The Triclinium of Julia Felix: A Virtual Reconstruction
Introduction
The potential for 3D modelling and visualisation for achieving research and didactic goals in an archaeological framework is astounding. Increasingly, the technology is becoming more accessible and commonplace for its use in archaeology (Kanter 2000; Porcelli et al. 2013).I have fictionally been approached to produce reconstruction images of the Triclinium of the Praedia of Julia Felix for a production company undertaking an hour-long documentary series about the Praedia. This will entail producing a 3D model of the room as it might have been before Pompeii’s devastation in 79 AD in an appropriate 3D program. The end goal for the production team is to create a short animation of the room to inform the audience about the emotion and aesthetics of this room as it might have been in the past. The idea is to put the viewer within the context of the past. Computer-generated animations have become commonplace in popular media as a means of presenting ancient cultures and relevant archaeology.
Background
The world-famous city of Pompeii has provided archaeologists and historians with invaluable information on the lives and times of those living in the city. Within its walls, the Praedia of Julia Felix provided a unique look at a private house converted into a public space. It was a massive and unique complex of baths, shops, gardens and dining rooms on the Via dell’Abbondanza north of the amphitheatre on the western side of Pompeii (Figure 1). |
Figure 1. Map of
Pompeii with important and notable buildings and sites designated. (Source)
|
After a severe earthquake wracked the city in 62 AD, the complex was seriously damaged. Work was undertaken almost immediately to repair and renovate the buildings to provide public spaces and housing for those effected by the quake, to the benefit of its owner, Julia Felix, an affluent property owner who inherited her wealth from her family and astute business operations (Dobbins 2007; George 1986; Parslow 2007).
At a prominent position within the Praedia (Figure 2), a barrel-vaulted summer triclinium and nymphaeum with a water-stair fountain allowed patrons to dine and recline while gazing out onto the well-adorned garden. The water would have pooled around the marble-faced couches and created a beautiful effect within the room (George 1986).
 |
Figure 2. Plan of the
Praedia of Julia Felix. The triclinium is room F.
(Source) |
Aims & Objectives
The aim of this reconstruction is to recreate the visual perception of the room illuminated with natural lighting and the emotions evoked through the personal experience of being within the room as it might have been prior to its devastation in 79 AD. More specifically, I would like to recreate the effect of the water pooling at the outer edge of the triclinium and the caustics generated by the natural light which kept the room illuminated during the summer months.This will require a detailed geometric model, meticulous surface materials and textures, realistic spectral representations within the room and several renderings of the model with physically correct lighting. As my audience will primarily be those who may only possess a (suspected) cursory knowledge of Pompeii, Roman dining customs, triclinia or nymphea, I intend to construct my model as accurately as possible.
However, some compromises must be made and these will be described to dissuade the false assumption of ultimate veracity that often accompanies 3D reconstructions, as they tend to convey a strong suggestion of reality no matter the quality of data on which they are based (Kanter 2000). Ultimately, an animation will be story-boarded from rough sketches and render from which the production company can work.
Metadata
Appropriate metadata is essential for any project for its maintenance and dissemination. Accordingly, project-level and file-level metadata have been attached at the end of this report as Appendix B. All 3D files have been saved both as a 3D Studio Max Design File (.max) and Wavefront Object File (.obj), a standard open file format which is not software dependent.3D Modelling in Archaeology
3D data is an increasingly important visualisation and analytical tool within the field of archaeology. Virtual archaeology sprang from “a melting pot of computer science and archaeological aspiration” (Earl 2009). Although its considerable advancement over the decades has facilitated analysis and research goals, it is not without constraints and underlying interpretive problems. These will be outlined in the following section.There are many benefits to creating 3D models of archaeological data. It allows archaeologists to explore a particular research question or predict the behaviour or perspective of an object or feature which has been damaged. The ability to transform archaeological data collected in the past into novel interpretive and portable formats of elaborated data (as I have done with this project) is an invaluable tool for retrospective research. Above all, perhaps, is its innate ability to facilitate interaction between the data and its human users by producing an interactive virtual tool for analysis and interpretation (Hermon 2008; Papadopoulos & Earl 2009).
