|individual (unique) interpretation of hyper-reality that was carried out by computer processing algorithms.
The nature of VR as an interpretation model involves two main elements: a perceiving subject (e.g. researcher) and an object of knowledge (e.g. modelled archaeoreality). At the production stage the link creator – data, sustains this dual system. When an external viewer is engaged, one substitutes a model creator, while a model becomes an object of knowledge itself. From the other side, a model is a medium of interaction between a perceiving researcher and recorded archaeological data. The productive interaction can be executed through two main channels: analytical functions of a model and/or visual perception. To support the first channel the model is to be an integrated part of a GIS analytical environment, as for the second, it is based on the notion of a researcher being actively involved within a model. The dynamic model generated by the VRML technique, with a mobile viewer (engaged person) navigating freely through a modelled archaeoreality environment, seems to suit the requirements. Any pre-programmed set patterns of viewers dislocation and movements are to dramatically decrease the potential of a model.
The last theoretical issue I would like to examine is the mentioned concept of realism of a model. We must realize, that ever increasing visual sophistication does not bring a model closer to reality. Realistic, in this case, does not mean real. Rather then spend time and costly efforts on stunning but static re-constructions, featuring ever unparalleled visual details, I would propose to concentrate on an alternative notion of realism. In the mainstream of formulated conceptual underpinnings the term realistic is referred to the initial data. Its integrity and homogeneity, along with volume and quality, provide true realism. Thus, the demands for authenticity and faithfulness applied to a model are incorrect, and to be redirected to the initial raw data.
While working on the (for Russia, pioneering) project to virtually model one of the pilot sites of medieval time in the North Caucasus, these theoretical issues have been evoked. The low project budget made digital surveying equipment unaffordable. In this case, the initial spatial data was carefully manually obtained with simple optical theodolites and recorded and stored in a paper form. With the integrity analysis carried out through the data, it was considered sufficient for creating a preliminary virtual model.
3. The Kiafar project
The Kiafar site is situated in the Karachai-Circassian Republic in the Russian North Caucasus. In the 1st and early 2nd millennia A.D. the Iranian-speaking Alans inhabited the region. By the 10-th century A.D. several dozens of strongholds had emerged there. Nearly all of them were equipped with complicated fortifications, well thought-out and adapted to the environment. The well-organized Kiafar township is the largest of the fortress towns of the period between the 10th and 12th century A.D. The site with all its 1.8 km length is stretched along the narrow Kiafar-Argun river valley, and lies on the mountain ridge. The highest point of Kiafar, the peak, is some 180m above the bottom of the valley, totalling approx. 1200m above the Baltic sea level (figure 1).
Kiafar was chosen as a pilot site for 3-D modelling and subsequent visualization for the following reasons:
1. This site can boast a comprehensive survey database indispensable for adequate modelling. The archaeological expedition headed by Irina Arzhantseva has succeeded in producing a detailed 1:500-scale topographical plan with