Towards Three-Dimensional Non-Invasive Recording of Incised Rock Art

Article excerpt

Traditional recording methods and problems associated with them

Recording is the essential prerequisite of any database compiled for research and conservation programmes for rock art. In the British Isles several methods have been employed which can be grouped into two major categories: two-dimensional and three-dimensional recordings. The first one includes techniques already in use in the nineteenth century such as free hand drawing and casts, and others appearing later: tracing, rubbing (Beckensall 1983: 32), photographs and digital image processing (Donnan 1999). Three major problems affect these techniques. Firstly, they document in two dimensions what are essentially three-dimensional surfaces and volumes which usually results in inaccuracies that can sometimes be important (Figure 1, see also Coles 2003). A second major problem, as Loendorf has recently highlighted, mentioning rubbing in particular, is that it is proven to have a damaging effect in samples taken for dating (Loendorf 2001: 57). Given the difficulties with dating engravings experienced in Foz Coa (Zilhao 1996) this is a problem liable to affect British rock art in the future, even if it does not now. Finally, casting, tracing and robbing are invasive techniques and may affect rock art preservation. Although in his article on recording Loendorf considers invasive techniques (especially in this case tracing) as potentially harmful, he acknowledges petroglyphs are usually more durable. However, this is not always the case, as the condition of some rock art surfaces in Britain (such as Achnabreck) shows. Preservation is a major concern. In the main report of the Rock Art Pilot Project commissioned by English Heritage, photography was recommended as a non-invasive technique (RAPP 2000: 88). Historic Scotland follows a policy of non-contact techniques for recording rock carvings (Yates et al. 1999).

[FIGURE 1 OMITTED]

The second category of techniques used in recording prehistoric carvings in the British Isles comprise those recording in three dimensions: laser scanning (Eklund & Fowles 2003) and three-dimensional modelling from photographs. Both are still in an experimental stage. They have the advantage of overcoming the problems associated with 2D recording mentioned above and require no contact. Despite the potential of laser scanning, it is still a high cost option and therefore the opportunity for its use is currently restricted. 3D modelling from photographs, however, is more accessible at low cost. In order to investigate its potential a programme of study was established at the Department of Archaeology, University of Durham, England. The aim of the project wig to test the feasibility and accuracy of a method of recording rock art in 3D, in order to maintain the "real-world" spatial relationship between motifs overcoming the inaccuracies of 2D recording.

3D recording of the Horseshoe rock

Relatively inexpensive commercial software, mostly based on single camera photogrammetric techniques, is now available for creating precise 3D models from photographs. Photomodeler (produced by Eos Systems, Canada) is an example of such a programme and its use is illustrated in a number of archaeological examples on the manufacturers website. (www.photomodeler.com and www.3dphoto.dk), in England a project developing methodologies for 3D visual representations of megalithic monuments was also based on the use of a previous version of Photomodeler (Gillings 2000). The study presented here is an investigation into the suitability of Photomodeler (version 4) for the recording of rock art.

Photomodeler uses identifiable points within a series of overlapping photographs to calculate the position of the camera. If the characteristics of the camera and lens combination are known, then the 3D co-ordinates of those points can be determined and a wire frame model can be constructed. Visual detail can then be laid over the wire-frame model from photographs taken without the control points. …