However, the creation of 3D models in archaeology remains necessarily constrained by certain facts within the practice. Firstly, that archaeological datasets are seldom, if ever, complete and a certain amount of conjecture or, more favourably, imagination is inevitably introduced into every 3D model produced for archaeological purposes (Barceló 2002; 2010; Earl 2009; Kanter 2000). Even the most complete dataset will never present a sufficient reconstruction of the past (Papadopoulos & Sakellarakis 2010). (This position, of course, excludes laser scanned data of extant sites or objects which is often accurate within the sub-millimetre range.) It is for this reason that it is essential that descriptions of the primary data are included, in addition to any interpretive leap that might be made to build a model (Eiteljorg 2000; Kanter 2000).
Perhaps more theoretically, it has been argued that 3D data favour the visual aspect of any site or object, at the cost of the other senses (Eiteljorg 2000; Gillings & Wheatley 2001, Hamilakis 2013). Luckily, we do not live in a world dictated purely by what we see, but rather one that is affected by all the interactions of our senses in fluctuating combinations. 3D visualisation inherently ignores all other senses.
These constraints and others, such as the goals of a particular model and its intended audience, affect the way in which we model archaeological data in a 3D environment. The levels of realism portrayed in virtual environments must always be considered and described.
3D Studio Max as the Tool
Virtual reality is particularly dependent on the software and hardware used in its creation (Papadopoulos & Earl 2009). For the completion of my model, I have chosen to utilise Autodesk’s 3D Studio Max 2014 as the primary program for construction. By using geometric primitives, and subsequently adding materials, textures and lighting, a 3D model is thus produced. It allows considerable flexibility in model creation and excellent rendering capabilities with even the most basic package renderer.Model Creation
The general workflow for the creation of the resulting model can loosely be represented by Figure 3, although no excavation data was used. The key point to mention here is the minor degree of imagination is essential is reconstruction because an archaeological dataset is seldom complete. |
Figure 3. Schematic Diagram of 3D modelling stages. This
figure shows the typical stages of 3D modelling and its basic components.
(Courtesy of Hermon 2008)
|
Data Sources
My primary data source was Michelle George’s dissertation on the nymphaeum from 1985 at McMaster University. This work included three meticulously detailed section drawings of the room from which I derived the measurements for the geometry of my model. Of course, data uncertainty is always an issue in any archaeological study.Unfortunately, I had to make a considerable compromise in modelling my geometry in consideration of time constrains. I modelled half of the required geometry with the intention of mirroring the structure, thus creating a complete model. Unquestionably, this has introduced errors into my resulting data as nearly nothing in human construction is impeccably symmetrical, especially a building that is over 2000 years old. This unearths another inherent problem with modelling a reconstruction of a room. Although careful attention was made to model accurately, by ‘turbosmoothing’ my model, I have created a perfectly and uniformly smoothed model. It may look good, but I have knowingly introduced a certain amount of error.
I based the materials and general feeling/evocation of the room on my own personal experience (Figure 4) of visiting the site and photographs from the internet.
 |
Figure 4. A picture of me in the room taken in
2011.
|
Numerous books, tutorials and descriptions of models found on the internet were essential for the creation of this model (Bekerman 2011; Brinck 2005; DiClementi 2007; DM Multimedia 2008; Master Zap 2013; Nagakura 2011; Pabico 2011; ScratchPixel 2012).
Geometry/Meshes
A highly-detailed geometry is the first step in the 3D reconstruction process. An effort was made to create a reconstruction with as few errors as possible. Quadrangular topology was favoured for modelling as this tends to be more stable than triangular counterparts.Once a very basic model was obtained, the geometry was made more complex with the intention of adding a turbosmooth modifier on the interior. This was accomplished with swiftloops to achieve a subtle curvature to the model, as nearly nothing in reality is made up perfect ninety degree angles. This added a subtle yet significant degree of realism to the model (Figure 5).
 |
| Figure 5a. A comparison of the model without turbosmoothing applied. |
 |
| Figure 5b. A comparison of the model with turbosmoothing applied. |
The final polygon count for the model is 46,684.
