1. Introduction

The last decade was marked by a significant rise of interest in virtual 3-D modelling among the archaeological community. Some of the potential of the impressive development of virtual technology has been explored in a constantly widening range of its application in archaeological projects and case studies and presented in a number of weighty publications (Forte 1996, Earl 1999, McCullagh et al. 1999). These enthusiastic efforts were mainly concentrated on the realism of modelling, or model authenticity. This resulted in the fact, that while the presentational power of 3-D models is indisputable, their contribution to archaeological interpretation is still problematic. The lack of theoretical underpinnings of existing VR models leads to a contradiction between what we think a model is, and what does it truly represent. In this paper, I intend to focus briefly on the conceptual shortcomings of VR modelling of archaeological sites. The alternative conceptualizations of VR and approaches to modelled data are to be discussed. Another issue of enormous significance to be taken into consideration at this point is the demand for initial data from which models are produced. The currently existing digital survey data archives containing satellite images, GIS databases, digital terrain models (DTM) and other types of spatial data are fragmented. This makes the utilization of paper copy plans, drawings, hand-managed measurements and photos that have been collected over numerous decades, inevitable. In Russia, where there are no digital data archives at all, and the survey data is stored in paper form, this problem is especially vital.

A brief outline of the Kiafar project will be given in the presented theoretical framework, and the potential of data that was initially not digital, nor was it ever supposed to be digitally processed and analyzed for the creation of 3-D virtual models will be demonstrated.

2. VR – seeking new approaches

The concept of VR has been recently discussed to a great extent in a number of publications (see Gillings 1999). Instead of returning to these debates here (which would take a lot of time), I intend to focus your attention on the nature of the interrelationship between reality and virtual reality. The mentioned trend, manifested by VR case studies to make produced models as realistic as possible, originates from a misleading concept of VR being a kind of reality replica (Gillings 1999, Burton et al. 1999). According to this notion, the more realistic, or authentic, the replica seems to be, the less it is considered to be a fake. Thus, the ideal VR model is supposed to be indistinguishable from what it represents. The paradox is, that the most detailed and carefully constructed model will never match the original. At first glance, the explanation of such a phenomenon lies in the specifics of an object of archaeology as a discipline. An archaeologist deals with traces of reality that have been transformed by time, and can be described as archaeoreality. This implies a notion of multiple interpretations caused by individual characters of perception. It brings us to the thesis, that any VR model is an interpretation, rather than a representation of reality.

Further on, it poses a question: is what we interpret, or better say - model, reality? The answer is negative. In fact we interpret (model) an interpretation itself. A VR model is nothing more than data processed with a given algorithm. The data we extract from reality through perception, recognition and, finally, description can be expressed with the term meta-reality, or hyper-reality. The process of modelling this data (hyper-reality) introduces also a second level of interpretation. The resulting model reflects an in-