• Spring 2024 meetup in Calgary - tentative date Saturday, April 20/2024. Other regions are also discussing meet ups. If you want one in your area get going on organizing it! discussion
  • We are having email/registration problems again. Diagnosis is underway. New users sorry if you are having trouble getting registered. We are exploring different options to get registered. Contact the forum via another member or on facebook if you're stuck. Update -> we think it is fixed. Let us know if not.
  • Spring meet up in Ontario, Newmarket, April 6/2024. Discussion

Reverse Engineering from a 3D Scan with Fusion360

PeterT

Ultra Member
Premium Member
This might interest some of you Fusion-360 users, although the thrust of the video is 3D CAD modeler agnostic once you have the magic wand & freebie software mentioned. With the mesh scan imported & aligned to appropriate planes as he shows, the rest of process is analogous to building a 3D model from scratch. I foresee projects like this in my future one day but have not delved into the hardware scanner. The price range is crazy wide & I'm not quite sure what fits the purpose. Anyone else messed around with this?

 
During Covid (or during a break last fall) I was at the machinery show in Mississauga and saw the free hand 3d scanning laser systems in use, amazing....to say the least.

I also saw an automated 3d measuring probe set up. The accuracy OMG (well beyond my abilities to machine at this point). $25K allows you to play.
 

Dan Dubeau

Ultra Member
Interesting video. I'll probably have to watch it again to extract all the info.

I bought a revopoint Pop2 during their kickstarter last winter, but aside from playing around with the demos, and a few other quick attempt at scanning the kids heads and a few other household objects I haven't used it since. He makes it all look easy. I hope to get to that point someday.
 

PeterT

Ultra Member
Premium Member
Keep me posted as you start messing around with yours. My intent is much like the video - basically importing a 3D scan from an object, but with the specific purpose of developing a ground up 3D cad model, more or less using the scan as a background reference layer to turn on/off visually as the modelling progresses. So some level off scan resolution/accuracy is required, but not to the Nth degree like some other applications. I know a lot of folks, what I'll affectionately call 'creature modelers' import the scan & they want to capture as much of the detail as possible & physically manipulate the mesh before 3D printing or whatever. That's not me. If you find any good links on the hardware or usage, feel free to post. I'm in no particular hurry but curious.
 

Tom O

Ultra Member
I’ve got the Makerbot scanner that I’ve only used a couple times there are plans out there for using the video head sensors (most pawn shops 5/10 bucks) and lidar on my phone I haven’t played with that either. Lol
 

Mcgyver

Ultra Member
I know a lot of folks, what I'll affectionately call 'creature modelers' import the scan & they want to capture as much of the detail as possible & physically manipulate the mesh before 3D printing or whatever. That's not me.

You and I, we're 1%'s. The one percent of 3D scanning and printing folks with no interest in Orcs.

I recently bought a Revopoint mini, which is their newest, highest resolution offering, .02mm resolution. Its a blue laser which is better for small detail (apparently). Its less well suited for larger objects, faces, cars etc - Dan's is better for that, but the mini seems just the thing for fine detail. While its not exactly cheap, its 5 or 10% of what the high detail industrial models are....consumer priced 3D scanning seems to me is an emerging area that can't help but be of intreset to us shop types.

Their test bust scans well, but its been really difficult to get scans of other things. I've had to buy 3D scanning spray, and its still very challenging. There is a huge will to help from the vendor, but an also a language barrier. Software is ok, but its not slick. Lots left unsaid in documents and manuals. For me, it wasn't plug and play with great results, its a struggle and I'm still in that phase. Between this and my casting adventures, many days I push it aside and head to garage to machine something knowing its a 95% chance of success vs, well, almost single digits. (but I shall perserver....another bronze coming up in a few hours!)

Great video btw, really good to know how go from mesh to CAD
 

PeterT

Ultra Member
Premium Member
Interesting comments @Mcgyver please keep me informed as you progress.
My Spidey sense was its probably not that slam dunk a process as its sometimes made out to be. I suspect the devil is in the details - hardware, software, technique...

