[vtkusers] refinement of iso surface

[Dale "Luke" Peterson]

> Subdivision wont add any more detail. It will just make the artifacts worse
> I think.

I tried it along with refining all the new points by solving the
function (for x, y, or z depending on the normal at each point) using
a Newton-Raphson (the initial guesses are very good so it converges to
machine precision in about 5-7 iterations).  This works well in the
sense that the resulting surface is very accurate and reasonably well
represented, but results in way too many points, especially in places
where the curvature is very small.  Subdivision doesn't weigh
curvature when generating new points, so you get too many points in
some places and not enough points in others.  Even without the
Newton-Raphson approach, it was really slow.

I also tried the decimation approach you suggested, but having to
increase the number of samples in the volume also made things quite
slow.  For now, I think I will just have to live with it.

>> > VTK does not have an adaptive isosurface algorithm.

This seems like it would be a useful filter to have in VTK.  I found
some work by V. Pascucci that seems promising but I'm definitely not
an expert on the matter.  Are there "standard" or well accepted
isosurface refinement algorithms out there, and if so, would they fit
into the VTK framework without major adjustments?  If you have any
papers on the subject send me a link and I'll check them out.

Thanks,
Luke