Thanks for your input. I dont know if you have used scanning Bureaus before. But I have. For example this was a persons head and we were issued with a Step file http://flynn-product-design.com/portfolio/electronics-enclosure/
So I know it can be done, its done all the time for Reverse Engineering purposes.
As chopmeister said. It can be done. But with human help and alot of work. Its like you modeling a surface over the mesh from the stl.
So a simple converter program will not bring a useful result on complicated shapes i guess.
But i would be glad if someone proved me wrong 
Chopmiester said there was no solution to what Im looking for..but all were looking for is getting objects scanned to receive STEP or IGES. Which clearly is part of most scanning services.
As Desktop Digitising is growing, I was hoping that the group may have experience in doing the same . Just noticed that I could buy a Cubify Sense for the same cost as a single scan via a Bureau...though a large difference in quality.
Rhino has a Mesh to Nurbs command, but it is slow and I've never tested it with complex geometry.
There are plenty of ways to do this though - they just require some work. One workflow people use T-splines for, for example, is to pull and stretch a t-splines object to the important vertices of a mesh file.
There is reverse engineering software out there which enables you to use an scanned meshes as a base point for solid modelling. I think even SpaceClaim can maybe do something like that. It doesn't depend that much on polygons (since a flat plane can have 100k polys) but on the features. But you still have to model most of the stuff if you want it to be any good. And reverse engineering is mostly used in... well... engineering! So it's a particular kind of objects which are suited for solid modelling anyway, which makes it all a bit easier. Organic shapes are a different beast entirely.
Imagine a rough rock wall. While you can easily model something like that, or 3d scan it, there is no way to do it in solids, since it would require hundreds of thousands of surfaces, which (unlike a polygon which has only 3 vertices to it) have loads of data describing them. In contrast, a face like on your link would require two dozen or so carefully placed surfaces to recreate (which isn't really that bad), and nonetheless somebody must have put some manual labor into it.
So in reality the answer is yes and no. Yes, it's "possible" but on a limited number of scenarios, and no, there is no easy way of doing it. I hope that clarifies what I meant in the first post. Or to elaborate: while it is certainly possible to recreate an object in solids based on meshes (which is what your scanning bureaus do), it is virtually impossible to "convert" it in a standard way like converting video or audio formats, which is what I thought you were asking about. 
And no, I haven't used a scanning bureau myself, but I do 3d modeling for a living, so I usually try to help with 3d related subjects here on the forum.
Still, I think my previous question has merit - why do you need that shape in your solid modeler? Maybe we can find an easier and more effective solution if you elaborate your specific situation a bit.
Just tested the Rhino command with a 18,000 face mesh - the results are pretty good.
It was a http://www.thingiverse.com/thing:157592 I had previously designed in T-splines, and then converted to a mesh file.
Rhino was able to bring the mesh back to a nurbs surface very cleanly and without any noticeable loss.
Original mesh:
Converted Nurbs:
https://www.dropbox.com/s/3ryacd44g0fzmde/Urban%20Puukko%20v1.7%20-%20Converted%20Mesh.stp
Nice to know Rhino can handle itself nowadays, haha. That wasn't really the case years back when I had some experience with it during college. Back then, it would usually just hang. On any command.
But still, that's brute force swapping each small face with a surface, which isn't usually what one would want (and which is exactly what one of those two apps in the first post does). You can't use that STP file to perform basic solid body actions like modifying fillets or chamfers and such, since there are none.
And SpaceClaim's Repair feature on that STP still reports 195 tangency errors, 39 extra edges, 8 inexact edges, and completely hangs when trying some other repair functions. :mrgreen:
But of course, that is beside the point if all you're looking for is that the object simply loads into your app of choice, to use for reference or something similar. In that case it's more than good enough of a solution.
thanks to all. Chopmiester, I understand your getting into accuracy and semantics. video interpolation is not like for like, more a transformation is taking place.
And Thanks Nick for going to the effort. I couldn't open the file first try. Will try again.
Cant go into specifics, but it is useful in hand tool design to be able to work with a material like airdry clay and then get it scanned to bring into the product development workflow. If the surfaces are averaged / smoothed as a result then this is a win.
Will investigate the options you mention. and also have emailed Instep to see if they can handle organics.
Appreciate your input.
And Thanks Nick for going to the effort. I couldn't open the file first try. Will try again.
Long story short ( the step file does take a while to re-import in solidworks, but it definitely builds fine) the converted files look identical to the knife handle in the photos of my first link.
I have lost touch with Rhino. are there any tools there that could help even out the mesh. or average. Have you used http://www.meshmixer.com/%20?%C2%A0
So we have a more usable Nurbs file in the Solid modeller.
reducemesh is what you're looking for in Rhino, though I have heard from others that Meshlab (free software) is a better option for finely controlled mesh reduction.
