Hi mshideler,

> I have wondered how simple the concepts are on
> fillets when looking at them from the mathematical
> perspective.

Well, there are several steps involved, like creating offset surfaces, intersecting those offset surfaces to get sort of guide rails for the fillets, then building fillet surfaces, and extending and intersecting those fillet surfaces.

There are potentially tricky parts to each of those stages.

But the really tricky part is constructing corner junctures where many fillets are trying to meet up (like where you have several edges coming together into a shared common point). There isn't really a single mathematical solution for that, it's basically handled by a lot of special case code that sorts things out into different categories of junctures and knows how to handle each of those special cases.

The filleting engines that work better tend to have had a lot more time invested in doing a whole bunch of those different kinds of cases.

But that's why it's difficult to make a really robust filleter, it's not just guided by one single code path that can just be refined.

- Michael