Visibility analysis in fully 3D spaces, as opposed to 2D and 2,5D spaces, refers to a set of spatial analysis techniques that were introduced and developed by members of the Archaeological Computing Research Group at the University of Southampton for investigating visibility in historic and prehistoric built environments and landscapes.
These techniques draw upon developments on GIS-based viewshed analysis that has been utilised in landscape studies (for example Wheatley 1995, 1996; Fisher 1992), as well as visibility analyses applied in built environments, such as isovist analysis and Visibility Graph Analysis (Benedikt 1979, Turner et al. 2001, Batty 2001). The analysis is performed on digital three-dimensional representations of the spaces under investigation and involves the combination of common functionalities of GIS and 3D modelling software.
The benefits of visibility analysis in 3D spaces over other established methods of visibility analysis applied in 2D or 2,5D (GIS-based viewshed analysis) is that it can explore the visibility of objects of any form and shape modelled in 3D. They can be used to analyse the visibility of surfaces in the built environment, including vertical wall surfaces, and could permit the calculation and analysis of 3D isovists (Benedikt 1979). The methododology developed at the University of Southampton was originally aimed to investigate the visibility of the surfaces of 3D objects, rather than the visibility of open space in between walls, as for example does Visibility Graph Analysis (VGA) . Nonetheless, the same methodology has been recently used succesfully to analyse the properties of open space and identify patterns of co-presence in 3D building interiors, such as Late Antique churches (Paliou and Knight, in preparation) . Also the methodology presented here differs from other methods of visibility analysis employed in 3D spaces (see Bishop et al. 2000; Bishop 2003) in that it can take into account thousands of viewpoints that virtually cover all observer locations in the study area. Furthermore, it offers many possibilities of spatial mapping and analysis of visual information that can potentially explain spatial patterning in the archaeological record.
The analysis is performed in two stages. The first stage, the visibility recording, is carried out using 3D modelling software. It requires firstly the creation of a 3D model of the space of interest. Within this digital environment the visible and non-visible surfaces of a target object from a given observer location are identified using a light source placed at the viewer’s eye level that emits light rays in all directions (Paliou and Wheatley 2007). In that case illuminated and shadowed areas in the 3D environment correspond to visible and non-visible surfaces from the specified observer location respectively. The description and calculation of visible surfaces in a 3D environment using a light source has also been suggested in the past by Benedikt (1979), although for visualisation purposes only.
A reconstructed Late Bronze Age room embellished with mural decoration (Room 3, ground floor, Xeste 3, Akrotiri, Thera)
A view of the wall painting from a camera located at a specified viewer´s eye level (155cm) in Room 3.
Light distribution on the wall surface when a light emitting rays in all directions is positioned at the location of the camera
The textures (raster) of the target object that incorporate information about the illumination of a scene can then be extracted using tools for texture baking that are now built-in in most 3D modelling programs (Earl 2005).
As applies for other methods of visibility analysis in built spaces (for example Visibility Graph Analysis (VGA) , Turner et al 2001), the selection of observer points in the study area is made using a grid (Paliou and Wheatley 2007), whose centroids correspond to the position of a stationary or mobile observer. After the analysis grid has been defined, the visibility characteristics of the target object can be quickly sampled by animating the light source over the grid centroids and extracting the texture that corresponds to each specified observer location. This will result in a set of raster images that incorporate information on visibility, which are similar to the binary viewsheds known from landscape research (Wheatley and Gillings 2002).
A set of binary viewsheds of the wall surface that used to be embellished with the wall painting of the ‘Adorants'(north wall, Room 3, ground floor, Xeste 3, Akrotiri, Thera).
The second stage of the analysis involves a series of map algebra operations performed in a GIS, that aim to summarise information incorporated in the binary viewsheds into a single map. Many types of spatial mapping can be produced in this way:
A) ‘Times seen’, namely maps in which each cell indicates how many times the target object is seen from the areas an observer could have been located. These are similar to ‘times seen’ used in landscape research (Fisher 1992). ‘Times seen’ can describe the likelihood that a particular feature will be encountered by a viewer moving in the space of interest (less obstructed features are more likely to be seen). They can also help to identify invariants in the divergent perspectives of multiple stationary observers, for example during public gatherings. Moreover, ‘times seen’ can be created by summing up only viewsheds from locations that intersect a walking path. This could enable the creation of useful visual summaries that can reveal the most exposed features in the course of particular routes (Earl 2005).
The wall painting of the ‘Adorants’ (north wall, Room 3, ground floor, Xeste 3)
The ‘times seen’ of the ‘Adorants’ from 213 location in Room 3 (Xeste 3, ground floor)(Paliou 2009)
B) Scalar fields that show the changes in the amount of visible area of the target object through space. These are performed by mapping information on visible area which is incorporated into the binary viewsheds back onto the specified observer locations. These maps can give indications of how the visibility of the target objects alters in the course of the observer’s movement, when they are intersected with linear features corresponding to movement paths.
