First off let me say that I really like MoI. I've almost never used a spline modeling tool before, and in just a few hours I am happily modeling away:-)

My only problem at the moment is that I find it hard to predict if the filet tool will work at all. For example, I tried to create a simple shape with a button sitting flush with it. Here's my workflow:

Create a spline and extrude to the shape I want.
Create a curve and trim the shape of the button. (In order to keep surfaces for both button and shape.)
Select trim curve and extrude to use as sides hole in which the button goes.
Select extruded sides and join with trimed surface in order to create a filetable "corner".
Now when I select the "corner" curve and try to filet it, it will usually fail no matter what radius I set the filet to.

Is this a weakness of the filet tool, or is there something substantially wrong with my modeling workflow?

It would be nice if there were some error messages in the fillet tool that basically told me, I'm doing something wrong, like "radius is too large" or need to pick at least two perpendicular planes or something. It's hard to know what foolish thing I did that is resulting in no fillets.

> Is this a weakness of the filet tool, or is there something
> substantially wrong with my modeling workflow?

Can you please post an example model? Nothing that you're describing jumps out as being bad, but if I can examine your model it would help to determine what's going on.

- Michael

> It would be nice if there were some error messages

Yup, I would like to do this eventually.

There are other types of operations that could benefit from this type of error feedback as well.

I just haven't quite figured out a place to put them in the UI yet, it might be a while before the design of where those messages go in the UI comes together.

- Michael

Hi, I'm sorry for not following up on this earlier, I've been away for the week-end.

Now I've not been able to reproduce the problems from last time (although I've used the exact same methodology and measurements)
Anyway, here's the model I was trying to make, a (rather sloppy) reproduction of the front-panel of my mobile(cell-) phone.

Thanks for following up. Hmm, well I guess that's good that it is working now anyway.

If you happen to run into it again, please post the problem model.

My best guess is that there was some problem in the join process in the case where you ran into a problem - filleting won't work well if something went wrong with the join and edges that you thought were fused together really weren't attached to each other. A situation like that could fail to fillet kind of like you described.

- Michael

Hi, it's me struggling with some fillet problems again!

:-)

I think I've found a sure-fire way for filet to fail, and I don't think it's a user fault.
1: draw an open curve
2: extrude the curve
3: trim the corners of the resulting object
4: create a solid object by using Shell
5: try filleting the edges of the resulting solid object.

Another filet problem:
I made a closed curve with some corners, duplicated a few times and did a loft with straight edges. I couldn't then fillet most edges on the resulting object. and when I tried filleting just some of the circumfering edges this happened:

In your first example, there's no continuity on the curves chosen for fillet. You can see that by selecting the curves you want to fillet (the ones in yellow in your example image) and copy them to the clipboard. Then hide everything in your scene,
paste your curves back into the scene, select them all and press join.
The result will not be one single curve.
From there the problem in your filleting I think...

Pilou, yes I got filet to work on the first level too, but in a somewhat surprising way, by increasing fillet radius. (And by selecting just two of the circumfering edges I got that weird bug with the fillet sticking out from the object)

You're propably right Schbeurd. I don't think there should be a reason for the edges to be discontinous, though, as everything is modeled with snap to grid. Also I can't find any micro edges or surfaces that mess it up.

Something appears to have gone wrong with the trim, here's a zoom-in on one corner.

Yes, I just verified it - you didn't model anything wrong, trimming those surfaces by the curves has a bug where it doesn't do a clean cut right near the end where there is tangency.

It looks like the mechanism that projects curves on to surfaces doesn't deal with grazing tangent conditions that well (which tends to be a quite difficult area of calculation). However, there is a proper handling of these things in actual surface-surface intersection rather than curve projection. So as a workaround, I extruded your trim curves out (using both sides option) and used the extruded surface as the cutting object. this then generated a good cut. If this seems to be a recurring problem area with projected curve trimming, I may need to internally switch to this to an extrude and surface-intersection method instead of a projection method.

Anyway, I've attached a model using the surface trimming workaround, it behaves well now.

- Michael

Looks like you're right.
Testing a few times now it looks like the workflow I described will always produce a bad trim on the first try. *However* Hitting CTRL-Z and then redoing the trim always seems to fix the problem. I've done this 4-5 times with different files now.

1: turn on snap to grid.
2: draw a dome-like curve
3: extrude it
4: create an arc for trimming the corners (Like in the file I supplied earlier)
5: do a trim of the extruded object with the ark

You now have an object with a malformed trim at the corner(s)

6: Hit CTRL-Z
7: Redo trim (Not CTRL-Y)

You now have an object with a good trim at the corner(s)

(Edit: looks like we both found a workaround at the same time:-)

> You now have an object with a good trim at the corner(s)

Hmmm, I was able to repeat this over here too. That is quite odd.

Maybe it is a clue that some setting on the object (such as a tolerance value for an edge) is not set properly and gets fixed somehow when the object is restored through the undo process....

- Michael