Hello,

I have been running into a weird filleting problem. In the top pic I am trying to fillet an edge. for some reason it ends up mangling the end (middle pic). This happens more often than not, and on different edges. Sometimes changing the shape in the fillet properties helps, but not always. How can I get filleting to work correctly? Also, as seen in the bottom pic is the browser window stacks beside the toolbar. Is there a way to have it stack beneath the toolbar inline with the rest of the tools to save screen real estate? Thanks.

Hi OSTexo,

As for the fillet problem, it's hard to solve issues from screen dumps. if you could post the 3dm file or even part of the model that would help the forum troubleshoot the problem.
What I can make out from the picture, it looks like filleting is having a problem because of that small surface joined at the end and the fillet is intersecting that join, maybe if you get rid of that small surface at the end by reconstructing the main curved cylinder to be one surface it might do the trick, but without the model I don't know for sure.



> Also, as seen in the bottom pic is the browser
> window stacks beside the toolbar. Is there a way
> to have it stack beneath the toolbar inline with
> the rest of the tools to save screen real estate? Thanks.

There are a three options now in the June 23 beta, go to, options> general> scene browser position: and in the drop down you have the options; Adjacent, Inside or Opposite.

Cheers

Hello,

I have attached the sample. It also seems like it happens when the edge is broken up into multiple curves. I don't remember doing this, but as you can see the end it actually broken into 3 edges. I'm not sure why that happens. I also have this problem when I difference or union certain objects, the end edge not being contiguous. Thanks.

It fillets fine for me. as Danny said there are 3 edges there. THe other thing it looks like to me is (And something I like) is the filleter doesnt just "fail" anymore with larger values. It actually "DOES" the fillet under the surface.(I suppose as it becomes more sofisticated it will get that trim right also) This happens alot when filleting an egde "past aother edge". Something I look forward to.

Anyway, a good practice to make it work, is to start with the samller number and work up. For instance with your model, if I start with 1.5 inch fillet, I get the funcky fillet. Cant get it back unless I quit fillet and start over. Start with .5 and see what happens.

Hi OSTexo, is there a particular fillet radius value you are trying to target?

I loaded your file and for the most part I seemed to be able to get fillets on it, for example here is one with radius 0.2:



But one of the big problems here is that your surfaces are significantly more complex than optimal, with a lot of poles and irregularities in them.

For example your surface here that looks like a simple extrusion:



Is actually a complex surface that collapses together with poles in these areas:



That kind of squished together structure at those pole areas is bad, it is likely that the surface actually goes back and forth over itself in that little area, that is a bad thing for fillets.

Similarly this surface that looks like it may be a part of a sphere or something:



Is actually a kind of asymmetrical warped surface:



That's actually got a pole and actually even worse a kind of "anti-pole" where 2 edges of the underlying surface are tangent where they touch one another rather than being a distinct quad surface patch.

That's a lot of twisting and compression in little zones in each of those surfaces, that is going to very negatively affect additional calculations that try to trace along those surfaces like fillets do.

You'll get a lot better results with cleaner and more simple structured surfaces, you should try to get more things like trimmed extrusions for pieces that are straight in one direction, rather than trying to do it by something like a loft between 2 halves that touch each other at a point.

If your object is a solid, the best way to get a clean structure is to do the initial cut with a curve from the side view, which will automatically produce an extrusion for you as the capped off surface. It will give you a very much cleaner result and will actually be faster to model as well.

Check out this previous post for an explanation and some illustrated steps:
http://moi3d.com/forum/index.php?webtag=MOI&msg=2614.10

Also some of the previous discussion there may be worth checking out, since it pertains to the same kind of technique that you are using here that is giving you bad results later on:
http://moi3d.com/forum/index.php?webtag=MOI&msg=2614.7

Hope that helps,

- Michael

Hi OSTexo - just to clarify a little bit, it's not necessarily that you have to avoid poles everywhere, if they have a clean and more distinct structure where the things come together at the pole area, it can be fine. Sometimes you even have to have poles like for example a regular sphere object has poles in it.

So like the ones on this area of your model should be fine:



That's got good symmetry, not a chaotic clustering of points in the pole area, etc...

When something is becomes more sort of stretched flat though, like in the bottom cap piece I showed those screen shots of earlier, then the surface tends to be much more jumbled right in the pole areas which is what is not good.