I keep stumbling on these camera (or in this case phone) methods that look surprisingly good. Now obviously the software is driving the 3D image stitching. At one point it shows a missing or unresolved sliver. But dang, if my purpose was purely as a background mesh image to reference for a ground-up CAD process, that looks pretty good. I have seen some examples of objects where it failed miserably & they had to get into sprays & reference dots & other higher end techniques. If I recall it was something like a motorcycle & it kind of blew its marbles on head fins & other tricky, dimensionally finer features. They came out smeared & distorted. But that was a long time ago by development pace standards.

 

Matt-Aburg

Ultra Member
Interesting comments @Mcgyver please keep me informed as you progress.
My Spidey sense was its probably not that slam dunk a process as its sometimes made out to be. I suspect the devil is in the details - hardware, software, technique...

I keep stumbling on these camera (or in this case phone) methods that look surprisingly good. Now obviously the software is driving the 3D image stitching. At one point it shows a missing or unresolved sliver. But dang, if my purpose was purely as a background mesh image to reference for a ground-up CAD process, that looks pretty good. I have seen some examples of objects where it failed miserably & they had to get into sprays & reference dots & other higher end techniques. If I recall it was something like a motorcycle & it kind of blew its marbles on head fins & other tricky, dimensionally finer features. They came out smeared & distorted. But that was a long time ago by development pace standards.

Did you ever buy hardware for this type of work. I think the photometry is only for artistic stuff. This would not be accurate enough for manufacturing. As far as the cheap scanners go. I think you will get so frustrated that you will fold. on a crealty. Einscan has been out for several years and last year they came out with a cheaper scanner. Their software is proven too. I jist price checked it.. at 1300 still pretty serious investment.

On the CAD part.... I have not every used fusion. I am a Unigraphics addict. I also use Geomagic Wrap for cleaning scan data before it goes into CAD. I usually draw simple mechanical stuff. but also have some freeform surfaces right out of wrap. I will combine them to make a "hybrid" model. That is mostly simple, with some wrapped surfaces as needed. A good example of a real world project was a mold I designed for siding. The original mold had been cast from real rocks. The old mold was like 20 plus years old. No prints, no data.... only scan data at low res. We had to send it out to get the part scanned, then develop each of the 42 rocks as an individual part. This was because of file-size.... very detailed and time consuming job... Took over 6 months for that mold. They had to upcharge the customer over 500 K because they had NO idea what they had misquoted...
 

Attachments

  • hd1.JPG
    hd1.JPG
    272.9 KB · Views: 4

PeterT

Ultra Member
Premium Member
No, I never went further than casual interest. Yes, vintage objects would be a good example where this type of technology would really be beneficial. Not as a magic wand to make a copy, but a good approximation to constantly compare a re-creation against. I've spent many hours developing things in CAD from 2D images & its not easy, 3D would be even more challenging.
 

bigHUN

Member
Even for $20K don't expect much from a 3d scanner. It is not meant for revers engineering no matter what they say, you still need to do the CAD manually.
3d scanners are mainly for duplicating with 3d printers, for figurino's :)
I've been exposed first time to scanners back 20+ years ago, many things have changed but the main thing is the point cloud is limited.
If you really wanna spend some money go for CMM style, at least you got some paramaters=numbers to work with.
Btw, to clarify, I started with Unigraphics back in '98 and worked with Catia and ProE as well, really a NX addict.
 

Matt-Aburg

Ultra Member
Even for $20K don't expect much from a 3d scanner. It is not meant for revers engineering no matter what they say, you still need to do the CAD manually.
3d scanners are mainly for duplicating with 3d printers, for figurino's :)
I've been exposed first time to scanners back 20+ years ago, many things have changed but the main thing is the point cloud is limited.
If you really wanna spend some money go for CMM style, at least you got some paramaters=numbers to work with.
Btw, to clarify, I started with Unigraphics back in '98 and worked with Catia and ProE as well, really a NX addict.
I just had a customer ask if I can QC three parts / cavity from each mold. Their main customer is now requiring a report back on each new part. I asked what kind of volume do they expect / year. Currently they are doing 35 - 40 molds / year with this customer. Their customer wants to place much more work with them. Some molds have 8 cavities, so 24 parts to scan, inspect and report on that type of tool.

I responded back that I will need to level up to VXInspect, an addon for VXelements. This is a 14000 investment now, plus I will have to pay my maintenance to do the upgrade (which I did not and it has lapsed). Bottom line is it will cost a bundle to do this kind of work. Good thing I have a large plate with 1350 holes for fixturing...

