Anyone done 3D scanning?

A guy was in the shop recently talking with me about a project I am working on, giving me some advice on controls/drives etc and asked a question about the parts. I showed him some parts and he asked how I made a model, or how I made something from the parts. I said I painstakingly inspect them and make a 3D model.

He asked why I didn’t buy a laser scanner and scan the parts?

I’m an ignorant dinosaur, that’s why.

Anyone do this or have this done? The parts are not high tolerance by any means, and will need finish work regardless. But he seemed to think a cheap laser scanner would work OK, maybe with a little tweaking.

What I could find online was not encouraging. A “cheap” laser scanner runs $1500, with a high quality one pushing $50k.

Fal Grunt said:

A guy was in the shop recently talking with me about a project I am working on, giving me some advice on controls/drives etc and asked a question about the parts. I showed him some parts and he asked how I made a model, or how I made something from the parts. I said I painstakingly inspect them and make a 3D model.

He asked why I didn’t buy a laser scanner and scan the parts?

I’m an ignorant dinosaur, that’s why.

Anyone do this or have this done? The parts are not high tolerance by any means, and will need finish work regardless. But he seemed to think a cheap laser scanner would work OK, maybe with a little tweaking.

What I could find online was not encouraging. A “cheap” laser scanner runs $1500, with a high quality one pushing $50k.

Click to expand...

I have never done it, hopefully someone with experience will chime in.

My understanding is a cheap one will give you poor results that will need alot of "fixing" to get you a good solid model (or surfaces?)... I don't have to build anything too complex, but I can normally build the models (when needed) in an hour or so. I think unless you are doing alot of freeform type stuff, using a print and extruding/revolving/cutting with solid tools is the way to go.

We have 2 Hexagon Romer arms in the plant for stamped part verification (comparison with true model) for a quick once-over every certain number of parts stamped. It works well.

What you're wanting is an exported model and will take special software like PolyWorks, which will give you a STEP or Parasolid (or whatever) to work with. That software is not cheap and is in addition to the cost of the arm. This is mainly because the base software is incapable of generating a NURBS surface.

I only replied to make sure you ask for the model creation software if/when you talk to a dealer about the arm.

I only use 3d scanners for mesh surfaces when I am doing aesthetic work and I usually have to clean them up a bit before they are totally usable.

Like Rewt above said, if you want a model with features that takes another whole level of filtering (read: cost$$$) so like you I prefer to just re-engineer the model myself or find whatever remnant of cad the designer may have given if it exists and go from there.

my experience is nearly 10 years old with this, so there are probably more software solutions out there now to make the work flow easier, but back then the scanner itself (supposedly it was some higher end hand held thing, which I never saw myself) will output a point cloud data, then, as Rewt said - it is up to post processing software to generate a solid model, Solidworks can import the point cloud data and you can then start to make sketches to create the model using those points in space for reference, but I had to scale the point cloud to match a known dimension on the part, but the biggest problem was that a high detail point cloud data can be huge (and Solidworks would stall trying to process it), I was provided with a 200mb file for a rear differential cover from a BMW to make a solid model for, and rear subframe scan from the same car was around 800mb (and less detailed than the diff cover), I remember doing some filtering on that data to make it easier for SW to swallow, don't recall what software, might have been Blender

overall it was handy to have the scans, but not really necessary for the work I was asked to do, might be more useful if the part had features sticking at odd angles in every direction without a good flat surface to reference off, something like a thermostat housing for a car engine, those usually have quite "irregular" features, cylinder head scans might be another good thing have to create models to port said heads etc.

I have a David 2 scanner (I believe HP owns them now) that works quite well. It is set up on a tripod and uses a projector and a camera (structured light). I've only scanned smaller objects on a rotary table. I found the tighter you control the rotation in the scene the better the mesh output.

A few of the others hit on the most important point that I've found thus far: the post-processing software. I've used a few different types and everything pales in comparison to Geomagic. Geomagic costs somewhere around $10k. There are other ways to get to a solid model but if I had to do it on a semi frequent basis, geomagic would be what I'd purchase.