Map showing the visble area of all three figures of the ‘Adorants’ that would be exposed to viewers located in Rooms 3 and 4. The figures would have been fully visible form areas marked in red (Paliou 2009).
C) Visibility maps showing different angular and distance thresholds that could have affected the visual clarity of the target object can also be created (Earl 2005; Paliou 2009)
Incomplete data and error modelling
It cannot be overemphasized that, as in the case of landscape visibility studies (Fisher 1992), in most situations there is some degree of uncertainty as to whether a target feature is visible or not from a particular location. This lack of confidence in the results of the analysis is caused by possible errors in the geometry of the reconstructed historic or prehistoric space under study. Errors in a proposed reconstruction are likely to be caused by imprecise recording of the extant built features on site, or during the digitisation process (e.g. when calibrating site plans). The main source of uncertainty, however, is the partial preservation of the archaeological record and the lack of knowledge concerning features that are no longer preserved. On some occasions assumptions can be made regarding such features with a certain degree of confidence. Even then, however, details of the missing elements, such as the exact dimensions, cannot be known with precision. Ultimately, it has to be acknowledged that a great deal of detail about the geometry of past built structures which may have influenced visibility in space is simply unknowable.
That said, it is important to establish whether uncertainty in a particular context can substantively affect statements and interpretations related to the visibility of the features of interest. For that, probabilistic viewsheds can be created that aim to show the likelihood that a certain feature is exposed to the viewer, and the propagation of possible error in the analysis outputs. This can be done in a number of ways. For example, probabilistic viewsheds can be created by, firstly, summing up visibility maps deriving from alternative reconstructions and then dividing the output by the number of model variations, as suggested by Fisher (1992). The possible errors in ‘times seen’ can be propagated either by calculating the range of difference in the case that maps derived by only two alternative reconstructions need to be compared or, for a greater number of alternative reconstructions, by calculating the root mean square deviation of the cell values of the different possible outcomes from the mean (Paliou 2009). Analyses of this kind could help define the areas of the model that are most affected by lack of data and the degree of confidence in which certain interpretations are made.
Detail plan of the ground floor of Xeste 3 with indication of the position of the pier (in red ellipse) reconstructed by Palyvou (2005, p. 120, fig.166) in Room 3a. The wall painting of the Adorants was located on the north wall of the room (i) b, c) The visibility of the wall-paintings of the Adorants with (c) and without (b) the wooden pier (blue-low visibility/red-high visibility d) the range of difference between the two maps (blue areas are not affected by possible errors/red areas are those that are most affected.The visual exposure of the figure at the left is more affected by the presence of the pier.
The visibility analysis techniques described above have been used for the study of prehistoric spaces as part of a PhD research(Paliou and Wheatley 2007; Paliou 2008, 2009, Paliou 2011, Paliou et al. 2011) carried out at the University of Southampton that aimed to study the visibility and potential social significance of the wall paintings that used to embellish public and private buildings in the Late Bronze Age settlement of Akrotiri (Thera, Greece, ca. 1640 BC). Visibility analysis was used to explore the reception of mural decoration in a visually complex public ritual space (building Xeste 3). The method proved to be particularly useful for examining the relationship between iconographic meaning and the visual exposure of individual pictorial elements in three different painted scenes in Xeste 3: the wall painting of the Adorants, the Crocus Gatherers and the male scene. Through the creation of total viewsheds and maps showing the changes in visibility of each pictorial feature within the study area it was demonstrated that there is a correspondence between visual emphasis in pictorial space with visibility in actual space in all three scenes. In this case the analysis reinforced certain iconographic interpretations of the paintings and highlighted meaningful patterns in the archaeological record that would have otherwise remained unobserved. The results of the analysis were also suggestive of movement and circulation during ritual performances that could have taken place in the building. Furthermore, the same methodology was used to investigate whether pedestrians traversing the street network of Akrotiri could have seen, through open windows, the wall paintings that embellished the interiors of elaborate private houses. The application of visibility analysis offered insights into the social significance and functions of Theran murals, illuminating their possible symbolic role in establishing power relations in the prehistoric society of Akrotiri.
Furthermore, Earl (2005) applied similar methods to assess the impact upon visibility of the removal of a rood screen and balconies between the Georgian and Victorian periods of Holyrood church (Southampton, UK). He also provided examples of visual summaries (maps) that take into account distance and viewing angle.
