Hi Lemo, thanks for posting the model that helps a lot.

Yeah normally you will want to join pieces together to fillet their common edge. It is possible to do fillets without joining but only if you have separated the objects into completely unjoined individual surfaces. Then when you pick 2 surfaces you can get a slightly different surface/surface type fillet calculation which can sometimes work in situations where the edge fillet will fail. But the joined-edge-based filleting tends to be more convenient so that's normally where you start.

For your new model on the top ridge edge, there are 2 things going on that cause fillet difficulty - one is when surfaces collapse down to a point, the filleter can get confused about that. Also when the surfaces don't meet each other in a sharp crease, that can also cause difficulty - here you've also got that happening since at the tip the 2 surfaces flatten down and become tangent with one another.

The top edge will fillet if you hack off about this much off the tip (here I drew a line and use Boolean / Difference to cut it:



Then you would have to fill in the tip part by doing some manual trims and maybe a sweep of the end fillet piece down to the tip. Let me know if you would like some more detailed steps on that part.

For the indented piece - that's again got the surfaces collapsing down to a point.

If you can model those as larger surfaces that are trimmed down, instead of ones where the entire surface collapses down to a single point at the tip, it will have a better chance of filleting.

Like for instance here I have made some cutting surfaces that are extended to be bigger than just the exact cut area:



If the indent is created by trimming these surfaces with one another (select them all and run Trim and push "Done" at the select cutting objects prompt to indicate they will cut each other), and the pieces joined together, then that result should have better chance at filleting.

Another difficulty will be the sharp corner - trying to fillet such a sharp corner will probably result in pretty messy blended corner areas that are trying to stretch too far around. To put less stress on that corner you need to make it rounded first before filleting. The easiest way to do that is to first fillet the inner edge with a larger radius:



Now the outside piece is all rounded and you can fillet the outer edges with a slightly smaller radius than the one used on the inside edge:



It should be possible for the filleter to get tuned up in the future to handle collapsed-down surfaces better, but for V1 it is pretty sensitive to that.

Hope this helps show you some work-arounds, let me know if you need more details about any part.

- Michael

Fantastic advice! I only end up with the 'collapsing to a point' challenging geometry because I generated the sharp carved out part with an extrusion, bool, and then subsequent addition of a curve (valley center) and surface generation from there. That leads to 'sharp' edges. It did not occur to me that I can use another surface to trim the existing surface, thus creating already the valley profile I actually want..... From there, controlling the 'rim' of the valley is much simpler and straight forward....
Thanks a lot for your comments!
Lemo

Nurbs are tricky things isn't it? :)

You're welcome Lemo!

There can be many situations where it can be easier (and produce more reliable geometry), to kind of imagine a larger extended surface that gets trimmed back to make your final shape, rather than trying to create a surface only by following outline curves.

If your outline curves have a very natural 4-sided type layout, then surfacing them directly is fine, but otherwise building extended and trimmed back can be better.

Here's another example from a recent thread: http://moi3d.com/forum/index.php?webtag=MOI&msg=1127.4

In that case it helped to build a single larger slightly curved surface as the cap of the object which then gets trimmed, rather than trying to sweep using the existing outline curves which came to a kind of difficult corner which caused a sort of bunching and overlapping of the surface when trying to sweep.

- Michael