Rapidworks exclusively for Nextengine 3d scanner

[Elrick]

Good day,

We recently got the Nextengine 3d scanner to help transform parts into accurate CAD data at a faster, more reliable pace for our assemblies. Theres a bit of a variety in aditional software and hardware to help with the scan and alignment process whereof Rapidworks is one. The "3ds Geomagic" badge is at the top of the 897 page user guide but I cant find many usefull forums and threads related to this combination. CNCzone have some helpfull information but without enough opinions ( http://www.cnczone.com/forums/3d-scanni ... solid.html ) Is there someone here who could point me in the right direction or have some experience with this combo? How accurate is Rapidworks?

Its important to know wheter it will be a good decision to buy Rapidworks to fill in for that which Scanstudio fails to do (Scanstudio is the software used to operate the scanner). It doesnt have a lot of trouble outo aligning small 360 and bracket scans made with the turntable but manual alignments needed for the single scans seems to be problematic. If your alignments arent perfect you cant expect accurate results. Which is where the main concern lies for me. I need to scan large objects and there arent many proof out there to relate to.

At the moment Im not convinced about the accuracy of the single scan alignments done with Scanstudio. The scans and parts arent identical. Which could be seen in the geometry. I also measured imported parts to GMD and the difference over the width (760mm/30in) were about 50mm/2in. I might have made a mistake somewhere but I cant spend any more time playing around with Scanstudio. Its not very user friendly even though it might appear so.

I am however relieved to see Geomagic Design doesnt have as much trouble manupilating the surfaces generated by Scanstudio as those generated by Solidworks. It gives you 7 file formats to choose from when exporting. IGES has been flawless for GMD so far.

Any help questions comments welcome.

Regards

 

[Michael Moore]

Hi Elrick,

I've got an HD NextEngine with RW 3.0.3 (an older version) that I bought from a friend that I've been fiddling with for about 3 weeks.

I'm finding that learning RW is not going very fast -- it is another software package where the designers seemed to have "a better idea" which in spots invalidates common practices. An example of that is sometimes needing a double-click with the mouse to select items instead of a single click. I watched a short section in a tutorial video 5 or 6 times while following along and not getting anywhere before I finally heard the double-click going on. Much of what I've tried works fine and I have no problem, but the manual doesn't always give much detail on a topic that is new to me. A screen shot that doesn't tell me what each option is doing and when/why I'd want to use the different options doesn't get me very far. But then RW is not the only program to have that problem. I also suspect that the tech writers (who I don't think had English as a first language based on some of the errors) never sat down with a novice/naive user to see if they could make enough sense of the manual to let them do a task. I'm sure it would help if I had much more experience with surfacing so I'd be able to make informed guesses about what needs to be done.

You seem to have a lot of error on the measurement. I've got a motorcycle engine side cover that I've been using to learn on that is about 7" x 12" x 3" in size and the mesh is made up of 4 sets of bracket scans with small scans to patch a few holes. I aligned the model in ScanStudio. This was done in macro mode, and a spot that measures 7.13" is showing up in RW with a mesh measurement of 7.08 which is 99.3% of the true length. Perhaps the magnitude of your error is at least partially due to using wide mode with the lower accuracy?

I suspect that for reverse engineering I may need to treat RW mainly as a way to build a parametric model that is close but which will require validation and nudging of dimensions when completed. I think the money spent on the tools in RW is largely to extract features that are a good fit to a not-perfect mesh and not so much as a final CAD solution. After all, RW will do the "live export" to SolidWorks and other standard CAD packages/formats so that makes it sound like the work is expected to continue once you leave RW.

I've also got Rhino3D which is strong on surfacing and I wonder if it might be able to do much of what RW does, though of course that would give a "dumb" model and not a parametric model. But if all that is really needed is an accurate 3D model and there is no need for a parametric workflow Rhino might do the job for 1/3 the price of RW. But that's just speculation on my part as I'm far from being any kind of expert in either Rhino, RW or A/GM.

Oh, a NE support engineer told me to consider doing alignment in RW when there are large amounts of scan data. RW will handle about 3X the point/mesh data as ScanStudio. I've had some problems where SS won't go any farther with an alignment even though I can see the additional scan is there. But the problems were showing up when I was getting a 600-900mb scan family.

I don't know if that will be of any help to you but it is what I had to offer.

cheers,
Michael

 

[Michael Moore]

Elrick, you might look at the comments on RW here and elsewhere on the SW forum:

https://forum.solidworks.com/message/204280

For example:

Do not be fooled though, while rapidworks is extremely helpful in cleaning up your scan data and adding useful surfaces to the plots, it does not easily create a parametic model with a full feature tree. You are in essence creating all of the features and rapid works has smart tools to enable those features to 'snap' to your plotted data. Still useful, I just feel they made it seem in their marketing literature that it would magically build a feature tree.

I am surprised that NE doesn't have a user forum. This Alibre/GM forum has proven to be extremely useful with lots of helpful people. The NE support have been responsive but it would be nice to be able to get some feedback from customers who may have struggled with the steep part of the learning curve and have it fresh in their minds what helped them to get over the hump that was blocking forward motion. Sometimes experts are too far removed from the "thrashing around" stage.

You may find some information here:

http://www.laserscanningforum.com/forum/index.php

http://www.3dprintforums.com/showthread.php?t=191

cheers,
Michael

 

[Elrick]

Thanks for sharing your experience Michael! It would have been much more helpfull if Nextengine had a forum! It seems that the product is still in a trail stage for many reasons. Even though its been on the market for quite some time now. I also feel the user manual needs much more detailed explanations. I have no clue what all the settings do. Like they assume you know exactly what they are talking about. When I learned Alibre I never had so much confusion and uncertainty. On a logical base the user relies on the software to get the job done. User input only refines the results. After I didnt get any help from the sales person I asked on NextWiki support. They wanted me to send them all the scan data. 3GB for one part is too much. And you wont learn anything that way.

I used the extended scan to do the scans with lowest quality SD scans because its faster and uses less data. Only on parts I practiced on. But I think somewhere between buffing ans simplifying, the edge of the surface were trimmed too far. It is a surface I wanted to thicken in GMD. Another part I scanned in wide mode werent so far off. Over 700mm/27.6" it had a difference of 3mm/0.1". Its just a bit strange that NE claims to reach macron accuracy and then you get this. I believe the alignment functions arent sound.

Unfortunately we decided to have a refund. Saving time doenst appear on the list of assurance. At this time in our project we cant afford something that doesnt benefit our goals. I tried really hard to decipher this but it simply werent that easy.

For whose whod like to see I will add examples of the one part I believe was a success.

 

[Elrick]

More

 

[Elrick]

 

[Elrick]

 

[Elrick]

 

[Elrick]