Somehow the surfaces are inside out. This makes the Boolean work wrong way round. Difference acts as Union and Union acts as difference. There is a Flip command but I can't make it work.

Hi Yenmonger, it looks like it is doing a boolean union instead of a difference - that usually means there is some problem with the surface normal orientation of one of the objects.

If you select them both and actually run a boolean union instead, it will actually do the difference then.

The problem one is the green object, and it looks like there may be some self-intersecting surfaces in that one which can tend to mess up intersection calculations.

Notice here at the tip where things take a kind of sudden swoop inwards:





So I think that side wall piece is actually folding back over top of itself right in that tip area - those kinds of things will tend to mess up surface intersection calculations. In this case it probably messes up the ray intersection tests that are done to determine which side of an object is the outside portion.

If you delete that one side wall face that folds over itself with a trimmed planar face instead, then it should be better, I've attached a version that is set up like that. This version should now behave normally with boolean difference.

- Michael

Hi Steve,

> There is a Flip command but I can't make it work.

The flip command doesn't do anything on a closed solid, because there is a mechanism that performs intersection tests on solids to determine the outside.

But that mechanism can get confused if there are self-intersecting surfaces in the model, which there was in this case - the side wall portion of the green piece is a surface with a bunch of control points in it and it folds back over top of itself near the tip.

If that area is repaired to not have a self intersecting surface (see attachment above), then things will work properly with it.

- Michael

Heya guys, thanks for the investigation!

I chose to cut this Gordian knot by remaking the base ship using a looser loft instead of normal loft. Evidently that must have cured the wacky fold that you pointed out.

Hey Yenmonger,
I looked at making this shape a little better with a "rail revolve".. Here is the result.. The revolve curve I had adjusted the 2 endpoints to be tangent, as opposed to the little wiggle at the end. I also made all curves planar.

Thank you, sir.