Materials
As the materials were one of the most important aspects of my work to achieve my aim, I primarily focused on them in the creation of my model. I used Arch & Design shaders for all my materials.I intentionally used BRDF by Index of Refraction instead of manually setting facing and perpendicular reflectivity to ensure a realistic reflection falloff was obtained. The decision was made to map the materials visible to the cameras using image-based texturing instead of procedural texturing (Ebert et al. 2003). This allowed for a more accurate representation of the materials as they might have looked.
In adding materials to the walls, the decision was made not to recreate the frescos which might have been on the walls. The entire Praedia was effectively looted upon its discovery in 1755 and it is unknown which frescos ornamented the walls (Parslow 2007). In addition, the condition of most frescos today, having been removed from their walls several hundreds of years ago and hung in museums, are severely degraded. Cracking, fading and warping are all effects which would be irreparable, even if I were to hazard a guess as to which frescos were present in the room. Instead, a red painted plaster material was chosen to represent the walls, which is known to have adorned the walls, albeit not exclusively. This was a severe compromise, although one for which I could not find a solution.
As the room was primarily used in the summer months, it is unlikely that anything other than natural light illuminated the room. Thus the built-in photometric 3DS Max daylight system was set up. As the system cannot calculate the position of the sun before 1583 AD, it was arranged at the pre-set time of 15:00 on the summer solstice (June 21) in the year 2014 AD. Sky portals were also aligned with the two openings in the room, the main entrance and a skylight above the water-stair fountain on the opposite wall. This directed the photons created by the daylight system through these openings.
Rendering
The rendering process was comprised of extensive trial and error. Through several tests, I was able to render out an acceptably realistic image of my model and its materials, textures and lighting, including caustics.Mental Ray
Mental Ray was the rendering engine selected for the rendering of this model. This is the package renderer that is included with the 3DS Max software and it yielded excellent results for its relative ease of use and speed (O’Connor 2010).Caustics
Caustics, the way light rays are reflected or refracted by a curved surface or object, such as water, was enabled in my rendering. Caustics add a great deal of realism to my scene. However, it took additional setup, a high number of caustic photons (1,500,000) and very long render times to produce a satisfactory solution.Anti-Aliasing
As there are several types of filters for anti-aliasing, I’ve chosen to use the Mitchell filter as this is often the most accurate (according to Grant Cox). This seems to have produced the crispest version of my render yet.Post-Production
The renders produced were still somewhat flat and the shadows not as defined as I would have liked (Figure 6). Thus, the decision was made to create an ambient occlusion pass within 3DS Max and incorporate it into the original render using Photoshop's multiply feature (Figure 7). |
Figure 6. The final render output from 3DS Max. As one
can observe, the shadows in places such as the shrines and the water-stair
fountain are falling flat and do not look realistic.
|
 |
| Figure 7. Ambient Occlusion Pass rendered using Mental Ray in 3DS Max. |
Results
The virtual reconstruction of the triclinium of Julia Felix offers valuable information about the visual perception and aesthetics of the room as it might have been in the past (Figures 8, 9, 10 & 11). For its purpose, I believe my reconstruction met my goals, at least superficially. However, the model and its subsequent renders are too pristine and crisp. There is no accumulated stains, dust, scratches and objects or mildew in the water or the water-stair fountain from everyday use. Nor is there background noise from dust or haze in the model. Admittedly, due to time constraints, I was unable to render the model with these effects. The final renders would have taken on a more believable realism from them real-world effects. |
Figure 8. The final render looking into the room from the
entrance.
|
 |
Figure 9. A final render looking into the room from the
inside, as if standing near the edge of the couch.
|
 |
Figure 10. A final render of the room looking toward the
water-stair fountain.
|
 |
Figure 11. A final render of the room as if sitting on
the right couch of the triclinium.
|
Conclusion
The potential for archaeological analysis of sites and objects using 3D virtual modelling is exhilarating. The purposes of this project were to recreate the room as it might have been with the water pooling around the triclinium to understand the general ambience and visual evocation of the room using caustics from the water. In this regard, I think I have succeeded in doing so.Luckily, we live in a world where technology is constantly changing, advancing and ameliorating the way we see archaeology. The prospects for future research are unknown, yet tremendously optimistic.
