"Body would intersect itself" when sweeping, but only with spline path!

"Body would intersect itself" when sweeping, but only with spline path!

jandyman
Advocate Advocate
782 Views
6 Replies
Message 1 of 7

"Body would intersect itself" when sweeping, but only with spline path!

jandyman
Advocate
Advocate

I've been fighting with Fusion 360 trying to get it to do parametric shaping the way I want it. I've run into problem with variable radius fillets - I hear it might be a bug. I've tried complex lofts, but they require a lot of setup work and also they run into the problem I'm describing here with sweeps. And the problem seems to be limited to using splines as the path, which is what I want to do. Here's a video, and I'm including the model as well:

I've been having a lot of trouble with the shaping I want to do, but the example that caused me to dig in and troubleshoot was trying to achieve a result like like this:






jandyman_0-1725134672843.jpeg

 

0 Likes
Accepted solutions (1)
783 Views
6 Replies
Replies (6)
Message 2 of 7

TrippyLighting
Consultant
Consultant

If you pull the arrow back to the start and then slowly forward, you can see when the sweep starts failing.

Hopefully, you can see why this will create self-intersecting geometry. 

 

TrippyLighting_0-1725135740638.png

 


EESignature

Message 3 of 7

TrippyLighting
Consultant
Consultant

Here is one approach to how to solve this. I'd be happy to show this in a screencast.

You can see on the timeline that this is a bit more complicated:

 

TrippyLighting_0-1725137123153.png

 


EESignature

0 Likes
Message 4 of 7

jandyman
Advocate
Advocate

@TrippyLighting I'm always happy to learn more techniques, so sure a screencast sounds good. But one of the points I was making was that treatment of spline sweeps seems different than linked lines and arcs. It sure looks to me like the sweep on the other side of my example model - which consists of straight lines linked by small arcs - also "intersects itself". Am I wrong about this? - if so then I don't understand the definition of "intersects itself". (BTW I do like that blue indication of what is happening though - I don't see that, how do you turn that on?).

 

Another way of saying this is "so what if the sweep intersects itself, it does when not using a spline and everything seems to work". Of course maybe I'm not understanding something.

BTW I've noticed further experimentation that some of the problems I've been having with variable radius fillets also seem to go away when I replace spline paths with linked arcs and lines. I'm guessing I'd find similar behavior with lofts.

0 Likes
Message 5 of 7

TrippyLighting
Consultant
Consultant

@jandyman wrote:

... I'm always happy to learn more techniques ...


Before I show techniques, I feel the need to explain concepts:

 

First, CAD software uses two main sets of mathematical descriptions to represent geometry.

  1. Analytical geometry: spheres, tori, planes, boxes, prisms, pyramids, etc., and sections of such geometry.
  2. NURBS for arbitrarily curved surfaces.

In the case of a solid body, those are stitched into a BRep.

 

Analytical geometry is mathematically much simpler than NURBS! That includes calculating self-intersections. 

 

Filleting lines and arcs with circular or constant radius fillets creates analytical geometry.  In essence, cylinders or sections of tori. A variable radius fillet with 2 different radii along a straight edge can still be represented as a section

of a cone, so "simple" analytical geometry.

However, once you add a third radius somewhere in between, the resulting surface will be represented by NURBS.  

 

When you use Splines (NURBS) to create the outline of your instrument, and then fillet those, that creates a very complex surface. Sometimes, the geometric modeling kernel of Fusion can deal with self-intersections of such surfaces, but more often than not, it cannot. 

 

Ultimately, to create the "sculpted" or "organic" or "freeform" shapes of your instruments,  you will need to move so surface modeling and , yes, that includes lofts 😉

 

 

 


EESignature

0 Likes
Message 6 of 7

jandyman
Advocate
Advocate

@TrippyLighting This is a HUGELY useful reply, and though I didn't know the specifics that you spoke about, I suspected something of the sort. I'm a nerdy engineer guy, so if you have any references to get some background on what you explained, I would love to read it.

 

One comment you made near the end is about using surfaces. I have done some of that in the past, but moved away from it because there are so many more steps with all the patching stitching and so forth. But - are you saying that if I use some of the same techniques (e.g. lofting and sweeping) but in the surface domain, that the math is easer for the kernel and I'm less likely to run into these issues?

0 Likes
Message 7 of 7

TrippyLighting
Consultant
Consultant
Accepted solution

@jandyman wrote:

...But - are you saying that if I use some of the same techniques (e.g. lofting and sweeping) but in the surface domain, that the math is easer for the kernel and I'm less likely to run into these issues?


No, what I am saying is that some geometry just cannot be created with solid lofts.

Indeed, working with surfaces often requires more steps. That is the price you pay for creating more sophisticated models 😉


EESignature

0 Likes