Hi niko, also, re:

> It's a pretty straightforward case I think...

The part that makes it more complicated than what it looks is that the surface seems to be slightly creased along the top:



This type of surface configuration where things come close to being smooth but are actually off by about 5 to 10 degrees or so tends to complicate filleting quite a bit because the fillets in those areas will also be slightly differently shaped where they collide together instead of having a totally natural shared end. The amount that they're off will increase as the fillet radius increases and when it gets to be off by enough it will attempt to build a corner juncture patch in that spot but that's difficult to do with tiny slivery pieces.

You'd most likely get a better fillet result if those things were totally smooth in that area - probably the easiest way to do that would be to construct the entire surface all in one sweep rather than trying to sweep just half and then mirroring it. When you build things in separate chunks it's fairly easy for it to not be seamless when mirrored, with 2 rail sweep it will tend to happen because the "in between" profiles for your sweep do not necessarily stay perfectly vertical like your source profiles are, the sweep slides things along both rails by distance traveled along the rail and that can generate some slanting with respect to the mirror plane. At least I suspect that's what happened to you here. It tends to be easier to get smoothness if you build the whole thing with one sweep so try mirroring the profiles to make closed profile curves and sweep those closed profiles instead of halves.

- Michael