Full Version: Loft then boolean weirdness

First off, if there's a better way of modeling to achieve what I want to do, please let me know.

Thanks Michael, I took your suggestion and am now tracing the DXF shapes of the wing ribs using through-point freeform and then working with those curves. I'm using inset to create the parts of the skin and rib that I want to 3D print. I printed a test using extrusion and it's amazing. You wouldn't believe how much strength this thin bit of plastic has!



Now we're at my next challenge.

The wing that I'm building is tapered - it narrows the farther out it goes. That means I want that D shape at the front of the wing to change size as it goes from one rib to join up with the next. My thought was to loft, and within MOI that looks great, but when I export it to STL and import into my slicer it's messed up.

I made what's in the attached 3dm by separately lofting the outer curve and then the inner curve, and then doing a boolean subtraction of the two. I tried making the loft as simple as possible, using 'straight' and 'exact'.

If I try to export the STL of that to my slicer tool it says it's not manifold (not a solid object)

I then tried doing a boolean merge with a straight line to cut off the D section I want for the leading edge, and that leading edge piece DOES import into the slicer correctly, but then if I try to also cut off a piece of the trailing edge, it won't even work in MOI - I basically get a copy of the whole lofted piece minus the leading edge part.






Clearly I'm doing things in a bad way... Is there a different technique I could use to achieve the same result? I tried to do the loft using the outer and inset curves one one rib like I can do with an extrusion so I'd loft just the thin piece I want but the loft worked on the 2 curves immediately and didn't let me select the other 2 curves as a destination.

I just got the idea that maybe the problem was the inset curves and so I selected show points on the inset curves and indeed they are formed by a lot of points connecting straight lines. I tried tracing over the inset curves once again using through-point freeform but unfortunately I still get bad results when I export as STL and import into the slicer.

I've attached the 3dm file I'm using which is now littered with experiments I've been running. :) I've drawn an error at the 'main' set of curves so you can distinguish from my various experiments.

Attachments:
Wing1200mm.zip

Image Attachments:
badSTLExport.jpg  booleanFailure.jpg  ribExample.jpg 

---

Hi Supagoat, it's a little hard for me to figure out which one is the boolean problem. Is one you marked with the arrow is the one that you're having trouble with the slicer?

What mesh settings are you using for the STL export?

I don't see anything wrong with the geometry in that arrow-marked one, but the issue may be that your slicer does not like to have very thin areas in it, like this spot here may be thin enough that it is deciding that the polygons there are close enough that it considers that they are touching each other:



Does the slicer have any tolerance values you can adjust?

If it doesn't like such thin sharp shapes you might need to have a pretty small chopped off flat piece at the end rather than a razor sharp point like that.

- Michael

Image Attachments:
supagoat_stl.jpg 

---

It was hard to describe so I made a video demonstrating the problem:
(fixed the URL)

https://youtu.be/nYwWqQHxZKQ

---

OK I think I may have found a workaround. If I trace the outer profile as well as the inset in the form of a C, and then separately make a rectangle to close it and make a D shape, then I loft that curve and rectangle separately and then boolean union the lofts together it appears to work. A little labor intensive but not too bad.

---

Hi Supagoat thanks for the video, I did a closer examination and it looks like there may be an anomaly in the back inner curve here:



There are a couple of spots where the curve is making a sudden tight bend in shape, these spots:



It would probably be good to try and get the original curves to be a little simpler and smoother than this, without this type of sudden changes in shape in highly localized areas. That's not generally good for offsetting.

This is the spot that seems like it may be a problem especially:





You can see there where the curve's control points do an abrupt reversal of direction. I'm not entirely sure but it may be that the curve is back-tracking on itself or very close to back-tracking in that spot. I think that's what may be messing up the boolean.

One way to make this shape without doing a boolean is to do 2 lofts as you did, but with "Cap ends" turned off (unchecked) so the ends are open, and then select, then select all 4 curves and run Construct > Planar, that will build 2 end caps and now you've got 4 surfaces that can be glued together using Edit > Join. Doing a boolean will attempt to intersect surfaces with each other and that backtracking area is probably confusing the boolean.

But even with the non-boolean method that area may generate polygons that are back-tracking as well so it would be good to get the curves smoother and more simplified I'd think.

- Michael

Image Attachments:
supagoat_curves1.jpg  supagoat_curves2.jpg  supagoat_curves3.jpg  supagoat_curves4.jpg 

---

Hi There

I think also that because the inner surface that you boolean out shares the same surface as the outer you sometimes get a conflict.
If you move the inner profiles .1 mm outside of the outers then you get a clean cut as its not trying to cut a surface from a surface on the ends.
Also you get a better result if the inner and outer curves have a similar number of control points, also Offset can throw up some confused curves with way too many control points , I often rebuild them to simplify it.

You can also sweep the inside curves longer than the outer solid to cut with



cheers
shayne

Image Attachments:
2018.04.26-15.28.02-[3D].jpg 

---