Hi OSTexo, Hi Michael,

I run into these type of thing a lot when I'm using fillet.
Most times I try until something works or try something else.
I've been doodling for most of the time now.
Trying to learn the features and tools of MOI .
So now that I'm really starting to make actual models I want know if im doing things wrong or how to fix things.

So thanks for the detailed description on how to salvage the fillet OSTexo.
I will give certainly give it a try.


The problem with the bottom part is clear to me Michael.
Thank you to for the detailed explaination.

The problem with the top part is what I encounter the most when using fillet.
But if I understand correctly there nothing I can do different to prevent this from happening?
There is nothing wrong with the way I modeled the solid or something?
If it does happen, the only thing I can do is to try OSTexo's method of trying to salvage the fillet parts?

If it's a bug that can be fixt in the future.
Maybe it's helpful for you, If I send you a file each time I encounter such a problem?

Regards,

Glenn

Hi Glenn,

> The problem with the top part is what I encounter the most when using fillet.
> But if I understand correctly there nothing I can do different to prevent this from happening?
> There is nothing wrong with the way I modeled the solid or something?

It doesn't look like you've really done anything wrong there, I would have thought that should work.

But you may be running into some general problems related to constructing objects at a somewhat larger scale, you may be able to get some more reliable results if your objects are something like maybe 10 units across in size rather than hundreds of units across.

I've attached a reconstructed version of your object, which I did by scaling it down 1/10 from your original size, then extracting the edges and building new lofted surfaces between them. This reconstructed version does not seem to have any problems with filleting anymore.

If this is something that you're running into frequently you might try just generally drawing and creating things at a little bit smaller size, something like having things at around maybe 1 to 50 units across.

Also from the structure of the surfaces, it looks like you're building this by arranging several 3D curves into place and then constructing surfaces between them. I wouldn't really call that "wrong", but it may be slightly better to try and use 2D extruded surfaces when possible, so like in this case here starting with a block and then doing booleans with 2D curves to slice off chunks of the block rather than doing it all with surface construction between 3D curves.


> If it's a bug that can be fixt in the future.
> Maybe it's helpful for you, If I send you a file each time I encounter such a problem?

It does look like it's a bug. Right now I'm not sure exactly what it's sensitive to, unfortunately it's very difficult and time consuming to do a proper diagnosis. Also it's a particularly easy area to introduce new bugs into when trying to change things to fix a particular case. If you run across cases like this in the future it would be helpful for you to send them to me (please e-mail to moi@moi3d.com), if I get several test cases to work with that does help a lot when I would try to do a fix for it.

- Michael

Hi Glenn, I also tried reconstructing your object at it's existing size by lofting new surfaces, and this new version also seems to behave ok for filleting (attached).

Maybe what you have been running into is a previous bug in Loft where it would sometimes not generate a tight enough fitting tolerance and not make a quite accurate enough result.

That was a bug that was fixed though in the most current v3 beta, are you running the Jun-27 v3 beta version or are you possibly running an earlier version than that?

If you happen to have the original curves that you used to construct that top piece, could you please send those over and tell me which specific method you used for the construction? That might help me to figure out what was different about your original model you posted versus the one I constructed and attached here...

- Michael

Hi Michael,

I will look into the scale thing in the future.

How I constructed the top part:
-I started from rectangle, made a plane and extruded it to a box.
-then I drew my curve in front view , mirrored them, and trimmed it on the surface of the box.
-then I used the extrude tool to push the side pieces away until they disappeared.
( I started out modeling with sketchup, so the new extrude auto Boolean thing kind of makes sense to me = push/pull tool in sketchup )
- next I copied and orated one of my trim curves to line up with my Right view and adjusted it a bit.
- I trimmed this adjusted curve on the solid
- did another push extrude until the part disappeared and had my solid as. It is.

Since had some problems with the fillet tool.
I checked some things out and noticed that the front two curves of my solid was made out of multiple pieces and had to be selected separately ( I think because of pushing the trimmed surface in 2 directions?)
So I joined them, giving me a new curve.
I deleted the front plane, deleted the original (broken) curves and replaced them with the new ones.
At first I did a network on the top front to reconstruct that curved surface, but that didn't help the fillet.
So I went back and did a loft instead.
If I recall correctly, I think I lofted all 3 curved surfaces like that.
At least 2.

I can't send you a file right now,
But I should have a file with the original curves in it, so i can send it to you tomorrow when I'm back at my office.

Hello Glenn,

It just so happens that you're able to get those last two bottom fillets to be a value of 1 and have it end up a nice watertight solid. The key is to extend the bottommost box along the long axis so you can Fillet the lines you indicated with a value of 1. Then just create the bottom box again and select a Fillet operation of 1 on the lines you indicated. Remove the internally created faces and Join the two surfaces together into a Solid. Those fillets along the top edge between the two major parts of your model will probably require a bit more thinking to blend in nicely.