CAD Forum

Can is scan to find a user that is just selling their product in posts?

RichGergely wrote: ↑Tue Feb 14, 2023 3:44 pm Can is scan to find a user that is just selling their product in posts?

Right? This forum seems to be a nice little obscure corner of the internet. It's a shame to see that shills have made their way here even.

Creality got one too:
https://www.creality.com/products/creal ... header_1.1

NeRF is the new thing coming down the 3D Scanning pipeline. It's capable of much more than what some of the more traditional methods are offering...i.e. Photogrammetry, LiDAR...etc. It is not yet something that is viable yet for any practical applications but it is amazing how fast it's evolving in terms of capturing speeds and machine learning/training AI to create volumetric displays of models.

It's at the point where it's even being used in some commericals and videos. Here's a quick example where they're seamlessly blending in-between the video footage and the NeRF model. https://www.youtube.com/watch?v=X1F4mg6EqKE.

If you want to try it out on your phone there's an app called LumaAI. I've gotten some amazing results already.

The original authors post has been removed (leaving me as the author ). The authors removed post was just sales clickbait.

In the month since I bought my Creality scanner, there are 2 new models I didn't find in my search. The Mole Paul posted above, and now this from Revopoint.

So this is building off of the already well on it's way technology called NeRF which is now expanding into LeRF!!! (Ya gotta luv us Nerds, we know how to make acronyms!!!!) https://www.lerf.io/ . The best part is that this can all be done with the Phones.

Here's an example of some models...

Uboat Ancho https://captures.lumalabs.ai/phenomenal ... -cp-221580

Big Connect Four Game https://captures.lumalabs.ai/firmer-strong-8-55876

Diving Helmet https://captures.lumalabs.ai/classy-reassure-af-148863

zxys001 wrote: ↑Sun Mar 26, 2023 3:08 pm

One thing that I'd caution people about, when it comes to buying these sub $1,000 scanners, is that working with mesh data...aligning, erasing extra noise, basically cleaning up the files, don't all come with the best software. And they're all generally missing the biggest aspect which is the mesh to CAD/BREP/NURBS conversion. Though most CAD softwares have gotten much better with dealing with mesh data they're still a VERY far cry from being able to handle mesh data well. The tools necessary to work with mesh data can be very different than the normal modeling tools found in base CAD softwares.

I am working through that process now. I'm not sure if I should spend time with Meshmixer or meshlab, or just commit to learning blender, or investing in rhino. The little bit of time I've spent with the first two has produced less than desirable results, but there's lots of room for improvement in my scans.

SPerman wrote: ↑Wed Mar 29, 2023 9:37 pm I am working through that process now. I'm not sure if I should spend time with Meshmixer or meshlab, or just commit to learning blender, or investing in rhino. The little bit of time I've spent with the first two has produced less than desirable results, but there's lots of room for improvement in my scans.

What scanner are you using? The first question I would ask is what is your intent in terms of what you want to do with the data once scanned? Do you at least have a watertight model or are you still trying to align the different meshes into one model?

Based upon that I can at least make some recommendations as to where the next best steps can be.

Let me start by saying my whole project is for the sole purpose of understanding the technology, so that when I do need it, it doesn't turn into a 6 month project. I am attempting to reverse engineer a complicated mechanical device with lots of small parts. I realize this is a challenge for any of the current scanner technology that I can afford. I purchased a creality scan lizard. (I've also done some experimenting with Polycam.)

Even when I had access to a 6 figure Romer arm with a laser scanner, the scans were only for reference when 3d modeling parts. That is my intent with this project as well. I don't think it is feasible for the size of the parts I am scanning and the quality of the scans I can produce, to expect to generate a watertight solid, but I would be happy to be proven wrong.

Arthur NY wrote: ↑Wed Mar 29, 2023 7:46 pm One thing that I'd caution people about, when it comes to buying these sub $1,000 scanners, is that working with mesh data...aligning, erasing extra noise, basically cleaning up the files, don't all come with the best software. And they're all generally missing the biggest aspect which is the mesh to CAD/BREP/NURBS conversion. Though most CAD softwares have gotten much better with dealing with mesh data they're still a VERY far cry from being able to handle mesh data well. The tools necessary to work with mesh data can be very different than the normal modeling tools found in base CAD softwares.

Hi Arthur, absolultely, even with expesive systems, if you get into this you'll quickly learn, depending on what you need/want,.. or the software you may choose (stock or other tools) there is more work , methods, techniques and workarounds for the subjects/lighting/details you're trying to capture.

