3d scanning to shapeways files

[Jamesn320]

And, when you scan you get a solid object and 3D printing gets expensive when its a large amount of material.  Cad models normally are hollowed out.

Juts my 2 cents! James

[wcfn100]

Another way to look at this is to think about 2D scanning.  I would imagine most have learned that to get the best reproduction of and old photograph is to scan it in at a very high resolution.  So to print something at 300dpi you would try and scan at at least 1200dpi or 4x the output.  The same idea applies to 3D scanning accept even more extreme, maybe 10x would be needed.

So if your printing FXD at Shapeways at .1mm, you'd probably want a scanner that could go down to .01mm.  And even of you did that, and you got a very accurate model, the file may be so massive it could be unusable.

Just learn to draw.  With 3D printing going the way it is, you'll use the skill for something beyond trains.


edit:  I just wanted to add that anything I say is absolutely open to correction or updating.  I started in 3D modeling before there were even smooth curves (vector based), everything was point/line based mesh.  Shortly after I started, NURBS became a thing (look it up if you want  ).  If they've overcome any of the limitations scanning has by way of hardware of software, I'd certainly be interested in hearing about it.

Jason

[C855B]

Another way to look at this is to think about 2D scanning.  I would imagine most have learned that to get the best reproduction of and old photograph is to scan it in at a very high resolution.  So to print something at 300dpi you would try and scan at at least 1200dpi or 4x the output.  The same idea applies to 3D scanning accept even more extreme, maybe 10x would be needed. ...

I was trying to gather the thoughts to say that earlier but at 2:00 a.m. the brain wasn't fully engaged.

Once more, we're repeating history from 2D. Scanning at a high enough quality to use for unedited output lagged behind printing by at least 10 years, probably more. I remember looking at $250-300K drum scanners in the late '80s to do press-quality halftones and color seps on our high-end PostScript film imagers. I couldn't justify the expense with the "knowledge" that affordable quality scanning from a desktop flatbed was "just around the corner"... I was out of that business by the time the tech finally caught up.

3D printing for non-organic shapes is going to be a vector world for a while. Probably a long while.

[JMaurer1]

Adam: Seemed like an easy enough question. Short, non geeked out answer is learn the software.

[MK]

But you guys are talking about different things.

Organic shapes like fruit or a tooth can be much better represented by a scan and the resulting mesh will be much easier to adjust to match the original object better.

To get down to maybe the simplest issue, right angles. 



Obviously the drawing is very simplistic, but it's just to convey the idea. A scanner can only map what it sees.  so in the drawing, if it can't see the two points in the red circles, it will take the next point it can see (the blue dot) and create the green line instead of the right angle line that should be there.


A tooth or an apple won't be as affected by this slight angle change or cabe modified much easier if it's a dimensional issue, but multiply this 100 time across one of our models and you have a pile of mush.


Jason

Teeth are actually quite hard to scan accurately with highly irregular surfaces, deep cuts and the death kneel, undercuts.  But modern day scanners can hangle that easily.  Your example above is assuming that the model is fixed and the laser is not highly articulated.  Modern scanners articulate the object and the laser to give quite a bit of accurate coverage.

Check out this 2013 YouTube video of Nobel's 2G scanner.  The video doesn't show it's max articulation that it can do.  It'll surprise you.  That's 4 years ago so technology has advanced since.  Scanner is roughly $50k.

[wcfn100]

Teeth are actually quite hard to scan accurately with highly irregular surfaces, deep cuts and the death kneel, undercuts.  But modern day scanners can hangle that easily.  Your example above is assuming that the model is fixed and the laser is not highly articulated.  Modern scanners articulate the object and the laser to give quite a bit of accurate coverage.

Still comparing apples to oranges though.

The problem with articulated scanning is now the information needs to be interpolated by the software.  You're going to get different reference points for the same spot which won't exactly line up and the software needs to decide what to do.  And going back to file size, you can scan a tooth down to microns and still get a file you can work with.  Something even the size of an N scale box car would create a massive file with all those reference points.  If you ever want to see what a difference it makes, you should be able to adjust the steps of iterations/steps the programs outputs curves with and see the jump in file size.

