couple of local firms, but no response yet. Maybe too small a job to be interesting.  Owning my own scanner, I have better control over time and other factors. I will continue to explore all avenues until I come up with an acceptable solution. Probably should have done this a few years ago.

Link to comment
Share on other sites

I saw a good scanner, probably around 5K, at CES 2 years ago.  I will look for my old notebook for the name.  They already had scanned some guitars, and several had spoken to that company re violins.  Other scanners were too small, too slow, or didn't have good resolution (sub millimeter). Others were way more expensive.  Here's an interesting site - there's also a link for pro scanners, and most there seem to be between 2 -8K.  https://all3dp.com/1/best-3d-scanner-diy-handheld-app-software/  Or try a cheap one, like Revopoint.  Lots of hacks for the cheaper ones...

Link to comment
Share on other sites

8 hours ago, l33tplaya said:

I saw a good scanner, probably around 5K, at CES 2 years ago.  I will look for my old notebook for the name.  They already had scanned some guitars, and several had spoken to that company re violins.  Other scanners were too small, too slow, or didn't have good resolution (sub millimeter). Others were way more expensive.  Here's an interesting site - there's also a link for pro scanners, and most there seem to be between 2 -8K.  https://all3dp.com/1/best-3d-scanner-diy-handheld-app-software/  Or try a cheap one, like Revopoint.  Lots of hacks for the cheaper ones...

Thanks, that's a step in the right direction! I've seen the results from some of the rinky-dink ones, so there's definitely a minimum acceptable standard.

Link to comment
Share on other sites

Definitely on the right track now. Seems I can get what I need for around us$750. Several competing brands and models at that price point that will handle violins and mandolins. Need to learn more and read more reviews and specs, but resolution of .01mm seems plenty. and volume of 500 x 1000mm also seems adequate, even for possibly more ambitious projects.

Link to comment
Share on other sites

I do a fair amount of 3D scanning for restoration work.  I think right now if I wanted to do your application fairly inexpensively, I’d look into photogrammetry.  Micron precision isn’t necessary, and with a little bit of sweat equity in stitching photos together you could probably be up and running.  There are numerous tutorials on YouTube on how to do this with largely open source software.

 

With that being said, I’ve seen impressive results of plates done with an Einscan Pro.  These can be found on eBay for +/- 5k.  The Transcan C from the same company also looks quite good, and what I would be buying for my application if I needed to buy something new tomorrow.   If you are feeling adventurous, you can build your own scanner using off the shelf components and flexscan software from polyga… which is probably what I’ll be doing for my next scanner.  

The new crop of budget scanners from creality and revopoint look very interesting, but I haven’t met anyone using them on instruments yet who are able to give feedback.

Link to comment
Share on other sites

I think that the Creality scanners can do the job I need, but I don't yet know whether they can give me a  plain IGS file that I can use with a CND router, so I have more to learn. 

I'll look in to photogrammetry as well.  I have a good camera and graphics card. I'm a little shy of Autodesk products, though. They're not exactly aimed at us one-horse operations.

Link to comment
Share on other sites

Michael, do you think that 15 violins and 4 people evaluating them is really a valid test? It is interesting result, but that is about all.

Before you spend money on a scanner, can you find out if there are scans of violins with similar arching already available that you could use?

Link to comment
Share on other sites

1 hour ago, GeorgeH said:

Michael, do you think that 15 violins and 4 people evaluating them is really a valid test? It is interesting result, but that is about all.

Before you spend money on a scanner, can you find out if there are scans of violins with similar arching already available that you could use?

It's pretty rare to get three players to agree so thoroughly, especially with such close competition, because I had selected each of those instruments with them specifically in mind. The fact that they ranked them in the same order was certainly worth exploring, so I'm interested in investigating those specific instruments. Education is expensive. As it turns out, Jerry Lynn's suggestion of looking at photogrammetry is pretty interesting. Looks like I may be able to achieve my goals with only a camera, my existing photo setup, and some open source software. There's time involved in learning the software and coming up with files that can be used on the CNC, but today it got a whole lot easier. How do you learn stuff without trying it out?  If you have a hypothesis, test it!

