more stuff

[Luka <luka@MIT.EDU>]

1. end to end pipeline and firewalk integration

the mit version of the end to end pipeline seems to be in good shape.
if you follow the instructions on the bmg page that patrick mentioned,
it appears to just work.  lots of room for improvement tho.  i did a
partial automated fetch-and-build (bldgs 1 and 2 -- just because i
saw no point to thrashing through the whole campus quite yet).

i then loaded the generated bldg 1 basement model at least into
firewalk, built a walkthru db out of it and viewed it in walkthru.
a few issues were revealed.  first thing, the UG parser in firewalk
completely chokes on two things in the generated UG files at least:
leading @-signs in names, and any slashes in names.  i did a quick
search/replace with funky strings to get around that and afterwards
it went through the build just fine.  we need to see if this is
outside the standard UG spec or not, and either fix our name
encodings or the parser, respectively.

also afterwards, some issues turned up in the walkthru viewer that
i will look at some more and try to figure out soon, but nothing
super critical.  i got a view and ran around, and the vis stuff
appeared to be working at least useably, even tho like i said i
definitely see some glitches.

so next up, i work on getting these pieces to plug together seamlessly,
which might just mean fixing that naming problem, and then integrating
everything into a good top-level build system.

2. basemap

i also managed to get on that windows box and view our dxf's for the
the baseplan.  here's the final summary:

  (A) BMG/end_to_end/2dbase.dxf :

    looks like a map of SE cambridge, with the MIT bldgs colored
    (separate layer?) and labelled with building numbers.  this file
    is about 7 MB, and has dos-style CR/LF newlines.

  (B) BMG/Basemap.ok/Contours/2dbaseASCII.dxf

    looks identical on screen to the file above.  but this file is
    only about 2 MB, and has unix-style

  (C) BMG/Basemap.ok/Contours/Campus_Basemap.dxf

    this file is about 7 MB, shows SE cambridge, and has all kinds
    of psychedlic colors all over the place.  looks like all kinds
    of subtle features (roads, curbs, sidewalks, etc, etc....) have
    all been separated onto separate layers.

  so, i'm guessing that A/B above are the "old" basemap (with A being
  the source straight from facilities and B being somehow trimmed but
  i don't know how yet), and C is the "new" more recent basemap that
  acad2ug supposedly does not grok yet.  does this seem right?  anyone
  have ideas what the change from A->B entailed.  other comments?

  additionally, there are two topo elevation files of unknown origin:

  (D) mit/src/heights/contour27.dxf

    this contains contour lines separating regions of differing
    elevation.  lines that are multiples of 10 (feet? meters?) are
    also labelled with text.

  (E) mit/src/heights/dtm27.dxf

    this contains just discrete points that have x,y,z values.

  both of those files are quite high quality ~40 MB files covering
  the entirety of cambridge.

that's about it.  there's some other non-relevant dxfs around, and
a few extra copies of the one labelled (B) above but i won't bother
detailing that here.

note also that this is a summary just of the stuff under the
walkthru module.  there may or may not be more data files elsewhere
that matter.  if there are and someone knows of them please pass
that info along to the rest of us...

3. dream

i just had this cool dream last night.  i was demoing firewalk on
a 3d cave display and it had the entire mit dataset loaded in, but
generated from floorplan and city data from 1958.  so we were
walking around in this fully rendered immersive environment showing
exactly what MIT may well have looked like almost 2 decades before
i was even born, deduced automagically from old architectural data.

just sharing, because i thought it was very cool, and maybe
the sort of thing we could actually end up able to do someday.

-pl