Fal Grunt said:

A guy was in the shop recently talking with me about a project I am working on, giving me some advice on controls/drives etc and asked a question about the parts. I showed him some parts and he asked how I made a model, or how I made something from the parts. I said I painstakingly inspect them and make a 3D model.

He asked why I didn’t buy a laser scanner and scan the parts?

I’m an ignorant dinosaur, that’s why.

Anyone do this or have this done? The parts are not high tolerance by any means, and will need finish work regardless. But he seemed to think a cheap laser scanner would work OK, maybe with a little tweaking.

What I could find online was not encouraging. A “cheap” laser scanner runs $1500, with a high quality one pushing $50k.

Click to expand...

i've got about 5 years of scanning/modeling/reverse engineering experience. its a pretty involved process but once you learn it, not too bad.
depending on what it is that you're scanning will determine what kind of scanner you need and how expensive itll be. some of the cheaper SLS (structured light scanners) can get pretty accurate (~5thou)but are limited to smaller parts. if you give me an idea of the types of parts you're looking to do i can give you a recommendation on a suitable scanner.

We always ask a specialized company to do the scanning an deliver the "cleaned files". Today the cost for this kind of work is pretty accepable and they always use state of the art equipment. They come on site or we send the parts to them.

We recently bought a scanner as we are doing a fair amount of pump impeller and pump reverse engineering. We have used it on a few jobs so far (not pump work) and there is definitely a learning curve. We have had a fair bit of work scanned for us in the last few years some pump jobs and some really detailed decorative castings. If at all possible you want to model on top of the parts as the rather than importing scanned surfaces. Some of the decorative work we had to use imported surfaces and the files were huge and often very difficult to modify and do work with.

The Scanning companies we used all used Geomagic which is really expensive software but the way to go if you need to import surfaces etc. I tried a demo version of Geomagic for solidworks which was a cheaper option but I did not have a lot of good experience with it. Solidworks crashed more on me in that 2 weeks than in I get in years of daily use. I had more crashes even when I was not using the add on but it was installed.

About 6 months ago I discovered a Solidworks Add in called 3dxtract from a company called Polyga. It was pretty cheap (around $1000) and we really like it for working with scan data. It works in Solidworks like solidworks. When we scan or get imported data from someone else we don't worry about getting a watertight solid scan. We just use the scan data as a reference to build a solidworks model on top of. Trying to do that by importing an STL file directly you choke solidworks up if it will even import. The first impeller we modeled with it we had someone else scan the part for us. The scan was missing a bunch of data and even with it being the first time we used the software the modeling was almost as fast as modelling from a drawing. https://www.polyga.com/xtract3d/

I find that businesses that actually need that kind of thing have it already and pay a lot for it.

I also find that a lot possibly book smart folk think that some kind of tool like that is a requirement or "the best way" to reverse engineer stuff when the reality is they don't have a clue what they are talking about.

Personally, I really enjoy reverse engineering stuff and can't see the need for any kind of scanning thing for the parts I work with. My reverse engineering tools are calipers, radius gauge sets, tons of gage pins and a really old open style VMC that's still plenty accurate with a DTI stuck in the spindle.

I've had other machine shops ask me where I got my geometry from for the some of the products I make and it kinda makes me scratch my head.

One of our customers has subbed it out on a couple machined castings that they didn't have prints for. Critical features(ie. bearing bores) were still measured and modified by hand, but I think they were getting something like +/-.005" on feature sizes and locations. Which is completely fine for the sand castings. I believe it cost about $1k-$1.5k. Parts were about 8"Ø by 10" tall, with a bunch of features. They seemed pretty happy with it.

Fal Grunt said:

....
. I said I painstakingly inspect them and make a 3D model.
He asked why I didn’t buy a laser scanner and scan the parts?
I’m an ignorant dinosaur, that’s why.
...

Click to expand...

