| ||||||
Appendix A: Application of Digital Photogrammetry
A.1 Approach
In any form of measurement and documentation, there is a tradeoff of speed vs. accuracy. For this study, the objective was to maximize speed: covering the key areas as broadly as possible, with less emphasis on accuracy. The study also sought to use inexpensive equipment, using a conventional 35 mm SLR camera, and off-the-shelf software costing less than $700, running on a personal computer costing approximately $3,000. This arrangement stands in contrast to some commercial applications of photogrammetry, where the combined cost of specialized cameras and software may exceed $30,000.
This rapid and inexpensive approach to photogrammetry naturally sacrifices dimensional accuracy. Accurate results are difficult to achieve in many areas of the Macellum because the accuracy depends on having photographs where the viewpoints have wide angular separation: something which is extremely difficult to do in some of the confined spaces and narrow streets around the building, so some models are much less accurate than others, depending on the availability of camera angles.
Although many of the models have evident dimensional errors, they are nonetheless useful in understanding the general patterns of failure, and in creating three dimensional models of the buildings at different points in the timeline of damage and repair.
A.2 Procedure
Figure A.1 below shows a screen dump from the Photomodeler program. After selecting several photographs of a region of the building, common points are marked on the photos, plus triangular faces defining the surfaces of the walls. The program then calculates the coordinates of the points and maps chosen photographs onto the triangular faces. The calculation of coordinates is within a scale factor; the generated models were then scaled to fit the plan of the building.
| Figure A.1: Plan of the Macellum showing regions of the ten photo-mapped models |
A.3 Documentation of the Macellum
Approximately three quarters of the Macellum perimeter was modelled with ten separate photo-mapped models. The models omit the southwestern quarter of the building, since there is little visible evidence of seismic damage or repair in this area. Figure A.1 below shows the areas covered by the ten models, and the listing below shows the photographs for each model, plus a link to a VRML model.
| Figure A.2: Plan of the Macellum showing regions of the ten photo-mapped models |
nwshops
23 Photos. VRML model
 |
|  |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
neshops
12 Photos. VRML model
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
 |
 |
nwall
7 Photos. VRML model
 |
 |
|
|
 |
|
 |
necorner
14 Photos. VRML model
 |
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
ewall-center
10 Photos. VRML model
 |
 |
|
|
 |
 |
 |
 |
 |
 |
ewall-south
9 Photos. VRML model
 |
 |
 |
 |
|
 |
 |
 |
 |
shrine
5 Photos. VRML model
 |
 |
 |
 |
 |
segate
13 Photos. VRML model
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
seshops
8 Photos. VRML model
|
 |
|
 |
 |
 |
 |
|
swwall
10 Photos.
 |
 |
|
|
 |
|
|
 |
 |
 |
| ||||||