Hi Burr, well I'm hoping to avoid spending years and years on it. Hopefully I will have something not too much longer, I think it may be possible that there won't be one single mechanism that will cover every kind of situation so there might be a few different modes each of which takes a different approach and works better for some different kinds of situations.


@Martin,
> I wish you all the best with what you're doing there, Michael.

Thanks very much Martin!

So far it's been mostly reading tons of headache inducing research papers and several failed experiments. I think that I'm getting to a point where I'm seeing the kinds of things that are most problematic though, so some of my newest experiments seem to be going better.


It's difficult because trying to force the surface through the boundary areas tends to result in wiggles in other parts of the surface, forcing things through a certain shape and also having things smooth are basically sort of opposing forces.


A lot of research results use some kind of special purpose surface that is not directly compatible with a regular NURBS surface. That's for example why the ACIS one behaves so differently when they export it. I found the paper from 1994 that describes the mechanism that the analytic ACIS blend uses, and it's only well defined on the interior of the polygonal region of the boundaries, it does not behave well for points outside the boundary so it has to switch to a completely different approximate fitting mechanism when it needs to convert to a regular trimmed NURBS surface.


- Michael