Internet Archaeol.8. Virtual 3-D Facial Reconstruction

5. Discussion

We still regard our research and models as preliminary in nature. Although all of the medical imaging methods currently in use allow data collection from living individuals, none is ideal for tissue depth measurement. Ultrasound methods (Helmer 1984; Manhein et al. 2000) allow measurement with the subject in an upright position, but require operator intervention for the collection of each measurement. Computed tomography (CT) is X-ray based and allows the soft and hard tissues to be distinguished easily (Phillips and Smuts 1996). It is not practical for data collection from sample populations, however, as clinical data tend to originate from patients with craniofacial abnormalities and ethical considerations preclude it as a method of data collection from healthy individuals. Following a pilot study (Nelson 1995), we have abandoned CT in favour of methods for extracting soft and hard tissue surfaces from magnetic resonance (MR) images of the head (Ratnam 1999). MR is suitable for the study of healthy populations, but measurements must be taken with the subject lying down, inevitably affecting facial appearance.

Although abandonment of the traditional landmarks in favour of circa 10,000 measurements collected from MR imaging represents a significant increase in the precision and accuracy of facial reconstruction data, we are encountering difficulties in automating the process by which the tissue depth landmarks are located on the skull even using the traditional landmarks. An appealing solution is to use 'deformation' algorithms to 'warp' a generic tissue depth dataset to fit each novel skull (Quatrehomme et al. 1997). Although this approach might be regarded as mathematically or statistically valid, deformation causes sufficient disturbance to the facial surface for it to lose much of its face-like qualities (Quatrehomme et al. 1997, 651).

Rendering of a primitive surface with a photographic quality facial image provides a pragmatic solution to the problem of making a computerised reconstruction life-like (Vanezis et al. 1989). A simple polygon constructed by joining the traditional facial landmarks into a 'wire frame' facial surface would show only a limited conformity with facial shape, but if rendered with a photograph of a face the outcome is quite life-like. This is a method of modelling of faces commonly adopted in the computer games industry. There is a danger that all facial surfaces will strongly resemble the person in the photograph, however, and composite photographic images constructed from a number of real faces are used in an effort to overcome this problem.

There is scope for substantial research on the simulation of skin colour and texture, and on the use of 3-D graphical editing software to enhance the finished reconstruction. Although the development of ageing and other interpolation models represent a significant advance, they presently lack sophistication and fidelity with 'real life'.

Preliminary work has been undertaken examining the value of studies in psychology, art history or portraiture in facial reconstruction. Bruce and Young (1998) have examined the psychology of face perception a fascinating science in its infancy. Our capacity to discriminate unfamiliar faces is extremely poor. Whilst it is interesting to speculate on the evolutionary origins of the demarcation between our capacity to discriminate familiar faces compared with unfamiliar ones this may be related to the overwhelming importance of relationships in the immediate social group in hominid evolution the implications for facial reconstruction are that close friends or relatives are by far the most likely to recognise the face. Furthermore, it seems we have the capacity to discriminate a familiar face even when a good deal of distortion is evident. For example, Sabir Kassim Kilu, who came from the Indian subcontinent, was recognised from a facial reconstruction deliberately given Malay or Chinese features (Prag and Neave 1997, 34). Forensic reconstructions tend to have relatively expressionless faces. Facial expression is a powerful conveyor of mood and personality (see Russell and Fernández-Dols 1997 and Cole 1998), something that can be exploited in archaeological reconstructions. Psychological studies also reveal the differing roles of shape and colour in distinguishing faces of different sex and of differing geographical ancestry (Hill et al. 1995). Caricature, the exaggeration of features peculiar to a particular face, conveys strong signals (see Enquist and Ghirlanda 1998) that could be exploited in facial reconstruction to increase the chances of recognition.

The methods adopted by portrait painters may also provide a useful basis for the development of methods for presenting facial reconstructions. The left or right eye is commonly centred in the portrait, for example (Tyler 1998). Borkowski (1992) begins an irreverent study of Leonardo Da Vinci's Mona Lisa by observing that the portrait is organised in a classic 'golden triangle'. Further examination leads Borkowski to infer that the Mona Lisa was edentulous, however, and that a small scar below the lower lip may be the result of trauma. The enigmatic smile was, in reality, 'an expression resulting from the loss of the anterior teeth'.

Comparisons of reconstructed faces with photographic images of the individual during life are so far rare. Vanezis et al. (1989) compared the likeness of reconstructions generated using traditional and computerised methods. Haglund and Reay (1991) compared 24 facial reconstructions by nine artists produced from the skeletal remains of 41 victims of the Green River serial killer. The authors found that 'interpretations of the same victim varied greatly' and for those subsequently identified 'resemblance of the facial approximation to the deceased showed considerable variation'.

Conclusions

As a field which straddles the humanities, arts and sciences, facial reconstruction offers a rare opportunity for inter-disciplinary research and cross fertilisation. The prospects for applying computerised 3-D reconstruction in archaeology are considerable. Computerised reconstructions should substantially reduce the cost of facial reconstructions to museums although 'hard copies' are likely to remain costly for the time being. The main application is likely to be in 'hands-on' interactive graphic exhibits, visual displays and virtual museums or in the efficient generation of images which can be finished off by an artist or sculptor. Facial reconstruction offers a route via which the public can be introduced to a more proper understanding of human biogeographical variation, the scientific failure of the concept of race and the fallacy of biological concepts of ethnic origin.

Empathy with people of the past will remain the strength of facial reconstruction in archaeology. The introduction of animation and speech to simulation models to create real 'talking heads' to deliver archaeological narrative is the logical step forward for computerised facial reconstruction.