I am having problems importing this STEP file. Not sure what's going on, but Alibre and Inventor import it ok.

The file is available here:
http://cid-b6766873553a783e.office.live.com/self.aspx/123D%20files/Triple%20Clamp%20v6.stp




Thanks for the help,
Nick

Hi Nick, unfortunately the STEP file format has a kind of different way of representing trim curves - the way the file structure is set up trim curves are only present as 3D curves and they do not necessarily map one-to-one directly to UV parameter space curves.

So in order to convert those kinds of trims into the more conventional trimming boundaries that MoI uses (where each trim curve has both a 3D edge and a UV parameter space edge to match it), those 3D curves need to be processed and one of the tricky parts is that 3D curves that cross over a closed surface boundary (like the seam edge of a cylinder for example) need to be cut at that crossing point.

That is a somewhat delicate operation and can be hard to resolve in cases where trim curves are kind of barely skimming along the seam.

It tends to be in that particular area where STEP imports can potentially get messed up when importing into MoI. If you notice those messed up pieces all have underlying surfaces that are closed surfaces with seam edges on them like a cylinder.

When that happens, it is can be possible to repair the file by doing some manual untrimming and retrimming on the pieces that have mangled trim boundaries.

There is a tutorial on how to do untrimming and retrimming here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=446.17

But probably the easiest thing to try in this case is to see what happens using IGES or SAT format instead, those formats do not have quite the same kind of trimming issues as STEP format, it is possible for there to be some other kinds of issues instead, but hopefully one of those other formats will handle this particular shape better.

Let me know if neither of those formats helps get it into MoI, and I will try to help you get the file in using some other program as an intermediate step, or I can help do the repair work on it if that's the only option.

- Michael

I actually already tried the *.sat file and it's has exactly the same results. I can try to use an intermediary, but I was hoping to keep the conversions to a minimum...

Thanks,
Nick

Also, was the file created in Alibre??? Alibre has some proprietary stuff in it's format for assembly purposes.. I had originally thought it was the ACIS kernel stuff and SAT, but I think it may be a version of STEP....

If it was the STEP, look if there is 2 or more STEP version exports, or if the model is an assembly somehow....

Hi Nick,

> I actually already tried the *.sat file and it's has exactly the same results.

Did you also try IGES? IGES usually has surface trimming boundaries that are closer in structure to how MoI needs them, so that will not as often have that same type of trimming problems. Instead the problem becomes processing the individual surfaces to join them together, but that tends to be most difficult if you've got some tiny slivery surfaces mixed in with other ones hopefully you don't have that in your particular model.


> I can try to use an intermediary, but I was hoping to
> keep the conversions to a minimum...

Let me know if you are still stuck and I can try to use some intermediate steps as well.

- Michael

Hi Nick, so looking at your model more closely it unfortunately does appear to have some little tiny slivery surfaces in it, which tend to create a lot of various problems in translation.

I opened your STEP file in ViaCAD and exported into MoI by IGES format, (turning off 186 type export), and it can import but it is not easy to get it to join up, because there are some little slivery parts like this:





I also zoomed in inside of ViaCAD just to verify that those little slivery surfaces are part of the geometry and not something generated by MoI at import time, these screenshots are from ViaCAD after the import into there:





Those little slivery bits are going to probably make things kind of difficult.

Where did you originally create this object? Do you know if there is any way in that application to produce a cleaner object structure that only has broader surfaces for those fillet areas instead of having a tiny sliver surface near the ends of some of them?

Probably this will have to be joined in strategic pieces, probably trying to join the little slivery parts with their immediate neighbors first before trying to join larger chunks of the model together.

- Michael

Hi Nick, I've attached a 3DM version of your model that you can open with MoI.

But it has the object as a bunch of separate surfaces, it's going to be fairly difficult to get it all joined together into a solid due to several little slivery surfaces in there.

Little slivery fragments make things much more difficult to join because it's kind of easy for the opposite side of the slivery bit to get fused on to an adjacent piece. One thing that can help with this is to join little slivery pieces to their smallest sized neighbor surface first before joining that to the larger pieces.

What is it that you intended to do to this model after importing it into MoI?

- Michael

May be helpful in this matter: Triple Clamp.stp import with SWX 2011
- 47 bad surfaces
- 36 gaps

Test with SWX "Auto Repair":
- 15 bad faces/surfaces
- 7 gaps

It was created with 123D ;-) It's suppose to be solid and in fact that is how Alibre and Inventor import it and the dwg in AutoCAD. I hadn't noticed the little slivers in the slot fillets...

I was importing to MoI to do some embossing. MoI is top notch for doing that sort of thing (among others).

Thanks,
Nick

Hi Nick,

> It was created with 123D ;-) It's suppose to be solid and in
> fact that is how Alibre and Inventor import it and the dwg in
> AutoCAD. I hadn't noticed the little slivers in the slot fillets...

Yeah it also imports into ViaCAD as a solid as well. One thing all those have in common is that they are all ACIS based. I wonder if their importers see that the STEP file was also created by an ACIS modeler as well and skip doing some of the more complex (and therefore more delicate) types of processing that can get messed up with little slivery surfaces.


> I was importing to MoI to do some embossing. MoI is top notch for
> doing that sort of thing (among others).

Will the surface model that I attached above allow you to get that job done? You could probably use those surfaces for constructing some additional cutting object pieces, and then you could import those cutting objects into 123D to combine them into the model there.

- Michael

Hello,

It looks like there are surfaces that haven't been trimmed properly inside the model. It was not too difficult to get rid of those slivers in SCE, but the surfaces stuck inside that model are tough.

OK, so I got the paw as a seperate solid from the main body (new solid). I cannot get the two to merge. When I try to Boolean union the pieces I get a pile of surfaces. What's going on?!?

