Looking into possibly getting a Sense 3D scanner for work to help with reverse design of existing parts.
I wanted to know if anyone had any experience with this scanner?
I see it exports to a .stl format. What would be the best way to turn this into a solid using Solid Edge.
Looking forwad to running ST9
Thanks in advance for the help and suggestions.
I tried one of the 3DSystems Sense scanners a few months ago when a coworker brought one to work. It is pretty finicky to use and really only good for organic shapes due to is fairly low resolution. If you scan a mechanical shape with sharp corners, it will tend to round everything off into a blob. It also loses it's tracking easily so you have to restart the scans often which takes a lot of time.
I'm far from an expert in this area, but turning the 3D scan into a usable CAD solid model appears to be a bigger problem than the scanner. Solid Edge does not currently include any tools to do this. I also know some folks that do 3D scanning for a living and they tend to use a hybrid approach of taking sections of the 3D scan, referencing the sections to develop sketches, and then turning the sketches into 3D geometry (I think they are using SolidWorks). I asked them specificity about other dedicated 3D scanning software such as Geomagic Design X (costs 4-5 times as much as Solid Edge) and they said they did not have much luck with it.
Thanks for the input.
With the growth and affordability of 3D printers and 3D scanners becoming, you would think that there would be more of a push to make CAD, Scan data, and printers more compatiable.
Maybe the Solid Edge brass and programers can come across this post and give us some feedback on using 3D scan data in Solid Edge.
They were actually soliciting feedback on this forum awhile back. See link below:
3D scanning is a HUGE can of worms.
First, you have to know exactly what model results you are looking for.
Second, you have to find the right software to process the desired results. Each $35,000 reverse each engineering software has a specific set of results in mind. for exmples:
Organtic 3D enclosed polygons: Geomagic
Extruded steel shapes: Cyclone
2.5D Tins for civil: Mixture of Imageware and Audodesk civil
Archtecture: Mixture of Cyclone and Couldworks for tracing.
Once you figure out the best software for your shapes, then you have to find a scanner that works with your software.
In general the accuracy of laser scanning is range/10,000. Fo that's a function of the scanners range.
Also you have to plan out the registration process (fitting the clouds together). Sometimes another instrument is used like a total station if working with large scales and civil target. If working with smaller scanners design for say Geomagic. they will use an automatic registration knowing how the scanner is set up.
Also moving the objects from the reverse engineering software to the desired result format can also be a challenge.
So if your answer is you want to make anything you can throw at it at any scale. You will need 4 different $35,00 pieces of software with 3 months learning curve each along with many at least one traditional modeling software to do some typical modeling work to fill in the missing pieces.
Oh, and get the fastest computer you can possibly afford. I'm not kidding. Plan on a at least a $5K budget per PC.
Attached are a few things about my experience with laser scanning. I will never touch that stuff again, I don't like waiting on computers.
STL format is nothing but a bunch of triangles between points. Files will be much too large to load in any CAD program. The hard part of processing scan data is reducing the number of significant points by making assumptions. Say you scanned a cube. 2 points define a perfect cube, 8 points define a box, but is your box really flat true square, normal, etc.... you may need 500 points to define it's real shape within a tolerance. and you likely started out with 8,000,000 points (8 scans 1 million each) to get complete coverage.
Here are the main problems with laser scanning.
1. A single scan is NOT 3D, it is 2.5D. You can't scan an undercut or surface behind in one shot
2. No corners are ever captured, they must be inferred using other comparative geometry (Like a 90 deg corner), Even if you calculate two planes and intersect to create a corner, the two planes will not be 90 deg and normal to the coordinate system......by just a small bit, unless you traced the corner looking normal to the cloud based on an established coordinate system.
3. Lots of noise. Reflections from water, mirrors, chrome are a problem. Also when the laser half catches a corner, you get lots of points between the corner and whatever else the other half of the laser measured. Scanners also have a range of tolerance that is larger than desired.
4. Files are HUGE. Most scanners are 1 million points each. that's good for 4 to 20 features, not 100's. Most solid objects require 8 to 16 scans to cover. Can you load and rotate 16,000 points at once on your computer to start segmenting the clouds for processing?