Re: BMG Roundup - 10/14 Mtg

[Seth Teller <teller@csail.mit.edu>]

after the 2d footprint of the building is computed, its
edges can be inserted into the basemap triangulation.
then all triangles within the footprint can be deleted.
as you say, this produces a footprint-shaped "hole" in
the basemap.

the "join" of the footprint and the basemap will be a
"1 and a 1/2-dimensional" contour -- a contour whose
projection onto z=0 is the building footprint, but which
is constrained to lie in the basemap, and so has varying
height.

thus, to insert the building and produce a "watertight"
resulting model, each vertical building face should have
a lower envelope that exactly matches (vertex for vertex,
edge for edge) the basemap, evaluated along the footprint.

as a first pass, we don't need the model to be watertight,
so i'd suggest simply making sure that the lower envelope
of the building's vertical faces be extended deep enough
(i.e., to a sufficiently low z value) that no gaps are
left between the bottom of the building faces and the top
of the join.

seth

Patrick Nichols wrote:
> All --
>
> Just a roundup from our meeting yesterday. We've agreed to do the 
> following:
>
> - Have BMG input data to LocationServer be pre-triangulated (reducing 
> pre-computation costs for the LS, and keeping a consistent triangulated 
> view of the world in the pipeline)
> - Think deeply about how to place buildings on the basemap for vitaly wrt 
> Z axis placement (do we carve out holes in the ground and then smooth over 
> after the building is placed?)
>
> Thats pretty much it. Applet should be online after this weekend for some 
> 2d route stuff.
>
> -Patrick