Hi flippyfloop, I took a close look at the geometry and used the GCon command in Rhino and 2 areas of the projected trimming curve are not quite fully smooth where they touch each other. They are off from tangent only by a small amount like 2.26 degrees but I think that is enough to be what is messing up the filleting. Filleting doens't like to have things that are close to being smooth but only off by a small amount like that. When they're not quite tangent it's going to try to extend the fillets and intersect them against each other and with a very shallow intersection that can be difficult.

Probably a good way to ensure accuracy would be to make your cutting curves as sharp pointed lines and use filleting to round those off too.

The steps that I'd use would be to go ahead and shell the cylinder to start and get 2 polylines set up like this (make sure both sides of them are on the same level):



Do a boolean difference with one of the polylines to get this:



Select that piece and use Ctrl+C to copy it to the clipboard. Now do an undo to get your full cylinder back. Then do a boolean difference with the other one to get this:



Now paste in the other part so you've got 2 pieces right on top of each other like this:



Now those can be combined with boolean union. Often times it is not so great to do booleans with pieces that have a lot of overlapping surface areas because it can be difficult for barely grazing pieces to be cleanly intersected. However an exception is if the pieces all come from the same original object and so are very exactly matching and all have the same underlying surfaces. Boolean union in that particular case will do a good job of combining things and unifying areas that share the same original surfaces. (note that rotating the first piece would not count as having the exact same underlying surfaces, you've got to apply cuts to the same original piece and not transform pieces for this technique). Then after that select these short edges:



And fillet those with a large fillet radius to get this:



Now select all these edges with one window select (it's ok to select already filleted edges for another fillet they are just skipped):



And then do your smaller radius fillet:



The nice thing about this method is no manual projection involved, just booleans and fillets.

- Michael

Thanks Michael. That works great.
Cheers.

Tried Phiro's method. Didn't know about the Rebuild command so that was great.

Issue I had was that when lofting big circle to bottom curve, for some reason I wasn't getting the Close checkbox. Usually it is there but for whatever reason it didn't appear ?
Tried on a simple two circle loft and the close checkbox is there again, weird ?

Hi flippyfloop,

re:
> for some reason I wasn't getting the Close checkbox. Usually it is
> there but for whatever reason it didn't appear ?

It will only show up if you do a loft with 3 or more profiles. A loft between just 2 profile curves won't have the option for making a closed surface.

I think your original problem came about from something like gluing together 2 projected curves at some spot other than the natural endpoints, like at a trimmed spot or at a spot where the projection ran off the silhouette. Those spots will be accurate enough to be joined but the curve end tangents at such areas may be slightly off because projection curves are the result of a fitting process. You would need to true up the curve's end tangents in such cases by editing the end and first interior points to ensure they are totally smooth to each other and that would make for better filleting.

- Michael