Filleting

Next
 From:  wibble
232.1 
Hullo again!,

When creating a fillet, I've noticed that there are some limitations. Initially, it was all a mystery to me, sometimes I could create a fillet, sometimes I couldn't. But as time goes on I'm slowly starting to see what's happening.

Example:
Filleting the intersection between two spheres that have been unioned together reveals the problem everytime. Where the pole of a sphere lies can determine whether a fillet is succcessful or not. In the attached picture you'll see that I've filleted the intersection and if I was to try to increase the fillet to reach beyond the pole of the smaller sphere, then the fillet would fail. The fillet can only go as far as the pole. Yet I can get around the problem by rotating the smaller sphere 90 degrees in any direction.

Is this something that you're aware of Michael?. Is it a problem that you can overcome on your end, or is it a problem for the guys who made the libraries you're using? Is there a simple rule that we can follow to learn how not to get into awkward situations like the one above?
Attachments:

  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
232.2 In reply to 232.1 
> Is this something that you're aware of Michael?

Yeah, I've run across it myself as well. It's not just the pole, a fillet won't go across that whole edge there that starts at the pole.


> Is it a problem that you can overcome on your end, or is it a problem for
> the guys who made the libraries you're using?

The second one, I'm afraid. I think that there is some stuff built in that is supposed to handle this situation with a fillet continuing across an edge like this, but it seems to not be working properly right now.


> Is there a simple rule that we can follow to learn how not to get into
> awkward situations like the one above?

Well the fewer edges that you have nearby the fillet makes the fillter's job easier. But sometimes this can be hard to avoid.

- Michael
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  wibble
232.3 In reply to 232.2 
I've noticed that filleting is a problem in all the 3D modelling programs I've tried. Although I've not really tried any of those big CAD applications like SoldWorks. But those things look far too cumbersome for what I want to do.

Despite MoI having some filleting problems, it can still manage to fillet things more reliably and easily than all the other 3D modellers I've tried. Then there's the fact that MoI feels so damn nice to use.
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
232.4 In reply to 232.3 
Yeah, it's a very complex problem.

Those "Big CAD" programs (ones based on the ACIS or Parasolid geometry kernels) are better at doing it, they've got quite a lot of people working on it and have invested a lot of effort handling a large number of special cases. But there can still be fillet problems even in there too.

I do expect that the filleting in MoI will improve over time, but it will take a while. But I think that this particular problem here probably won't be too bad to get fixed.

- Michael
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  tyglik
232.5 
Hi Michael,

I think that it's inappropriate to keep the radius distance between each calling the Fillet command (and maybe Chamfer, Offset, Shell... too).
For instance, I select some solid and fillet its edges with r = 2. Then I select something else and want to fillet with r = 0.5. ...whoops... MoI uselessly performs a computation (or even freezes with complex shapes) for last input, i.e. r = 2, before I'm able to change the radius to r = 0.5. (Note that the exactly value of r is not critical, it is only for the illustration).
I suggest,
(1) not performing a computation until the 'done' button is clicked or
(2) adding a 'preview' button or
(3) not retaining last input or
(4) ...



Thanks
Petr
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Frenchy Pilou (PILOU)
232.6 
For the critical filets, is not more tricky to draw by "hand" the profil and sweep it along curves wished? :)
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
232.7 In reply to 232.5 
Hi Petr, yes this is a problem.

The way I have been planning to solve it is with:

4) Perform the fillet calculation in the background and automatically cancel it and start a new calculation if a new radius has been entered.

Which has the only problem that it is quite difficult to achieve... :) However, I have been planning on this since there really needs to be a way to cancel long operations in general anyway even separate from this particular issue.

I've tried really hard not to have things like "preview" buttons, I really prefer to just do the preview automatically.

- Michael
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
232.8 In reply to 232.6 
> For the critical filets, is not more tricky to draw by "hand" the
> profil and sweep it along curves wished? :)

Hi Pilou, it is certainly possible to generate some fillets in this way, but you'll also have quite a bit of work to do to fill in corners and handle intersections between the different fillets.

But that's actually one of the nice things about a "hybrid" modeling system like MoI (where you can work on solids or also on individual surfaces and join them together later) - that you can get in there and construct a fillet manually in some cases where an automatic solver isn't able to handle it.

It is not exactly easy to do though.

- Michael
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
 From:  Frenchy Pilou (PILOU)
232.9 In reply to 232.8 
Sure and perfection is not for this world :D

EDITED: 4 Dec 2006 by PILOU

  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged
 

Reply to All Reply to All