I watched my former boss test different ground formulations on identical (as near as possible) violins for six years, comparing one formula against the other. I would never have believed a ground makes so much difference if I hadn't seen the development process myself. So I'm willing to invest the time and money it takes to test out my hypothesis and see whether it works or not.  I can keep the wood the same, cut from the same logs, aged the same, and shaped with considerable precision. To do it with CNC makes it possible to investigate different shapes with control, using a minimum investment of time and materials. I'm sure other makers have done this sort of thing, investigating other variables, but I haven't read about it.

Link to comment
Share on other sites

13 hours ago, Jerry Lynn said:

I do a fair amount of 3D scanning for restoration work.  I think right now if I wanted to do your application fairly inexpensively, I’d look into photogrammetry.  Micron precision isn’t necessary, and with a little bit of sweat equity in stitching photos together you could probably be up and running.  There are numerous tutorials on YouTube on how to do this with largely open source software.

That looks really interesting!  I'd have to get a different GPU for the open source software and CUDA, but the current demand for gamer cards seems to have driven the cost of used GPUs way down, even for 4 gb plus .

Link to comment
Share on other sites

7 hours ago, Michael Richwine said:

So I'm willing to invest the time and money it takes to test out my hypothesis and see whether it works or not.  I can keep the wood the same, cut from the same logs, aged the same, and shaped with considerable precision.

That does sound like it will be an interesting experiment. I hope you share the results. 

Link to comment
Share on other sites

I'll get a 6gb graphics card (GTX1060) that's compatible with the photogrammetry software for around US $100, and should be able to do some preliminary attempts in a few days or maybe a week, depending on how my brain and other workload/ customers treat me. There are some good tutorials on YouTube.  I'll post some short updates as things go along.

Link to comment
Share on other sites

3 hours ago, Michael Richwine said:

I'll get a 6gb graphics card (GTX1060) that's compatible with the photogrammetry software for around US $100, and should be able to do some preliminary attempts in a few days or maybe a week, depending on how my brain and other workload/ customers treat me. There are some good tutorials on YouTube.  I'll post some short updates as things go along.

There is free open source photogrammetry software around.  I have metashape pro which is great, but not really worth investing in u less you know you’re going to be spending a lot of time with photogrammetry. 

Link to comment
Share on other sites

What about just knocking door on a nearby hospital and asking for a CT scan? Some evenings the emergency doesn't have anything to do :-)

I'm not kidding here. My friend did just that with a Lloyd Loar mandolin and after few seconds of surprised looks from doctors they took the "patient" in just to kill the boredom. Now we have full CD of examination results of patient named Lloyd Loar born in june 1923...

Getting the 3D scans done is just beginning of a looooong route. What are you planning to do with that? CNC work?

Link to comment
Share on other sites

8 hours ago, Advocatus Diaboli said:

There is free open source photogrammetry software around.  I have metashape pro which is great, but not really worth investing in u less you know you’re going to be spending a lot of time with photogrammetry. 

I've watched a tutorial on Metashape and Blender. It seems I need a GPU compatible with CUDA, which I am in the process of acquiring. This seems relatively straightforward and does produce a point cloud that I can use to produce CNC programs necessary to machine duplicate plates, as best I can tell. There are other projects where the capability to scan shapes would be very useful.

Link to comment
Share on other sites

9 minutes ago, Michael Richwine said:

...This seems relatively straightforward and does produce a point cloud that I can use to produce CNC programs necessary to machine duplicate plates...

Before purchasing a CUDA capable graphics card specifically for this, it might be worth researching what is going to be involved in going from point-cloud to (presumably) G-code (the instructions needed for most home CNC routers).

While I obviously don't know what hardware / software you will be working with, I don't think it is guaranteed to be simple moving from the typical unstructured output of a photogrammetry package to the geometric primitives needed for good G-code.