Now the ball is back in their court.


 

Susquatch

Ultra Member
Administrator
Moderator
Premium Member
Even for $20K don't expect much from a 3d scanner. It is not meant for revers engineering no matter what they say....

I agree completely. In fact, I don't even think they know what they are talking about. You cannot reverse engineer anything by scanning it or measuring it. The best you could ever do is duplicate it and even that has its weaknesses.

Btw, as far as I know, you are the first member on here that has mentioned any history with Catia besides myself. Mine goes back over 15 years ago, but it was awesome software.
 

Matt-Aburg

Ultra Member
I agree completely. In fact, I don't even think they know what they are talking about. You cannot reverse engineer anything by scanning it or measuring it. The best you could ever do is duplicate it and even that has its weaknesses.

Btw, as far as I know, you are the first member on here that has mentioned any history with Catia besides myself. Mine goes back over 15 years ago, but it was awesome software.
I think there is no way at this point in time to magically click a button and turn mesh data into a solid object, Automatic surfacing creates a bunch of shit surfaces that cannot be extended, are very dense on the amount of points... just to mention a few shortcomings. "Wrapping" a point mesh is an approximation. If a few points are way out... the surface will deviate for the intended design. Mesh data can only be used as a guide to redraw
the solid object on a different layer.

I never went the craptia route... (sorry @Susquatch ) LOL... Not many toolmakers use the product down here... Those that did have switched over to NX. Same goes for stimitron.
 

Susquatch

Ultra Member
Administrator
Moderator
Premium Member
I never went the craptia route... (sorry @Susquatch ) LOL... Not many toolmakers use the product down here...

Not really surprised. Catia is a product design tool not a tool makers tool. Besides that, it's extremely expensive. 20 years ago it was way beyond what most small shops could afford unless they ran the seat 24/7.
 

Matt-Aburg

Ultra Member
Not really surprised. Catia is a product design tool not a tool makers tool. Besides that, it's extremely expensive. 20 years ago it was way beyond what most small shops could afford unless they ran the seat 24/7.
NX is pretty expensive, 50 K for Mach3 mold design license. My previous employer has 3 groups in India, so the licenses get used 24/7 .. 364.5 days / year...
 

Susquatch

Ultra Member
Administrator
Moderator
Premium Member
NX is pretty expensive, 50 K for Mach3 mold design license. My previous employer has 3 groups in India, so the licenses get used 24/7 .. 364.5 days / year...

Ya, that's expensive too. At least your previous employer understood the need to run the stuff 24/7 to pay for it. I've seen far too many who didn't and burned.

That said, 50K would not even buy the workstation needed to run Catia let alone the software.

I do miss Catia myself, but I don't miss working in the auto industry. Life is soooo much more enjoyable being a farmer and fixing farm equipment.
 

Matt-Aburg

Ultra Member
Ya, that's expensive too. At least your previous employer understood the need to run the stuff 24/7 to pay for it. I've seen far too many who didn't and burned.

That said, 50K would not even buy the workstation needed to run Catia let alone the software.

I do miss Catia myself, but I don't miss working in the auto industry. Life is soooo much more enjoyable being a farmer and fixing farm equipment.
I was looking into Catia since you saying the work station costs so much. Catia is now PC (windows based). They don't support SGI anymore. The workstation would cost no more that any other decently powerful machine. However, It is costly software. What version were you running?
 

Susquatch

Ultra Member
Administrator
Moderator
Premium Member
I was looking into Catia since you saying the work station costs so much. Catia is now PC (windows based). They don't support SGI anymore. The workstation would cost no more that any other decently powerful machine. However, It is costly software. What version were you running?

Even as I wrote that reply my mind considered that possibility. A lot can change in the 15 years since I retired. PC's continue to get more powerful each year probably at a rate much faster than the software requires.

Also, the environment was undergoing rapid changes back then with design work being done on workstations and system simulations being done on hyper fast mainframe supercomputers. I could see that evolution easily morphing into what you describe.

But I've been retired for half of a normal career so even the premise has probably seen a paradigm shift. I'd love to go back just to see what has all changed.

I have no slightest idea what version it was. Dassault used to come in regularly and update it for us and also provided update seminars. I probably knew the version numbers back then but then again maybe not - I've always had a poor memory especially for numbers and names.
 
Top