How we have prepared the model for VRML viewing

  1. We began with an AutoCad model of the stupa, which produced a file of about 1.2Mb;
  2. A program called VRMLout allowed us to export sections of the model into VRML, for manipulation over the web using Whip!;
  3. The basic problem is that to load a VRML model of the complete stupa into any machine, clothed with all the figurative and decorative reliefs, would place intolerable loads on the technology, and cause any movement to slow down to the speed of molasses;
  4. Again, we had constantly to bear in mind the speed of image transmission across the web, and tested this using an ordinary Pentium 133, admittedly at ethernet speeds;
  5. We conclude that it would be impossible to manipulate any portion of the Borobudur model across the net at any less than ethernet speeds, and hence will prepare CDROM presentations of the Project;
  6. A further complication if provided by the usual quality versus speed dilemma: just how far do we degrade quality in order to gain speed?
  7. One way of solving this is to allow image suites of different dimensions: in the finished project, the user should be able to select a suitable resolutoion just as a gear-change would be effected in a car; the smallest image suite offers a resolution of 366x278, with the advantage that the complete image suite, including details of each relief, occupies no more than 221Mb. The largest images are 1.6 megapixels, with the result that the full suite occupies more than two full CDROMs.
  8. For viewing the stupa under VRML, we have adopted a policy of fragmentation, as follows:
    1. General views of the stupa clothe only one-quarter or one-half of the stupa with reliefs. The remaining portions show the floors of each of the galleries or terrace, so the user can maintain bearings;
    2. The stupa consists of a basement, four galleries and a terrace, all rich in reliefs or, for the stupa terrace, 3D objects "expensive" to model in VRML;
    3. If you view the stupa, you will notice that each gallery (but not the basement) is clothed with reliefs both uphill (main wall) and downhill (balustrade) - too many reliefs for the current state of machines and VRML, hence the solution adopted below;
    4. For touring the stupa, the stupa terrace, each gallery and the basement have been sliced up into four sections or quadrants, each again divided into main wall and balustrade:
      • Basement: NW NE SW SE quadrants
      • Gallery1 main: NW NE SW SE quadrants
      • Gallery1 balustrade: NW NE SW SE quadrants
      • Gallery2 main: NW NE SW SE quadrants
      • Gallery2 balustrade: NW NE SW SE quadrants
      • Gallery3 main: NW NE SW SE quadrants
      • Gallery3 balustrade: NW NE SW SE quadrants
      • Gallery main4: NW NE SW SE quadrants
      • Gallery4 balustrade: NW NE SW SE quadrants
      • Stupa Terrace: NW NE SW SE quadrants
      • Total of 40 VRML "tours"
  9. Conceivably, these tours can be linked together seamlessly as machines/VRML get faster. Thus a user could moce seamlessly from Gallery One NW to NE Quadrants and, perhaps, view the VRML for both the main and the balustrade sides of each gallery at once.