Hi Nick, well one thing I notice at first is that your paw surfaces seem to be made up of a lot of little flat facets, like this area here is a bunch of little fragmented flat pieces:





And for example this little piece here which would ideally be made up of something like 6 faces is instead actually made up of 43 faces:



How did you go about creating those paw pieces, did you possibly use some original imported curve from a PDF file or something like that? If so you might check that file to see if it has actual curves in it or if the things that look like curves are actually just polylines with a whole bunch of little segments in them to make them look as if they were curved.

You generally don't want to directly use dense polylines like that directly as the basis for surface construction, you'll want to get them to be actual smooth curves so that you don't end up with things like small pieces made up of 43 little faces when just a few curved faces would make for a much cleaner structure.

A cleaner model structure would definitely make it a lot easier for the boolean mechanisms to do a good job - it's much harder for them to make sense of so many little slivery bits like you've got here.

I'd also double check to make sure you don't have any pieces that are so thin and slivery that they actually make for a kind of degenerate self-intersecting solid, anything that has totally collapsed degenerate surface fragments in it will cause difficulties with intersection calculations.

- Michael

Hi Nick, also looking at the underside of the paw pieces, the surface underneath (the one that you're trying to mate with the top surface of the larger piece with the boolean union) seems to be fragmented into a couple of surfaces:

I turned on control points for those surfaces, and it looks like the underside pieces are coming from a surface of revolution?

I turned on control points for this piece here:



And it appears to come from a surface of revolution like this:



The piece you're trying to mate it to with boolean union also is a surface of revolution but it is rotated to a different angle, the surface control points look like this:




Having that small piece at a different rotation than the large piece especially with the closed seam edge of the revolved surface running right through the middle of the piece (making it actually have 2 separate surface pieces for each face on the bottom of the paws) is going to make things more difficult for the boolean code to figure out - the booleans try to intersect objects with each other so it has to figure out that those pieces are supposed to be coincident - if they both came from the same original surface (instead of different rotations) that would probably be easier for them to figure out.

A couple of different methods which you could use that would probably work better would be to make the paw pieces actually push a little ways through the top surface instead of making them coincident - that will avoid complications in coincidence determination.

Also another possibility for stuff like this where the intersection is difficult is to use the Trim command to cut pieces by each other and then join the results together.

- Michael

Hi Nick, I was able to get your paws glued on properly using the Trim and Join method instead of the booleans for this case, I've attached a 3DM file with the result.

This technique is something that can be used to get a result out of situations where the booleans are getting confused for some reason - in this case the booleans were most likely confused by the coincident surfaces not being identical, particularly with one's seam running through the middle of the area.

So to use Trim I went to the paw pieces and selected these faces here:



And then just deleted those faces to make those paw pieces be open surface fragments that look like this:



Then I separated out the upper surface of the main body piece that you want to connect them to, and selected that surface and ran the Edit > Trim command:



Then for the cutting objects I selected all the open edge curves of the paw pieces to use as cutting objects. That then cut the surface so it looked like this:



Then I glued all those pieces together back into a solid using Edit > Join to get the attached result.

One trick for the trim stage is that instead of picking all the little edge curves one by one in the stage where you're selecting cutting objects, you can use a script to do it with just one keystroke. To do that, go to Options > Shortcut keys, and add in a new shortcut key (I use N for "select naked edges"), and then for the command part paste in the script shown here:
http://kyticka.webzdarma.cz/3d/moi/#SelectNaked

That allows you to select "naked" edges (edges that only belong to 1 surface and are not joined between 2 surfaces) with the N key after that, it's a handy way to select cutting curves in a situation like this, and it's also very useful for analyzing which particular areas of an object are preventing it from being a closed solid object.

Hope this helps!

- Michael

Geez... this whole thing is becoming a nightmare... the paw is from a pdf (vector) from the mfg. that I projected (direction) to the surface you created earlier. I then trimmed > extruded > exported to *.stp and iserted into an earlier version of my 123D file (pre-paw) Everything worked great up to this point. I then constrained (tangent assembly constraint) the paw to the spot it was located (worked great, bing-bang-boom). If it's any consolation, 123D would not boolean them together either. I thought it was a shortcoming of 123D so I exported (both bodies as constrained) to *.stp and opened in MoI. Here we sit.

;-\

I guess I should start over with tracing the pdf with some new curves...

At least the body imported correctly this time... ;-)

But next time around for something similar probably try to make the pieces push through each other a little bit instead of having them kind of skimming right along each other. Either that or try to get the exact same surface in the areas that are coincident instead of having one rotated or fragmented differently than the other.

Having them push through a bit tends to be easiest, since it then doesn't make too much difference exactly how the bottom pieces are structured since they will get sliced off anyway.

- Michael

Hi Nick, well I was actually able to get the pieces unioned together in MoI (using Trim + Join method described above), so if you're just looking for a solid result of what you had that's attached above.

But yeah it's good to be kind of wary of PDF data - often times PDF files are really set up just for printing and not really as a source of 3D geometric calculations in mind.

There are quite a lot of things that can be ok for printing but bad for 3D modeling.

Also if the PDF file was generated from one of those PDF printer drivers, those tend to fragment things that were originally curves into a bunch of lines, since that is pretty common for an application to only actually send lines to a printer. So that's why you may rather often see PDFs that have have only polylines with a dense number of line segments in them rather than actual smooth curves...

- Michael

Heh... I found out 123D will and does boolean them. I was selecting the parent of the paw instead of the individual blocks in the browser... Apparently it's something they are fixing. Anyway, if I select each solid piece of the paw it booleans fine. The problem however still lies with exporting to *.stp where virtually every program broke it back into individual blocks that need to be booleaned again... I've learned my lesson with pdfs.