> I have attached a reference image here, that I wanted to ask Michael
> for tips on the most efficient nurbs approach to modelling it.
> It seems like a good candidate for nurbs. I recently poly modelled it
> and found the process straight-forward but tedious.

Branching structures can actually be pretty difficult to model with NURBS.

One method that can work is to sort of break the model down into smaller components where each individual component is easier to surface. Keep breaking things down and you can usually get things down to a little hole in the middle somewhere that you just fill in.

In the attached model, I started by drawing some of the recognizable curves from the drawing.

Then there are a couple of pieces that are pretty straightforward, I started with those, one was an extrude and the other a two rail sweep. Then I drew some more curves to connect pieces, did more 2-rail sweeps to surface them (mirroring them to create the bottom parts), then finally I just filled in the small remaining hole with a loft between 2 joined lines and a smooth edge.

So this kind of has its own tediousness in NURBS as well, but at the same time this particular model actually shows off the amount of accuracy and control you have with NURBS as well, because the design of the profile in this case has exact circular pieces for the inside and outside curves. With this NURBS construction, that exact circular profile shape is preserved to within 0.01 units tolerance - that's awfully hard to do with a polygon model.

- Michael