Page 218 - Green - Maritime Archaeology: A Technical Handbook. 2nd ed

Chapter 6: Photogrammetric Techniques 197
measurements checked to ensure they made a good fix. Sometimes the person taking the readings was sent down on their next dive to take more readings or to recheck measurements. If the readings produced a poor fix, the program would indicate which of the four measurements was wrong, or alternatively if the orientation of the selected control points was producing a poor fix. Typically the RMS residual for individual objects was about 13mm (distance measurements typically ranging from 1–10m).
Using PhotoModeler, the procedure was quite different. Once excava- tors had uncovered an amphora, they were issued a mapping label with a designated number and type. After ensuring that the rim was clean and free from concretion, the mapping label was carefully attached to the mouth of the amphora, apex toward the handle. A photographic dive would then be scheduled with a list of amphora to be recorded. The photographer had to ensure that all the mapping labels were viewed from at least four (prefer- ably six) different positions. This was at times difficult to do, because the amphorae were often lying horizontally with their mouths pointing in random directions, occasionally pointing downward. Thus in a particular view some amphorae would be in view while others were pointing away from the camera obscuring the mapping label.
Similarly, high vertical photographs would often not show the mapping labels of amphorae that were horizontal. Generally the photographer took a large number of photographs around the site with the hope that cover- age would allow all the labels to be mapped. As a rule of thumb, eight low level (15°) views at 45° azimuth intervals, eight high level (60°), and one vertical would generally ensure adequate coverage. Photogrammetric tar- gets (a circle divided into quadrants of black and white) were also tried to compare with the three round dots (the latter worked well in the Photo- Modeler sub-pixel mode when they could be clearly viewed). These would have allowed more precise positioning of the point marker cursor in PhotoModeler, while not detracting from their visibility. The advantage of PhotoModeler was that once the amphorae were photographed, large numbers of mapping labels could be processed. So in one photographic dive it would theoretically be possible to plot over 100 amphorae, a task that would take several days using conventional tape measurements. In reality, large numbers of amphorae were not uncovered at one time, but progres- sively. So that during any one photographic dive one was usually re-record- ing amphora that had been covered on previous days.
The biggest problem in using PhotoModeler for this type of work is housekeeping. It is essential to establish a strategy for handling the images, the PhotoModeler files, and the output data. Potentially, there will be hun- dreds of points each related to a particular object and cross related to Photo-Modeler files and images. If a strategy is not developed the whole