Quite often the “Fillet edges” tool (! _FilletEdge) generates some surfaces with a number of unwanted duplicate control points at the end of the surface, resulting in a bad object or weirdly looking model. It’s tedious to deal with it when it occurs, because it requires multiple steps to be fixed:
The faulty surface must be extracted;
Then it’s rebuilt by manually deleting the extra control points with the “Remove a control point” tool (! _RemoveControlPoint);
Then joining it to the rest of the model again.
I made a short video that shows the issue and the way I fix it. Hopefully there will be an improvement in the “Fillet edges” tool in a future Rhino service release. Nearly 1/3 of the fillets I do appear to have such faulty surfaces, even though I create clean surfaces with minimum amount of control points and use Absolute tolerance of 0,001 mm and 0,1 degrees. Rhino 6 produced far less issues like this one in comparison to Rhino 7.
Sure, attached here is the part that had this issue at the lower edge while making a fillet with a radius of 1 mm or smaller (for some reason the fillets with radius above 1 mm were fine). I also uploaded a video showing how I created the part from quite simple surfaces (it should appear online in about 20 minutes after my post).
Note that this issue sometimes happens on straight extrusions created from curves with degree 1, 2 or 3 (or a combination of these), so it’s not only related to split and joined surfaces with degree 3.
I changed the Absolute tolerance from 0,001 mm to 0,01 mm, but the issue is still present with fillets less than 1 mm. However, when I changed it again to 0,1 mm, the fillet was fine.
I rebuilt the surfaces in this particular area while temporarily changing the Absolute tolerance to 0,0001 mm to split them more accurately. Then I reverted to 0,001 mm tolerance and joined them to the main model. Now the fillets work as usual.
I figured out that with 0,001 mm tolerance Rhino would still split the surfaces in a slightly wrong way and once the surfaces are joined together, there is an extremely small deviation where the 3 surface edges meet together in a T-shaped joint. In the past I wrote requests several times in different topics to use 10 times smaller tolerance only for the “Trim” and “Split” commands, while the rest commands should use the Absolute tolerance. That way, many issues like the one I mentioned here would not occur at all. I did a quick search and just found two samples of my request to have a more precise trim/split setting:
Rhino 6 produced far less issues like this one in comparison to Rhino 7.