Hi Seth & All, All, All,
Below is the information you asked me about. You can find more in my thesis if
you take a look at Appendix A, but please feel free to ask me if you guys have
any problems.
1. One place where you can find the correct contour information is the
/walkthru/mit/src/BaseGen/Input/ directory. I do not remember the name of the
file, but it is very likely to have .cntr or .cntr.fixed extension. It should
not be difficult to find it, since it is the only file in the .cntr format in
the directory (you can also find the description of the format in the very end
of my thesis). There also must be a copy (as well as the original, incorrect
version of the file) inside the /BaseGen/BldgMap/Data/ directory (Building
Mapper is the application that allows you to fix contours).
2. The simplest way to run the part of the pipeline responsible for the basemap
generation is to run the build_basemap.sh script located in /BaseGen/
directory. The script takes the data from the /BaseGen/Input/ directory
(therefore, it will automatically pick up the right contour file; you do not
really need to look for it separately, or you can just examine the script to
find it), and outputs the resultant model as well as space+adjacency info and
debugging information into /BaseGen/Output/
Because we are still using the old (and erroneous) version of the basemap, the
model that you get by running build_basemap.sh (i.e. running everything
automatically) is full of space-labelling errors. I remember I talked to MIC
about the possibility of adding a switch to one of our pipeline scripts to
allow fetching the good, fixed-by-hand version of the model of the basemap
until we get better data from the MIT DOF people. Because of the way the model
is generated, it is possible to pick up a fixed, 2D version of the model, and
then run it through the rest of the basemap-processing stages in order to
generate a decently looking 3D model + all the space-adjacency information
necessary for the route-finding algorithm.
I think this is it for the moment.
Regards,
Vitaliy
P.S. It may be so that I come to Boston for this weekend. I will not have a lot
of time, but I may be able to help if you have any problems.
Quoting Benjamin Vandiver <benmv@MIT.EDU>:
TODO
revisit binding labels to spaces in walls2
split walls into chunks
flag errors with visual markers rather than abort
+integrate vitaly's xform data into pipeline
+support route-finding service MITquest
-stata center floor plan conversion
-integrate furnit
+integrate VK's basemap into pipeline to produce basemap + buildings
+hand off knowledge of pipeline (MIC -> B, E)
group:
walls2 breakup
ben:
route-finding
emily:
VK xform data into pipeline
mic:
VK's basemap code into pipeline