Bibliography
Barceló,
J.A. 2002. Virtual Archaeology and Artificial Intelligence. In Nicolucci, F.
(ed) Virtual Archaeology. BAR International Series S1075, pp. 21-28.
Oxford: Archaeopress.
Barceló,
J.A. 2010. Visual Analysis in Archaeology: An Artificial Intelligence Approach.
In Elewa, A.M.T (ed) Morphometrics for Nonmorphometricians. Lecture
Notes in Earth Sciences 124. Berlin: Springer-Verlag.
Dobbins,
J.J. & Foss, P.W. 2007. The World of Pompeii. New York: Routledge.
Earl,
Graeme. 2009. Physical and Photo-realism: The Herculaneum Amazon. In Archeologica
2.0, Seville Spain, 16-19 Jun 2009.
Ebert, D,
Musgrave, F.K, Peachey, D, Perlin, K. & Worley, S. 2003. Texturing and
Modeling: A Procedural Approach. San Francisco: Elsevier Science.
George, M.
1986. The Triclinium-Grotto of Julia Felix: The Grotto in Roman Domestic
Architecture. Digital Commons @ McMaster University.
Gillings, M.
& Wheatley, D. 2001. Seeing is Not Believing: Unresolved Issues in
Archaeological Visibility Analysis. In Slapšack, Bozidar (ed) On the Good
Use of Geographical Information Systems in Archaeological Landscape Studies:
Proceedings of the COST G2 Working Group 2 round table, pp. 25-36. Office
for Official Publications of the European Communities: Luxembourg.
Hamilakis,
Y. 2013. Archaeology and the Senses: Human Experience, Memory, and Affect.
Cambridge: Cambridge University Press.
Hermon, S.
2008. Reasoning in 3D: A Critical Appraisal of the Role of 3D Modelling and
Virtual Reconstructions in Archaeology. In Frischer, B. & Dakouri-Hild, A.
(eds) Beyond Illustration: 2D and 3D Digital Technologies as Tools for
Discovery in Archaeology. Archaeopress.
Kanter, J.
2000. Realism vs. Reality: Creating Virtual Reconstructions of Prehistoric
Architecture. In Barceló, J.A, Forte, M. & Sanders D.H. (eds) Virtual
Reality in Archaeology. Bar International Series 843, pp. 47-52. Oxford:
Archaeopress.
O’Connor, J.A. 2010. Mastering
Mental Ray: Rendering Techniques for 3D and CAD Professionals. Hoboken:
Sybex.
Papadopoulos,
C. & Earl, G. 2009. Structural and Lighting Models for the Minoan Cemetery
at Phourni, Crete. In Debattista, K, Perlingieri, C, Pitzalis, D. & Spina,
S. (eds) Proceedings of the 10th VAST International Symposium on Virtual
Reality, Archaeology and Cultural Heritage.
Papadopoulos,
C. & Sakellarakis, Y. 2010. Virtual Windows to the Past: Reconstructing the
‘Ceramics Workshop’ at Zominthos, Crete. In Contreras, F. & Melero, J.
(eds) Proceeding of the 38th CAA Conference.
Parslow, C.
2007. Entertainment at Pompeii. In Dobbins, J.J. & Foss, P.W. (eds) The
World of Pompeii. New York: Routledge.
Porcelli, V, Villa, F.C, Senabre, J.B, Torres,
V.E. & Chapapría, J.E. 2013. Integration of 3D Laser Scanning and Virtual
Reconstructions as Research and Educational Tools for Representing the Past:
The Case Study of Roman Baths of Edeta. In Earl, G, Sly, T, Chrysanthi, A,
Murrieta-Flores, P, Papadopoulos, C, Romanowska, I. & Wheatley, D. (eds) Archaeology
in the Digital Era. Amsterdam: Amsterdam University Press.