Thanks to everyone for your help. I have found a vendor in the UK with a decade experience, Aerospace / F1 standard equipment who offer to create "automated surfaces" ie output a step. ( mesh to Solid ) which is what was discussed here. For just over the cost of a spool of filament. So looks like the search is over for the time being.
Great that you have found someone to do it at such an affordable price!
Well fingers crossed all will be well. I did receive an offer of help from a group member so thanks again for that.
on my travels I did contact the makers of Instep and received a indepth reply if people are interested I thought I might share it . At this stage, I don't think I will require this after all.
"Thank you for your interest in the InStep application.
I took a closer look at the file you provided and there are a few things to mention in regards to your request.
· As I am sure you are aware, parametric modelers will only work in a parametric manner with native files, so any time you import a file (Step, Iges, Parasolid, etc.) you will not get parametric data but rather surface data (basically a BREP). In most cases this is perfectly fine since the intent is to simply use the imported data as a starting point to then do something else with it, but I wanted to be clear about this distinction. Some packages will offer ‘direct’ modeling tools which allow you to interact with un-parameterized features…
· The source file is fairly large by any measure. There are some 260k facets which is above our recommended size (usually applications will be fine working with files up to about 100k facets; some more powerful applications can go far above that though). In order to work efficiently with the data, especially if this is just a starting point for some project, I would recommend reducing the density of the file to something more manageable prior to conversion. There are a number of (free) applications that can do this reduction in data with only little loss of accuracy (though in some cases issues can be introduced). Blender and MeshLab are the more prominent ones.
· The original data actually has some slight gaps and offsets which make direct conversion impractical. In order to close these gaps, I applied a tolerance value to the data which allowed the body to be recognized as a single, solid body.
In general, the InStep application does not manipulate the underlying data. Rather it performs a uniqueness conversion whereby vertices in space are collapsed and thereby features such as edges and surfaces recognized (the STL format consist of triangles only so several corner points can overlap or be in close proximity, the application merges these together and makes both vertex and edge information unique). Therefore the generation of smooth surfaces is (not yet) implemented and the output is just a different format of the original geometry with some corrective measures applied. This is fine for cases where the object is geometric in nature (boxes, flat covers, etc. where larger surfaces can be merged without loss of detail) but does not do well with organic shapes like the one you have.
In those cases, additional work is required to manually (with different software) generate NURBS surfaces from the data. The issue is still that there is a loss in accuracy in this step since these applications use some shortcuts to get the smooth surfaces where there were previously sharp edges.
I wanted to showcase the differences in these approaches and files and put together a sample of these (I’m afraid I cannot give you the full files for free since that is a service we offer).
Due to the file sizes, I went ahead and placed them on our website (I will remove them as soon as you can let me know that you have accessed them):
https://www.solveering.com/temp/chrisflynn_tempFiles/owl_010RED_Partial.zip
This is the file you provided, reduced to 10% of the original density (with Blender) and then imported into InStep, exported as STP (and then cut to the given size in NX7.5).
https://www.solveering.com/temp/chrisflynn_tempFiles/owl_InStep_Partial.zip
This is the full density file, converted in InStep, loaded and manipulated in NX7.5 and re-exported as STP.
https://www.solveering.com/temp/chrisflynn_tempFiles/owl_2_NURB.zip
Full density file, manually converted to NURBS surfaces.
There is obviously quite a difference in the file content and size for these different items, hopefully this explains the different options. It is our goal to be able to offer InStep with the option to generate NURBS surfaces directly from within, however that capability is still some ways out and currently only available as a service.
Please let me know once you have downloaded and viewed the files and if I can be of further assistance. Should you require us to perform conversions like this, I would be happy to provide you with cost options depending on the details of your project.
Kind regards,
Ben
"
ArunC posted a topic in UltiMaker Cura,
ArunC posted a topic in UltiMaker 3D printers,
Recommended Posts
chopmeister 0
You can't convert STL or OBJ to STEP. STL and OBJ are polygonal mesh formats, while STEP is a solid body format.
Because STEP doesn't work in a way polygonal meshes do (it's basically a lot of mathematical functions describing surfaces) you can't scan into STEP.
3d scanners output point cloud data which then gets triangulated into a mesh, thus you get your STL, OBJ, 3DS, or whichever mesh format you choose.
You can go from solid to poly, but not from poly to solid. The solutions you mention here are actually "hacks", ie. they work by detecting (with the user's help) features on the model that can be easily described with surfaces and recreate a solid model based on approximations of those features. It works with simple models, and low poly models, but not much else.
There is no solution for what you are looking for.. The better question is what do you need to do with organic forms in a solid modeller anyway?
Link to post
Share on other sites