SPerman wrote: ↑Thu Mar 30, 2023 7:33 am Let me start by saying my whole project is for the sole purpose of understanding the technology, so that when I do need it, it doesn't turn into a 6 month project. I am attempting to reverse engineer a complicated mechanical device with lots of small parts. I realize this is a challenge for any of the current scanner technology that I can afford. I purchased a creality scan lizard. (I've also done some experimenting with Polycam.)

Even when I had access to a 6 figure Romer arm with a laser scanner, the scans were only for reference when 3d modeling parts. That is my intent with this project as well. I don't think it is feasible for the size of the parts I am scanning and the quality of the scans I can produce, to expect to generate a watertight solid, but I would be happy to be proven wrong.

So this is a very challenging area that you're in. It would be like trying to learn Sketch Up to see if Solidworks is a good fit for you....

There is still a VERY huge gap between the price points that are affordable in terms of the 3D Scanners, the software they come with, and going full Reverse Engineering. But hey we all have to start somewhere. Depending on the work we're doing there are times when the software that comes with the higher end scanners is enough but I can also say that none of them come close to the dedicated RE packages like Geomagic, Polyworks, and Design X. These packages are just in a league of their own and for good reason. They have the necessary tools to properly take scanned/mesh data and make it something useable. To your point no 3D Scanned model is ever really the end, it's more often than not the base line reference to be used to whatever downstream application in required at the time.

The workflows going from scan to CAD is completely different than going to Architecture or VFX/Gaming. And within each of those there are subsets. My brain just hurts thinking about all of the combinations that I've hobbled together over the decades. Most of the CAD software still lack any real tools to properly deal with mesh data. Be it decimating, segmenting, converting Tri-mesh to Quad-mesh....etc.

My goal is to understand just what is possible with the tools available in my price range, and if the optimized results are worth the effort required to create them.

I did this test a few years ago, and the answer was clearly "not good enough." I won't be surprised if the answer hasn't changed.

SPerman wrote: ↑Sat Apr 01, 2023 8:49 am My goal is to understand just what is possible with the tools available in my price range, and if the optimized results are worth the effort required to create them.

I did this test a few years ago, and the answer was clearly "not good enough." I won't be surprised if the answer hasn't changed.

It really does become specifically what it is you need or the subjects/size/details/shape/....you'll be scanning. The tech is getting better, but....
Imho, I can assure you this, the $35k-$70k non contact scanning systems we tested could not get what we needed within 0.05mm consistently...but, they said, "maybe" their $100k system could work (it did not work either)... and this did not include special fixturing and a special line setup.
CT scanning was the only consistent consideration.
Being realistic about these tools regardless the hardware/software. It takes time, setup, finesse and expect issues.

SPerman wrote: ↑Sat Apr 01, 2023 8:49 am My goal is to understand just what is possible with the tools available in my price range, and if the optimized results are worth the effort required to create them.

I did this test a few years ago, and the answer was clearly "not good enough." I won't be surprised if the answer hasn't changed.

I think that much of this is about adjusting your expectations.....For example it still takes 3 hours to bake a turkey in the oven, sure you can microwave it but it isn't going to taste good....

Or think of 3D Printing, sure it has "sped up, but until it is a Star Trek Replicator it will never be fast enough!!!

That said...3D Scanning will never "Be" the part/model that's being scanned. There are a whole host of reasons for this but to just help keep it simple for now, whether it is a $100 or $100K scanner there will need to be post work done. The question is how long will it take to get to the desired results. We do amazing scanning projects for a whole host of firms ranging from Architecture to aerospace, movies to art work and none of the projects are ever just push button easy. With this in mind the tolerances you're asking for I'd actually ask why? Not that they don't need to be the question is always what needs to be done with the data and where.

It's definitely challenging to look for a process to achieve something when the "budget" is held to within a certain range. What may be more practical is to come up with a set criteria and have a firm help to prove through several types of scanning systems how to get to the results you're looking for.

Read the spec carefully.
0.05mm probably for something 10cm x 10cm.
The real resolution is % of the model size scanned.

Bought a old Sense 1st Gen (3DSystems) for $200.
Tracking is a problem with it. Probably better if the part is on a patterned grid.
I wanted to try putting markers on the part.
When it can get the scan, it's good.
The software can do simple editing. Remove bits and pieces. Patch holes. Fill and make it solid.

I'll say software is more important then hardware.
If you can get a solid after the scan, you're days ahead already.

I tried scanning manufactured parts and compare to model.
Put both in assembly and try to mate points on the model.
It's not for inspection but you can see how close it is.

Overall I got more failed scan attempts than usable one.

The company I worked for did have a laser scanner for inspection.
Don't know if it can create 3D model.

https://3dprintingindustry.com/news/rev ... ng-225376/

Looks more like a camera.

Is it possible to make one with Jetson?
Use ML to help get a better scan and create 3D model instead of mesh.