So generally with flat or straight things you want those to be formed by a plane or extrusion that is trimmed rather than having something that is going to have a bad "bunched together" kind of effect.

Filleting in particular will tend to work a lot better when flat things are simple planes or extrusions that are trimmed, rather than a surface that has the natural surface edges directly trying to hug irregular boundaries.

The natural structure of a NURBS surface is as a quad. That quad can be deformed and stretched, and pinched to make poles, but if you can produce the outer boundary by having a more simple surface that is then trimmed, that tends to be more desirable.

- Michael

Hello,

Thanks for the explanation. I do have a problem though when I upgraded to the latest build. I am unable to turn on show points. Thoughts? Thanks.

Hello,

After testing a little more it seems the biggest problem for me is to situate the seam of the object correctly. For instance when I Boolean diff a box with a tube and the seam is of the tube is cutting the box at some point it fragments the edge of the box and makes all sorts of problems for filleting that edge. Am I missing something obvious to prevent this behavior? Thanks.

Hi OSTexo,

> Thanks for the explanation. I do have a problem though
> when I upgraded to the latest build. I am unable to turn
> on show points. Thoughts? Thanks.

You can't turn on points for objects that are made up of more than one joined surface that are joined at trim edges, which will usually be what you get as the result of a boolean.

But you can use Edit/Separate to break your joined object into individual separate surfaces, and then you should always be able to turn on control points for an individual surface that is not joined to another one.

That's what I did to turn on control points in those screenshots I posted previously.

Please check out this for some better explanation and illustration of why it doesn't turn on points for joined trim edges:
http://moi3d.com/faq#Q:_Why_does_show_points_work_for_some_objects_but_not_others.3F

- Michael

Hi OSTexo,

> Am I missing something obvious to prevent this behavior? Thanks.

Well that's probably the single biggest problem currently with MoI's filleting mechanism, it won't sort of cross over past an edge like that so a small fragment of a seam can limit how far you can do an edge fillet.

It's unfortunately not an easy issue to me to fix, at least in a short timeframe.

If you notice that your seam on a tube is not good, you can rotate your generator profile curve by 90 or 180 degrees and then reconstruct your surface, the seam on something like a tube will be inherited basically from the seam of the initial closed curve.


One other thing that you can do is to separate things into individual surfaces, and then do a surface/surface fillet which is when you have 2 individual faces selected. That does a different style of fillet calculation which does not try to trace along the edge structure like the edge-based fillet.

So for example in your shape there if I separate out things into individual surfaces, and then select these 2 individual non-joined surfaces:



Then I can perform a surface/surface fillet that can go a further distance and is not limited by that seam edge or other surface fragment:



That can help to construct a fillet that the edge filleter cannot do, but the edge one tends to be more convenient since it can build corner patches where multiple edges collide, the surface/surface filleter does not do corners and may need to have some manual trimming done after the fillets are generated (but not when it forms a simple closed shape like here).

But before doing that surface/surface fillet, I also needed to do one cleanup set on your tube part as well - the tube fragment here:



Is part of a larger surface that has some bunched up areas that are places where the surface kind of folds over itself in a self-intersecting manner:




That kind of self intersection is pretty bad for a lot of operations, it will often time mess up booleans as well as fillets. It's the same issue discussed recently here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=2728.2


In this case, the active area of the surface is ok though, so applying a "ShrinkTrimmedSrf" (which you have to either type in or set up a keyboard shortcut to activate) will solve it by shrinking the underlying surface to the trim boundaries and actually removing those bad bunched up self-intersecting areas from the surface.


Unfortunately fillet as you've seen is a very sensitive function, it performs quite a lot of intensive calculations, including stuff like creating offset surfaces, calculating surface/surface intersections, making extensions and trims. It doesn't take that much irregularity in the geometry to mess up any one of these steps.

- Michael

Hello,

Thank you for such a detailed explanation. I am gathering that as I learn more about MoI I will begin to plan objects well from the beginning to avoid problems down the road. I am also coming to the conclusion that sometimes it is better to stop trying to tweak something to make it work and just create the same geometry a different way.

Hi OSTexo,

> I am gathering that as I learn more about MoI I will
> begin to plan objects well from the beginning to avoid
> problems down the road.

Yes, that tends to be particularly true when filleting is involved.

- Michael