Link to comment
Share on other sites

2 hours ago, notsodeepblue said:

Before purchasing a CUDA capable graphics card specifically for this, it might be worth researching what is going to be involved in going from point-cloud to (presumably) G-code (the instructions needed for most home CNC routers).

While I obviously don't know what hardware / software you will be working with, I don't think it is guaranteed to be simple moving from the typical unstructured output of a photogrammetry package to the geometric primitives needed for good G-code.

Working with a large industrial production CNC router. Not sure of its exact specs, but it has at least a 4' x 8' bed and has a 4th axis setup. The guy I'm working with, who owns the machines, says all he needs is a point cloud  file. I'll work with him from there, and learn as I go. I started with machine code in the 1960s, and then Fortran and Cobol on IBM 360s and many generations up to now, but there are plenty of gaps in my experience that need to be filled in. I need a CUDA compatible card to use Metashape, so it's a done deal as far as I can tell, always with gratitude for the caveats.

Link to comment
Share on other sites

4 hours ago, notsodeepblue said:

While I obviously don't know what hardware / software you will be working with, I don't think it is guaranteed to be simple moving from the typical unstructured output of a photogrammetry package to the geometric primitives needed for good G-code.

Pretty much all major (and most minor) CAM programs these days have zero problems dealing with any organic shape you can throw at them. Some, like DeskProto, you can use with next to experience and still get good results. 

Link to comment
Share on other sites

Autodesk Fusion is a good one to start with. It is free to use as an enthusiast - however to print anything you'll need to cough up for a subscription. The enthusiasts licences is a good way to go to figure things out and then pay for a single month if you need to do some printing

I'm lucky - I sat across from the team who implemented the g-code sectioning and UI into Fusion when I worked at ADSK.

Link to comment
Share on other sites

9 hours ago, Urban Luthier said:

Autodesk Fusion is a good one to start with. It is free to use as an enthusiast - however to print anything you'll need to cough up for a subscription. The enthusiasts licences is a good way to go to figure things out and then pay for a single month if you need to do some printing

I'm lucky - I sat across from the team who implemented the g-code sectioning and UI into Fusion when I worked at ADSK.

By printing you mean to export a file for a 3d printer, that’s included in the personal license.  Right click on the body you want to print and hit save as mesh, then import to your slicer of choice.  The big differences between the personal and the paid subscription is the personal is capped at 10 active documents (you can archive as many as you want), and for CNC cutting tools  ATCs aren’t supported. Rapids are limited to cutting speed.    While I love fusion, and I do a lot with it, I usually recommend a different CAM package for beginners. 

Link to comment
Share on other sites

FWIW, all I'm interested in right now is getting a scanned point cloud that my colleague can use in a professional cad/cam  application. I don't have any current interest in 3D printing. So far, I've spent about $120 on a used 6gb graphics card that is compatible with the free software that I will download as soon as the card comes in. Since I do a fair amount of photo and video editing, the GPU upgrade will be a nice side benefit over the 1gb unit I have currently.

Link to comment
Share on other sites

17 hours ago, Jerry Lynn said:

By printing you mean to export a file for a 3d printer, that’s included in the personal license.  Right click on the body you want to print and hit save as mesh, then import to your slicer of choice.  The big differences between the personal and the paid subscription is the personal is capped at 10 active documents (you can archive as many as you want), and for CNC cutting tools  ATCs aren’t supported. Rapids are limited to cutting speed.    While I love fusion, and I do a lot with it, I usually recommend a different CAM package for beginners. 

I should have been more clear - yes 3d printing is limited (you use to be able to print directly to a 3d printer hooked up to you computer)

But also note if you plan to use fusion as a drawing tool (the main use for me at least) one can't export 2D drawings as PDFs with the enthusiast licence. You can print a single sheet apparently but I haven't tried it. I usually export a dxf from fusion and bring it into ACAD to create PDFs

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share

  • Recently Browsing   0 members

    • No registered users viewing this page.


×
×
  • Create New...