Cl PA 2003 XIX 11 ' International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
the current case study edifices and virtual humans are fully 
covered by Foni et. al. 
This paper solely focuses on the real-time immersive 
visualisation and interactive narratives of the recreated 
ceremony with real-time virtual characters. Especially as none 
of the cultural heritage applications covered in Foni et al, is 
focusing on the ceremonial character-based restitution or 
provides immersive VR real-time visualisation (3D stereo 
viewing). The present case study illustrates the adoption in 
virtual heritage of not only architectural visualisation but also 
high quality virtual character recreation with face, body, cloth, 
speech simulation. Furthermore, the Cathedral of Hagia Sophia, 
one of the largest and most significant cathedrals in the world, 
has never been fully virtually restored and visualised in real 
time before. Ponder et al describes the main architecture, 
requirements, specification and uses of the VHD++ real-time 
framework. Finally Ulicny et al presents another application of 
the same VHD++ framework for virtual heritage dedicated to 
crowd simulation and rule-based scenarios. 
2. REAL-TIME CHARACTER SIMULATION 
2.1 Virtual Namaz pray 
The detailed scenario of the Morning Namaz Pray during the 
16 th century restored Hagia Sophia and S. Sergius and Bacchus 
edifices, was prepared by the [YTU] historians, architects and 
scientific personnel. Following that was the motion capture 
process, defined by [Molet et ai] utilising the VICON optical 
motion capture system and a real person re-enacting the Namaz 
pray thus being recorded digitally. After the needed post 
processing the individual movements that correspond to the 
various parts of the pray are available as separate keyframe 
animations, ready to be applied to the H-Anim created virtual 
characters. After the YTU and DTU (Denmark Technical 
University) teams provided the simulated echoic 3D sound 
speeches that correspond to the verses of the pray, it was 
possible to have a database of matched speeches and body 
animations. Thus the 2 essential parts of the Namaz pray 
ceremony, were then ready to be applied to the virtual Imam 
and other characters. 
2.2 Virtual Characters 
After the modelling of the H-Anim virtual characters (as 
described in Foni et al) the face processing starts according to 
the processes and tools described in Kshirsagar et al so that the 
virtual characters are ready for facial animation and speech 
simulation. Next step in the pre-processing stage is when the 
virtual body and clothes are being prepared for animation and 
deformation, according to the skinning process described in 
Cordier et.al. Finally the fully body-cloth-face-speech enabled 
virtual character is exported in the H-Anim format. The 
following figure illustrates some resulted virtual characters. 
Figure 1 Character source images (left) vs. 3D 
modelled characters (right) 
2.3 Evolution of the edifice throughout time 
Due to the hermeneutic need to revive the time-dependent 
context of the restored edifices, the need has arisen for 
restitution as churches originally and later as been transformed 
to mosques. However, the Byzantine restitution of the edifices 
proved of high risk as not sufficient architectural-historical- 
archaeological data were at hand. Thus it was commonly agreed 
after discussions with the CAHRISMA project cultural partners, 
that the Byzantine interior restitution will depict only 
architectural modifications without image-fresco restorations 
which were unobtainable in most cases. Thus the simulations 
resulted as depicted in the Results section. 
2.4 Rapid application prototyping using a real-time 
framework for VAR character simulation 
In order to proceed with the rapid development of such 
demanding high performance interactive immersive VR 
application, featuring advanced virtual human simulation 
technologies we adopted the VHD++ real-time framework as 
described by Ponder et al. (illustrated in the following figure) 
frui VHDtt Otivclopiltctu FfjmortOfA 
Figure 2 Concept of a generic VHD++ runtime engine 
with components (plug-ins) 
VHD++ addresses the most common problems related to the 
development, extension and integration of heterogeneous 
simulation technologies under a single system roof while 
combines both framework (complexity curbing) and component 
(complexity hiding) based software engineering methodologies. 
As a result a large scale architecture and code reuse is being 
achieved which radically increases development efficiency and 
robustness of the final VR and virtual character application. 
Figure 2 illustrates a graphical overview of the VHD++ 
framework. Without the adoption of such a framework it would 
have been extremely time consuming and technically not 
guaranteed that we would result in such a VR application 
featuring all these real-time heterogeneous technologies 
supported by the necessary processing tools: a) immersive 3D 
real-time graphics, b) immersive 3D sound, c) VR interaction, 
d) virtual human animation with skinning, e) real-time cloth 
simulation, 0 facial animation, emotion and speech.