Ben's House

I'd like to give a virtual tour of my house - from the native trees along the driveway to the shoji drop-ceiling that casts a soft light into the center of the house.

living roomThe architect, Gordon Motta, kindly gave me a copy of the blueprints for the house. There are a several plan and elevation views, and even detailed materials lists. However, I've learned it takes years of knowledge of architecture and construction in general to make sense of a blueprint - not all the 3D positioning and placement info is there, mostly it is left assumed.

Location

From looking the USGS topo DRGs, it looks like the house is at (UTM zone 5 NAD83: 237620, 2219325), which is 20°03'11" N, 155°30'30" W.  We can guess that elevation is around 750 meters because there are two contour lines passing near the property: the 2480' (755.9m) line at the southwest corner by the pond, and the 2440' (743.7m) line at the base of the driveway:

In Google Earth, using their imagery from December 2009, it's at -155.508362°,  20.053239°.  Previous imagery (2003) was shifted around 8 meters off from the 2009 image.

The Land

2.305 acres in a long L shape.  USGS 10m DEM, because it is derived from those two DRG contours that barely approach the land, lacks sufficient detail to get even a rough idea of the elevation of the parcel.  Surveying would be necessary to get elevation contours, but this could easily cost several thousand dollars.

In 2000, i tried using a consumer GPS to walk the property bounds and establish some major features such as the house corners and some larger trees.  However, the GPS device never got better than ~5-7m accuracy, probably due to the numerous trees present, and the results were generally not useful.

In 2002, i tried doing some manual surveying with a level and horizontal ruler, measuring along some lines in the SW corner.  Very labor intensive, and very limited because you have to be able to actually walk along these lines, and the land is too overgrown for that.

In February 2011, i used an IP-S2 system to laser-scan the land from the back of a truck.  The data is only collected near where the truck could drive, but this is roughly half the land.  The GPS trajectories are very accurate because of DGPS and IMU.  Relative to it:

There is an issue with vertical (elevation) coordinates.  According to a very accurate (OPUS) point measured just in front of the upper garden gate:

 REF FRAME: NAD_83(PACP00)(EPOCH:2002.0000)       ITRF00 (EPOCH:2011.1024)
        X:     -5455176.415(m)   0.029(m)          -5455177.372(m)   0.029(m)
        Y:     -2485079.832(m)   0.022(m)          -2485077.310(m)   0.022(m)
        Z:      2173499.339(m)   0.029(m)           2173500.362(m)   0.029(m)
      LAT:   20  3 11.65139      0.033(m)        20  3 11.68458      0.033(m)
    E LON:  204 29 29.08736      0.010(m)       204 29 28.99474      0.010(m)
    W LON:  155 30 30.91264      0.010(m)       155 30 31.00526      0.010(m)
   EL HGT:          773.960(m)   0.030(m)               774.146(m)   0.030(m)
 ORTHO HGT:          752.566(m)   D.N.E. [No official datum supported (FAQs 19,20).]

This "ortho hgt" value (orthometric) agrees very well with the USGS DEM:

Trajectory of truck IMU at a reference point on driveway: 773.4 m (vs. ellipsoid)
Height of truck IMU above ground: 2.1 m
Height of ground measured by GPS/IMU: 771.3 m (vs. ellipsoid)
Height of ground from LIDAR: 750.0 m (vs. geoid)
Difference (ellipsoid - geoid): 21.3 m
Difference (IMU - geoid): 23.4 m

House Modeling

  1. ideally, there would be a "homebuilding wizard"
    • would let you describe your home in general terms
    • procedural modelit would build the house, completely describing the structural elements like bricks and boards, and output the result into a realtime rendering environment
    • there are a few standalone packages like this on the market, but do they let you export the building?
    • so far, it looks like the answer is no.
    • i produced a procedural model using the VTP building support which looks OK, but interfaces poorly with the ground, since the elevation is wildly imprecise
  2. second choice would be to use a general-purpose 3D modeling program to model and render the house
    • 3D Studio MAX is fairly easy to get a hold of, but it's not easy to use for architectural work
    • it does support using English units (since version 1.2)
    • i started out use MAX to enter the floor plan as 2D lines, then
      • created a "box" entity for the floor of each roomhouse in MAX (22k)
      • used "extrude" to create 8' walls
    • but.. walls should have thickness - this gets awkward very quickly.
    • i'd rather start from the "ground up" - start with the foundation, then the main support posts, then the floor, much as carpenters would when building the house.   However, it's not clear what shape the foundation is from the blueprints - it mainly follows the topography of the terrain (a gradual, but significant, slope).  I don't have this topography information.  Where to start?
  3. third choice would be to use a full, traditional CAD/drafting program
    • e.g. AutoCAD, FormZ
    • would require a lot of learning and a lot of labor
    • somewhere between options 2 and 3 is 3D Studio VIZ

Source Materialsblueprint

what i have to work from:

Construction Details

Tax maps

Other useful data might be available in the Hawai‘i County 1997 Data Book, found on the County of Hawai‘i site.