CIPA 2005 XX International Symposium, 26 September - 01 October. 2005, Torino. Italy 
pillars that replaced one that collapsed. These elements can all 
be removed in the 3D model so the site could be viewed in the 
correct historical context. Many aspects of sensing and 
modeling must be understood before starting such a large 
project. The typical processing pipeline used for 3D modeling 
includes geometric modeling and appearance modeling. Here, 
we summarize the results of the virtualization of the Byzantine 
Crypt. The detailed technical aspects of the project are 
described in Beraldin et al., 2005. 
Figure 2. Entrance page of the CDROM CARPINIANA. 
2.2 CDROM and video animation: CARPINIANA 
We are currently working with three models: one 4.6 million- 
polygonal un-textured model (10 mm resolution) of the 
complete Crypt, a 12.8 million-polygon fully textured model (5 
mm resolution) of one half of the Crypt (contains the two 
apses), and, a lighter textured model with 0.4 million polygons. 
These different models were further transformed into 
orthophotos, isometric cross-sections and smaller 3D models in 
order to navigate through the information on the CDROM. All 
of these representations are aimed at showing the three- 
dimensionality of the site that is not visible in a typical visit to 
the site. A movie called "Carpiniana" showing a fly through of 
the Byzantine Crypt was also prepared. Snapshots of the 
CDROM and Video animation are shown on Figure 2 and 
Figure 3 respectively. When the animation was realized, i.e. 
year 2002, the computing power was not sufficient to deal with 
the high-resolution model. The software 3DStudioMax helped 
create the animation. The model used contains 400 000 
triangles, 1/5 of the maximum texture resolution, 13 lights, 
5000 frames at a 720 x 576 resolution. Today, the full 
resolution (shape & texture) could be used along with a more 
complex lighting arrangement. The presentation of the 
Byzantine Crypt is now available through a virtual reality 
theatre (with "il teatro virtuale" software) that can display the 
full resolution model and allows for a real-time navigation 
inside the Crypt for further study. A few large format holograms 
(1.8 m x 0.85 m) were also produced from the digital 3D model. 
3. TEMPLE C OF SELINUNTE 
The project that started in 2003 is divided into two broad steps, 
the first step saw the modeling of the frieze of temple C of 
Selinunte using 3D laser scanning and the second step will see 
the reconstruction of temple C of the Acropolis of Selinunte 
using photogrammetry and CAD tools. The 3D model will be 
based on historical information available at the University of 
Lecce and at the "Museo Archeologico Regionale" of Palermo, 
Sicily. In the first step, scanning and modeling of three Metopes 
from the temple were done in the regional museum of 
archaeology of Palermo, Italy. 
Figure 3. Still image taken from the movie included on the 
DVD: view of crypt without texture. In the video animation, the 
colour information is introduced half way in the movie. 
3.1 Practical considerations 
A visit to the museum and to the site allowed the team to plan 
the activities. One of the main concerns was the determination 
of the required spatial resolution and the technical difficulties 
the team might encounter. The spatial resolution depends on the 
level of details desired by the project but also by the equipment 
available and practicality of the choice (Beraldin et al., 2005). 
The spatial resolution picked for merging the 3D images 
together determines the size of the smallest triangle on the 
mesh. Other issues can come from the hidden surfaces hard to 
reach areas and vibrations induced when scaffoldings are used. 
Both scaffoldings and a mirror placed in the path of the laser 
beam were used to reach some of those 3D surfaces (Figure 4). 
Figure 4. Practical considerations when using a 3D scanner: 
mirror used to reach hidden 3D surfaces. 
3.2 The three Metopes of Temple C 
The 3D model of the frieze from Temple C including the three 
Metopes is shown on Figure 5. We continued the laser scanning 
work using both the Minolta 900 and Mensi laser scanners on 
different sections inside the museum room. The first scanner 
was used to acquire details in the order of 0.5 mm while the 
second scanner (Mensi SOISIC-2000) captured details in the