Current work in progress that is carried out at the Free University in Berlin (Topoi Excellence Cluster) aims to identify and analyse patterns of co-presence in Late Antique ecclesiastic space (using San Vitale (Ravenna, Italy, ca. 548 BC) as a case study), so as to highlight social aspects of the liturgy (e.g. focusing mainly on the relationships between male and female members of the assembly, as well as between the assembly members and the clergy) (Paliou and Knight, in preparation). This research also aims to combine visibility mapping with acoustic mapping that has already been produced by Tronchin et al (2007) and Tronchin and Knight (2008).
Batty, M., 2001. “Exploring isovist fields: space and shape in architectural and urban morphology”, Environment and Planning B: Planning and design 28(1): 123-150.
Benedikt, M., L., 1979. “To take hold of space: isovists and isovist fields”, Environment and Planning B: Planning and design 6: 47-65.
Bishop, D., Ian, 2003. “Assessment of visual qualities, impacts, and behaviours, in the landscape, by using measures of visibility “, Environment and Planning B: Planning and design 30(5): 677-688.
Bishop, D., Ian, R. Wherrett, Joanna and R. Miller, David, 2000. “Using image depth variables as predictors of visual quality”, Environment and Planning B: Planning and design 27: 865-875.
Earl, G., P., 2005. Texture viewsheds: Spatial summaries of built archaeological spaces derived from global light mapping. Paper presented in the 11th International Conference on Virtual Systems and Multimedia, Ghent, Belgium VSMM (2005). October 2005.
Fisher, P., 1992. “First experiments in viewshed uncertainty: simulating the fuzzy viewshed”, Photogrammetric Engineering and Remote Sensing 58: 345-352.
Paliou, E. and D. Wheatley, 2007. “Integrating spatial analysis and 3D modelling approaches to the study of visual space: Late Bronze Age Akrotiri”, in The world is in your eyes – Proceedings of the XXXIII Computer Applications in Archaeology Conference. 21-24 March 2005. Tomar, Portugal: 307-312.
Paliou, E., 2008. “An autonomous agent approach to the investigation of intra-site movement and visibility: The visual consumption of Theran Murals from the public spaces of LBA Akrotiri (Thera, Greece)”, in Proceedings of the Annual International conference of Computer applications and quantitative methods in Archaeology, CAA 2007, Berlin, 2-6, April 2007: p 328-335.
Paliou, E. 2011, “The communicative potential of Theran murals in Late Bronze Age Akrotiri: applying viewshed analysis in 3D townscapes”, Oxford Journal of Archaeology, 30(3): 247–272
Paliou, E., Wheatley, D., Earl, G. P., 2011 “Three-dimensional visibility analysis of architectural spaces: iconography and visibility of the wall paintings of Xeste 3 (Late Bronze Age Akrotiri)”. Journal of Archaeological Science. 38:375-386.
Paliou, E. and D.J. Knight, in press. Mapping the senses: Perceptual and Social apects of Late Antique Liturgy in San Vitale, Ravenna. In Proceedings of CAA 2010, Computer Applications and Quantitative methods in Archaeology, International Conference, Granada 6-9 April, 2010.
Paliou, E., 2009. The visual consumption of mural painting in Late Bronze Age Akrotiri (Thera, Greece): A computational approach to visibility analysis in three-dimensional built environments. PhD Thesis. University of Southampton.
Palyvou, C., 2005a. Akrotiri Thera: An architecture of affluence 3,500 years old. Philadelphia, Pennsylvania, INSTAP Academic Press.
Tronchin, L., Knight, D.J, Tarabusi, V. and Durvilli,L. 2007. “Sound Perception of Choral Music in San Vitale, Ravenna, Italy”, Conference Paper for the International Symposium on Musical Acoustics (ISMA). Venue: Barcelona, Spain
Tronchin, L. and Knight, D.J., 2008. “The Acoustical Survival of San Vitale, Ravenna, Italy Through Two Millennia”-Conference paper for the IOA – Institute of Acoustics Congress on Auditorium Acoustics, Venue: Oslo, Norway.
Turner, A., M. Doxa, D. O’ Sullivan and A. Penn, 2001. “From isovists to visibility graphs:a methodology for the analysis of architectural space”, Environment and Planning B: Planning and design 28(1): 103-121.
Wheatley, D., 1995. “Cumulative viewshed analysis: a GIS-based method for investigating intervisibility, and its archaeological application”, in Archaeology and Geographical Information Systems: A European perspective. G. R. Lock, and Stancic Zoran, ed. London, Taylor & Francis: 171-185.
Wheatley, D., 1996. “The use of G.I.S. to understand regional variation in Neolithic Wessex”, in New methods, old problems: Geographic Information Systems in modern archaeological research. H. Maschner, D.,G., ed. Carbondale, Illinois, Center for Archaological Investigations, Southern Illinois University at Carbondale: 77-103.