Last modified: 2011-12-16
Abstract
Research into the relation between space and decoration in Pompeii faces three key obstacles. First, the concept of “the Roman house,” derived primarily from literary texts, has imposed a traditional vocabulary (fauces, atrium, tablinum, oecus, etc.) that forestalls the collection and analysis of spatial data. This means that while we may be able to give precise physical location for a given fresco or mosaic (e.g. VI.15.7-8, room d, north wall, with GPS coordinates), we are not able to place it in conceptual or cognitive space, because the traditional vocabulary is imprecise spatially and generally unsupported by artifact finds, and we have no alternative spatial database (Allison 1993, 2004; Laurence 2007; Riggsby 1997). Second, while print resources like Pompei: pitture e mosaici (PPM) and Häuser in Pompeji (HIP) are invaluable, there is no accessible, searchable database for art in Pompeii, and therefore no way of systematically tracking its relationship to space. Finally, 3D visualizations of Pompeii generally consist of illustrative flythroughs that do not contribute directly to research questions. While 3D technologies have been widely used for conservation and recording, they have not yet had a major role in presenting and interpreting data.
In response to these limitations, the Digital Pompeii Project (DPP) is pursuing a threefold workflow that establishes a statistically derived set of unit and room profiles, links this with an art database designed to facilitate detailed spatial research, and incorporates the game engine Unity as a visualization platform and a tool for spatial analysis in its own right. Focusing in its initial stage on insulae 11, 15, and 16 in Regio VI, DPP combines space syntax with network connectivity analysis. For each unit, visibility, permeability, viewshed, and agent analysis graphs are generated through UCL’s Depthmap; for each room, area, convexity, depth, eccentricity, betweenness centrality, and eigenvector centrality statistics are generated through Gephi. The result is a nuanced, multi-layered spatial profile, separate (at least temporarily) from the traditional vocabulary. Second, DDP is creating a relational database based on images from PPM and HIP. This database records the level of complexity (and therefore labor expenditure) found in a given wall, ceiling, or floor, in terms of perspective elaboration and degree of ornament (cf. “convex boosting” in Weilguni 2011; Fisher 2007). Beyond more obvious features (color, style, mythological theme for central paintings), it also tracks decorative motifs found in vignettes, still lifes, and genre scenes, and their spatial location on the wall (socle, predella, middle zone, upper zone). This permits the evaluation of the distribution of constituent elements in wall painting against the spatial profiles of units and rooms. Unity models are then used to explore how space and art work together in realtime to articulate patterns of movement and stasis. Unity’s voxel-based occlusion culling and AI scripting features are compared with visibility graphing and agent analysis in Depthmap. Finally, analytics in Unity are used to test the movement patterns suggested by spatial profiles against the behavior of human “players” navigating the models.