Very bad fillet function

Next
 From:  cx (CX138)
4830.1 
Very bad fillet function (v1 v2 v3beta are the same)

This is a very simple model (a rotary switch), using the radius value is very reasonable!

Please explain why not generate the fillet?

Glad to see v3.0 beta version of new features, but some old problems have not been resolved! Please improve the use of these basic functions!

I do a round, or even make the model output to rhino3d deal with ...!

Very hard! ! !



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:  BurrMan
4830.2 In reply to 4830.1 
Any one of the continuity fillets work well.
  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:  BurrMan
4830.3 In reply to 4830.1 
seems to be something with the seam of the circles extrude being included in that cutout.. If it is not included, then a good circular fillet goes through also.

So for now, you can rotate that seam edge out of the area and get the fillet.

EDITED: 19 Jun 2012 by BURRMAN

  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
4830.4 In reply to 4830.1 
Hi cx - that definitely looks like a fillet bug, thanks for reporting it!

> Please explain why not generate the fillet?

You can actually see in your own screenshot why the fillet is not successful - it's because the fillet engine did not generate a long enough fillet surface, there is a gap between the end of the fillet here:



There's definitely some bug in the fillet surface generation step where it's not making that fillet long enough.

That's a somewhat unusual bug, I would have normally expected a pretty simple situation like this to work. Unfortunately the nature of geometry is that it comes in such infinite varieties that it's easy for there to be bugs that involve just some particular arrangements of shapes.

I'll put this on my list to investigate, but often times fillet bugs are quite difficult for me to fix up, the filleting in MoI comes from a library that I license called Solids++ and I'm mostly dependent on them to fix bugs in their library to improve filleting but sometimes I'm able to fix them up myself as well if I can figure out what exactly is going on. I won't be sure until I get a chance to look into it in more detail whether this one is something that I could have a chance at fixing up myself or whether it would need to be fixed by them.

In the meantime the easiest workaround is probably to do the fillet on the full cylinder before you cut the notch in it.

Thanks,
- Michael
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
4830.5 In reply to 4830.1 
Hi cx, in case you don't have an earlier version of the model available, I separated out the parts in the attached 3DM file into their pre-boolean state.

If you fillet the edge of the cylinder first like this:



And then do the boolean cut for the making the notch afterwards like this:



Then that should give the result you want:



There's definitely a bug though that was preventing it from working correctly doing the fillet after the boolean in this particular case.

- 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
4830.6 In reply to 4830.1 
The bug seems to have something to do with that particular situation though - other cases even simply repeating that same notch around seem to work ok, like this for example (3DM also attached):





Also if the notch is just a little bit wider it also seems to work ok, like this:





So probably the bug is something like it thinks that the fillet is nearly making a full 360 degree turn and it doesn't try to extend the ends of it in your particular original example.

- 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
4830.7 In reply to 4830.1 
Another possible workaround for this one particular fillet extension bug is to cut the edge up into 2 pieces (use Edit > Trim and the "Add trim points" button to cut the edge), that will then also avoid the bug that is happening in this particular situation, see attached model.

- Michael
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:  cx (CX138)
4830.8 In reply to 4830.3 
My previous workflow is this:
Draw curve -> Extrusion -> Boolean operations -> change of details.

Of course, I can change the workflow to avoid this error:
General shape obtained by rotating the curve -> do Boolean operations.

I know any one of the continuity fillets work well, but continuity fillets is not an exact arc, In many cases, G1 continuity and Circular continuity look very similar, but they generated polygon number is different. Moreover, this is not an advanced CAD software design intent.

This simple model can do, but complex models? Often difficult to change the design workflow。

Thanks BurrMan
  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:  cx (CX138)
4830.9 In reply to 4830.7 
  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
4830.10 In reply to 4830.8 
Hi cx,

> This simple model can do, but complex models? Often difficult
> to change the design workflow。

The bug in this case seems to be when you have one long edge that then has only one relatively small piece taken out of it.

If you had a more complex model it would actually have avoided running into this particular bug.

See my previous example where a model with numerous notches taken out works fine without doing any other workaround:
http://moi3d.com/forum/index.php?webtag=MOI&msg=4830.6

You happened to run into a bug that was specific to the particular configuration of that one model... That's going to happen from time to time.

I've added this one to my list to investigate to see if I can fix it up - it helps a lot having a simple and well constructed example like you originally posted.

For the most part this kind of simple case filleting tends to work well in MoI, I was pretty surprised that this one didn't.

Filleting is a quite complex process, it only takes one small bug in any one of a sequence of different operations to mess it up.

- 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:  BurrMan
4830.11 In reply to 4830.8 
Your welcome cx,
"""""This simple model can do, but complex models? Often difficult to change the design workflow"""""""

One simple rule that helps avoid these things is to keep seam edges out of your operations whenever possible... It's good to learn to pay attention to those as you create stuff. It also plays a huge factor in the polygon structure when you dice it up... A user here "PAQ" is very good at this! FYI
  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