Gordon bug?

 From:  Michael Gibson
592.2 In reply to 592.1 
Hi Pilou, you mean the sort of Triangular dimple that is created?

I believe this is actually the expected result of a Gordon surface. I tried it in Rhino and got a similar result.

Gordon is not quite like sweep, the curves that are coming together to a point there are forming a pole, but the different shapes kind of conflict at the pole, 2 are kind of convex shaped and the middle one concave. Gordon/Network surface kind of tries to average those opposite shapes together at that point, creating a kind of ridge in the middle of the averaging, that is what results in that type of triangular looking pinch.

So I think you generally don't want to use network surface with that type of arrangement - the curves should generally have a somewhat similar shape when they come together at a pole.

One possibility would be to use Network surface to build the upper part, and then do a sweep for the lower curved portion where it bends differently. Here is another method using the bottom part as a profile curve and then 2 upright pieces as rails for a 2-rail sweep:

The other way to do this for Gordon surface I think would be to slice up the curves such that the pole is at the very top of the shape, instead of along the bottom. The top does not have conflicting directions between the curves so that should work better. I tried this and it seems to fail though, I am going to be working on fixing up some Gordon surface bugs pretty soon.

It kind of looks like a fillet-type piece at the bottom. Sometimes it is easier to use fillet to put in this type of a blend instead of trying to surface it all in one piece.

- Michael