bul 2004 International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
  
e also 
g the 
m two 
3. FOLLOW THE GUIDE... 
We hope that the detail in the model makes it interesting for the 
public to explore this virtual building. It is impossible to show 
all parts of it in this paper, but we will take the reader on a short 
guided tour that highlights a few spots where fidelity to detail 
will be seen to matter. 
A first example is the 4 Medusa heads on the 4 middle 
pediments. When seen from a sufficiently close distance, it 
becomes immediately clear that their quality is quite different. 
Facing the monument, quality goes up from left to right. In 
ancient times, it would have been from the right that one would 
have entered or left the agora. Hence, it is the Medusa on the 
right that one would typically have seen most up-close. Fig. 15 
shows the first and the third Medusa head. As can be seen, the 
head at the top is more a bas-relief than a truly 3D head like the 
one at the bottom, that really sticks out of the back plane and 
tilts over to have it look down onto the square. The difference 
in artistic quality is obvious. These differences are most 
probably not accidental, but may have to be interpreted in 
relation to their relative visibility for the typical visitor of the 
square. 
Torre > 
  
  
Figure 14. Comparison of the two rendering methods: On the 
top ‘faked global illumination’, on the bottom ‘ambient 
occlusion’. 
  
Figure 15. Close-ups of the first (top) and third (bottom) of the 
Medusa heads, counting from left to right, showing the 
difference in quality. 
We have raised the issue of exaggerated regularity and 
symmetry that is typical for simplified models. In fig. 14 one 
can e.g. see that below the head the back plane surface is 
rounded near the bottom, rather than forming a straight angle 
with the border of the pediment like everywhere else. To an 
observer on the agora, this rounding off would not have been 
visible, however, and thus time and money were saved. At 
several places in the monument, the lower parts of the gables, 
not visible from below, were left unfinished in this way. 
Again referring to fig. 14, it is interesting to notice the 
irregularities in the relative placement and the sizes of the 
dentils on the back plane above the Medusa head. These virtual 
dentils follow the captured 3D data closely. This is an example 
of a deviation from expected symmetry and repetition. 
Similarly fig. 16 illustrates how the shapes of the arches above 
the niches show noticeable deviations from a pure semi-circular 
shape. Again, the actual level of symmetry is lower than one 
might expect. 
Furthermore, several aspects of the building have never been 
finished. Fig. 17 shows the second niche (from the left). The 
thyrsus staff on the left has not been carved out completely 
(compare with the complete pattern on the right). It was merely 
started.