What I do is that: I scan a volume by an octree; to decide if one cube has
to be subdivided I look for the values on the vertices of the cube.
At last level of detail, I have nodes-cubes and I know their configuration,
so with marching cubes I find the triangles that aproximates the surface
intersecting the cube-node, and I store the vertices of these triangles
in
my list.
So These triangles form a 3D volume, but the only information I reach
about
a 3D point-sampled is if
this point is inside or outside my volume, so I haven't a scalar value
that
gives me possibility to interpolate them and so to smooth the volume.
(The good think is that the triangles are almost of the same size.)
Perhaps the better think could be storing topological information while
building the octree.
Unfortunately ... I don't know nothing about 'scanline conversion'.
Thanks for the suggestion about 'winged edge structures', i will look around
for it, 'till now I don't understand how I can know, during my marching
c.,
if two triangles share one edge
|
