Hi, Michael! I've been mucking about a bit with MoI again, and I like the progress a lot:-) I think I've come across a bug/weakness with the offset tool though. I'm trying to make a pot for a perculator. It's made with two sweeps for the top of the pot, and a revolved profile for the "body". Each piece gets offset/shelled nicely when separate, but when they're joined I get this:

Your model works fine for me! No problem when I join the different pieces ???
Are you sur to have the last vesion of September?

PS A real percolator has a thickness :)

Uhm... It's the thickness that's the problem, I'm sorry if I was unclear in my initial post.

Join the 3 pieces together
Select Offset->Shell->Thickness 0,1

Object goes wonky, like in the image.

I have the 4th Sept Beta.

It must be a miscalculation of the surfaces when MOI joins them. Anyway, if you split the top and the bottom section, shell them separately, delete the common surfaces and then join, it works.
Hope it helps.
Daniele

Hi Linker, thanks for reporting this.

It seems to be related to a really tiny micro-fold in one area of the upper lip:







You can see there that 2 of the surface's control points are kind of crossed in position with one another. That creates a really tiny folded area of the surface right there which really tends to cause problems in shelling or offset operations.

Was that made with a sweep? If you still have the original curves could you please post them? If that little jog back was not in your original curves it may be something that needs tuning up in sweep.


It kind of looks like that general area around the micro-fold is kind of a little bit more lumpy and bumpy than it looks at first glance. Shelling in particular is pretty sensitive to that type of stuff because the ripples get magnified after offsetting, and then shell tries to take that offset result and intersect it with neighboring offsets.

That kind of intersection where things are kind of wavy and barely grazing through each other is difficult to calculate.

- Michael

Hi.

I redid the basic setup I used, and was able to reproduce the error.
I basically made 4 profiles and swept those along two rails. (See attached file)

Now I've tried using several methods to avoid the bug (Exact/Auto/Refit...) but they all produce some level of bugginess. I thought I had it sorted after trying Network instead of Sweep and it looked like a much cleaner mesh with show points, but still it produced errors on Shell.

that works again :)

When you've done the sweep, unhide the body of the pot (It's in the file), mirror the top and join the 3 elements. Now try giving thickness. Doesn't work, ergo there's something wrong with the object resulting from the sweep:-) (And as mentioned earlier in the thread, you can see the bug if you study the swept object closely with show points activated.)

Hi Linker, yeah unfortunately Shell is pretty much the least reliable function in MoI in general... But it also has a tough job to do.

In this case it looks like there is a difficulty in processing the juncture between your top sweep lip and the bottom part.

Even though you have carefully drawn those profile curves so that they are tangent pointing straight downward, the actual surface that is generated there does not have that same direction through its entire bottom edge.

That's because of the different lengths of and shapes of the rails - as those profiles slide along those rails they slide at slightly different speeds along each one, this causes a kind of twisting or torquing of the profile as it slides along, which affects the tangent direction of the surface.

So when the sheller tries to process that spot, it sees it as a sharp edge, not as a tangent edge. That means it tries to extend each surface, then offset them, then tries to intersect those. It's having problems calculating the intersection there, it can be difficult to calculate clean intersections when things are intersecting one another at a very shallow angle.

This can be a problem when things are nearly tangent but just not quite.

One method that I used that worked was to do one big sweep around the whole circumference, and then trim off a small bit off the top of the bottom part and then put in a blend between the sweep and the bottom - that then shelled well because the pieces were all tangent.

If you separate stuff into individual surfaces and then offset or shell those, that is working since there isn't any extension and intersection between extended surfaces happening in those cases.

I'll see if I can think of any other tips. In general if you can kind of weld your pieces together using either blends or fillets it can help to ensure tangency and make shelling work better...

- Michael