When I drew the box car end for the last RPM mini kit, I had to up the iterations of the curves because it was actually printing straight lines on the O scale version.  The file size went up over three times 3x just from adding some extra steps in the curves.

Yet another way to look at scanning is it won't be able to generate truly flat surface.  So a large flat box car side is going to get scanned in wavy and someone is going to have to go back and fix it which will take more time than just drawing it.

I'm not trying to say scanning isn't useful, but it's just a tool and it will not create a useful final model for our purposes.  Work will need to be done on that model and there may be much more cleanup than just doing it correct with a drawing.

Jason

[wcfn100]

I'll look more later, but here's a actual screen shot someone did of a Kleenex box.



I don't know the scanner or anything, and maybe this was the worst image I could possibly find, but just so there's a visual to what I'm talking about.

Not quite ready to put wheels on and roll down the track.

In fairness, here a very high res scan using a "Breuckmann StereoScan “white light” (or “structured light”) scanner", whatever that is.



Obviously much better, but I'm assuming the cost is up there.  It's also a scan of something of decent size (a carburetor).  We could debate if this would be usable as-is.

Jason

[MK]

The problem with articulated scanning is now the information needs to be interpolated by the software.  You're going to get different reference points for the same spot which won't exactly line up and the software needs to decide what to do. 

That's not a problem.  The software can handle that very easily.  Check the scans at the end.  They are for a full set but most of the time it's for one tooth or implant.

/>
I've used the software and scanner for a few years but no longer have access to it else I would plop down a freight and try it out.  I'm not saying 3D scanning is a cure all but most people here don't know the capabilities of it and just assume.

Here are some of the things machined directly from 3D scans.  No clean ups, etc.

[wazzou]

Obviously much better, but I'm assuming the cost is up there.  It's also a scan of something of decent size (a carburetor).  We could debate if this would be usable as-is.

Jason

If you're using the resulting piece on 1/24th scale model, I'd say the carburetor would be more than useful, but at what cost?

[wcfn100]

That's not a problem.  The software can handle that very easily.  Check the scans at the end.  They are for a full set but most of the time it's for one tooth or implant.

Well, again, that's for an organic shaped object, so it doesn't necessarily apply.  For irregular shaped objects you can use some simple tools to smooth out rough areas or just remove mistakes and let the software repair that area.


Here's a good read about the total process for anyone who's not picturing all of this.

http://www.absolutegeometries.com/3D_Scanning_file_output.html

There are also tools available to use the scanned object as a reference to make a parametric object (what we want), but you're basically just redrawing the object.

Jason

[wcfn100]

If you're using the resulting piece on 1/24th scale model, I'd say the carburetor would be more than useful, but at what cost?

I agree.  And maybe there are some items that could be scanned and would be usable in n scale without much extra work if the cost isn't prohibitive.


Jason

[GimpLizard]

And, when you scan you get a solid object and 3D printing gets expensive when its a large amount of material.  Cad models normally are hollowed out.

Juts my 2 cents! James

Not true. CAD files are not referred to as "solid" models for nothing. Section any CAD model, and you'll see that it's solid through & through. The only exception would be a shape that has been "shelled". But even then, the walls are "solid".

[Simon D.]

This guy in the UK goes to exhibitions, scans life size and creates 3D figures
https://www.modelu3d.co.uk/3d-scanning/.

[wcfn100]

This guy in the UK goes to exhibitions, scans life size and creates 3D figures
https://www.modelu3d.co.uk/3d-scanning/.

I wanted to do that for my son's Pinewood derby car this year but I couldn't find anything close by.

Jason

[thomasjmdavis]

This guy in the UK goes to exhibitions, scans life size and creates 3D figures
https://www.modelu3d.co.uk/3d-scanning/.

Can he "edit" the extra 50 lbs I've put on in the last 40 years?