Or because you are smarter,
Idiots think 3-D scans so easy and time saving.
They are not but can be .
So much depends on the part features and tolerances. Are you making a coffee cup, a face statue, or a valve body or crankshaft?
Try this with a oh-so very simple one cylinder lawnmower engine block.
But a thermostat housing for a small block. Outside is within such as a who cares and you only have to fix the mounting face so great use.
I do not know your parts but have built both laser and probe scanners for people. If you have the big money and time spent I can get you within a a handful of microns in the model.
You do not want this.

Laser scanning caries a whole big world of accuracy. From a trusted 0.100 to microns.
A great tool but holes you may not expect.
Bob

Is this like a Faro? I was wanting to do a V-8 engine block and a guy said he could do it.

rickseeman said:

Is this like a Faro?

Click to expand...

No, automated laser scan, miles different.

When I was referring to the Romer arms, those were for the steel room floor, and their tolerances are only give or take 10 thou on some and 4 thou on others. Depends.

I reverse engineer the tooling that makes those parts. We use a bridge CMM with a Nikon laser scanning head. I can get it down to a few microns if I take my time.

It all depends on what the OP wants to reverse, like Bob said, coffee cups or main journals?

This is all an interesting read

I had considered scanners a few years ago when the prices started to come down, thought I might have some alignment with my current business.

I think I am glad I didn't bother

CarbideBob said:

Are you making a coffee cup, a face statue, or a valve body or crankshaft?

Click to expand...

My neighbors had a business doing this several years ago ... they mostly did faces and other artistic stuff. They used Maya and some special software they wrote to clean up the point cloud and it was quite a job. I'm with Garwood on this, if you make a model you can manipulate it. A scan is a single entity that you can't do much with and it's no less work. For something swoopy maybe but for normal mechanical, way too much work with a less-then-optimum model as a final product.

EmanuelGoldstein said:

My neighbors had a business doing this several years ago ... they mostly did faces and other artistic stuff. They used Maya and some special software they wrote to clean up the point cloud and it was quite a job. I'm with Garwood on this, if you make a model you can manipulate it. A scan is a single entity that you can't do much with and it's no less work. For something swoopy maybe but for normal mechanical, way too much work with a less-then-optimum model as a final product.

Click to expand...

At IMTS 2018, I think it was Hurco that had a laser scanner on a robot arm, and they'd scan your face and make you a model of your head, in color. It was neat to watch, but the line was way too long for me to care about having it done.

I just had a crown made by my dentist. First step is they use a handheld scanner the size of a small flashlight and run it all over that part of the teeth. It generates a 3d image of 3-5 teeth. that image is sent to the lab who casts a gold crown to fit. They make a rubber impression after the tooth is ground down.
I assume the bottom is milled to fit the rubber impression while the top is machined per the scan. They may scan the impression at the lab and do it all with machines and no handwork. I assume the crown is cast near net with lost wax?
Bil lD

My very-limited experience with laser scanners is on medium-to-large objects, in an industrial environment rather than architectural or geographical use. And it's been several years.

That said, the Nikon "Laser Radar" was then the absolute belle of the precision ball. Point-location uncertainties of the measurements, which are taken in scanner-native spherical coordinates (horizontal angle, vertical angle, distance from scanner to point), were 2 arcseconds mean-square-error for the angles and (0.001 inch + 0.001 inch per 10 feet of distance) . . . IIRC. At the other end of the spherical-scanner spectrum, Leica offered a Disto scanner that was marketed to the home remodeling contractors to measure the as-is dimensions of rooms that are to be remodeled . . . with point-location uncertainties of 8 millimeters or so.

As I'm remembering prices, the Nikon Laser Radar started at around $US 800,000 while the Leica Disto sold for around $US 8,000. FARO and others offer a number of different models in between the two I've mentioned.

There are also a number of hand-held laser scanners that can be used either on the end of an arm-type CMM or in conjunction with laser-tracker or photogrammetric measurement systems. Not really familiar with either of these technologies, but they do exist.

Furthermore, there are laser-scanner probes for conventional "Cartesian" CMMs. Again, not familiar with these systems.

I suspect that the laser-scanner CMM probe would suit your application, but I suspect that the investment in time and money would make such a system impractical.