The quest to reconstruct al-Qusur in 3D (2/2): how did we create the model?

During MAFKF 2023’s excavation campaign, a 3D model was produced representing three buildings from the al-Qusur’s settlement: the church (building A1), the refectory (B23), and what was most probably a monk’s cell (B16). Each one of these had been the object of archaeological excavations during the past years, the first one was excavated in 1988-1989 and 2007 by the French Mission in Kuwait and the last two were excavated in 2012-2015 by the French-Kuwaiti Archaeological Mission in Failaka. In order to recreate the context of these buildings and enable the understanding of their location in the site and their relation to each other, the whole settlement also had to be modeled, although with a lesser degree of detail and accuracy. This modeling process was based on the master plan developed from 2011 to 2019 by Jean Humbert through surface surveys and aerial photography (Humbert 2020). Despite the accuracy of the plans drawn for these three buildings, it was still necessary to extrapolate their elevations as only a limited amount of material had survived on the ground, the wall remains being no higher than 50 cm. In the current paper, the second and last of this blog post series, we will focus on the methodology which was used to recreate these buildings and the results that were obtained.

Structural logic

The different estimated heights for the model’s walls were based on direct observations, dialogue with specialists, and architectural logic (e.g. the type of roofing, the structural strength of mudbrick walls etc.). We also started from the principle that an average-sized person would have been able to stand anywhere within these buildings.

Regarding structural logic, mudbrick is a very versatile material that makes it possible to create different shapes such as arches and pillars. However, it was taken into consideration that the average height to thickness ratio for mudbrick walls is 7/1 and that using this logic the walls’ height could not be greater than seven times their thickness.

Even though no windows have been preserved, these were added following the logic that the people using the buildings would have needed ventilation and a minimum amount of light. As no material remains such as window glass have been discovered on the site, this is also coherent with the hypothesis that the openings were also used as vents. Their size was decided taking the constraints of the Gulf environment into consideration, i.e. the heat and dust meant that large windows were highly unlikely, so small openings were added to the model.

The current lack of wood on the island and the lack of any physical evidence of its utilization raised questions about its use as a construction material in these buildings. However, comparisons with traditional construction in the Gulf area and structural logic meant that we believe it is highly probable that it was used. Our hypothesis is that mangrove and palm wood were used for the woodwork, as studies have shown that these used to be common in this area (Willcox 1990). Many of the fish bones which were found during excavations are related to species which populate mangrove areas and these may have been fished from Failaka’s shores. Date stones have also been discovered on the site, suggesting that palm trees also grew there previously. There is also evidence that reeds were present thanks to an imprint on a thick earthen coating which was discovered in building B16 (Bonnéric, Herviaux, and Rivoal, forthcoming). Following structural logic, it is almost impossible to imagine the buildings’ roofs without some sort of rudimentary wooden carpentry. The hypothesis of arched structures was considered, but these were deemed to be too heavy for the walls that were present on the site (except in rare instances, like in the church).

Existing traditional buildings were of great help for recreating the elevations of the buildings, as the mudbrick technique is still in use in the region (fig. 1). This gave us examples of elevations, windows, gutters, and roofing systems that were coherent with our structural hypothesis. This meant that the evidence was sufficient to allow us to recreate the elevations with a good probability of realism.

Fig. 1. Photograph of Umm-Salal’s Mosque, Qatar (source: https://www.alamy.com).

Comparison with other buildings in the region

When several structural hypotheses were present or when no evidence allowed us to imagine what the construction could have been like, we had to refer to other buildings in the region. It was difficult to find examples which perfectly corresponded to the buildings we were looking to model as no complete mudbrick churches have been found in the Gulf region: however some examples of stone-built churches do exist and these gave us clues about their general appearance, in particular the one on Kharg Island. Contrary to the previous model created in 2016, we did not consider that the evidence for the inclusion of a cupola was sufficient and thus we decided not to recreate it in the new model. This is because the traces which have been interpreted as a cupola were found in al-Kharg and no similar evidence has been found so far in al-Qusur.

As the sample of Christian cult buildings studied in the region is very limited (only a dozen Christian sites have been identified, out of which only five churches have been investigated: al-Qusur, Akkaz, Sir Bani Yas, Jubayl and Kharg, the latter being the only one which have to be recreated with a 3D model), it is particularly difficult to identify building types and classify them (fig. 2). However, what can be observed is the similarity in spatial structure between al-Qusur’s and Kharg’s churches which suggests that they most probably had the same spatial organization (fig. 3). This East-Syriac layout is characterized by a narthex preceding three naves, leading to a square apse framed by two square annexes, and a straight wall which terminates the apse, that is different from the semi-circular wall of West-Syriac apses. Potentially closed apses and side aisles, separated from the central nave (traces of door frames have been found in the wall openings), are very typical however it is difficult to say whether this type of layout was due to liturgical or climatic reasons.

To widen the sample, examples from the southern Mesopotamia area can also be used (fig. 2), especially given their morphological similarities with al-Qusur’s church which have been highlighted by K. De Langhe (2008) and J. Bonnéric (2016, 2023). For example, the Ayn Shay‘a church has a very similar layout. This can be explained by the fact that the Church of the East founded the majority of the early Christian sites in the Gulf. However, unfortunately, as in al-Qusur, none of these churches contain any complete surviving elevations or roofs so this means that even comparisons with other buildings present a number of significant limitations.

Fig. 2. Map showing pillar-wall and flat-apse churches and churches of other types in Southern Iraq and the Gulf region (J. Bonnéric, background map H.David-Cuny, 2021).

Fig. 3. Masterplans of pillar-wall and flat apse churches discovered in southern Iraq and the Gulf: Sir Bani Yas, Ain Sha’ia’, Kharg, al-Quṣur, tell V in al-Ḥīra (J. Bonnéric, 2020, based on Elders 2001, fig. 2; Talbot Rice 1932, fig. 2; Fujii et al. 1989, fig. 4; Steve 2003, pl. 12).

Form, function, and purpose

It is important to remember that the act of building, at a time when infinite energy was not readily available, was complex and time-consuming meaning that purposeless work would generally have been avoided. As a consequence, we should consider that everything present in the building is meaningful and had a specific purpose, even if this purpose eludes us.

To be able to recreate a space it is necessary to understand its function and use. This is especially true in the case of a church, as not all spaces are equal in importance, whether they are liturgical or functional. Indeed, even if it might sound obvious, a church’s narthex and choir do not have the same value. Thus, the emphasis must be put on the heart of the building and to do so one must understand the liturgy which was followed and the possible uses of the various spaces (Lassus 1950). This hierarchy is easy to identify for a church as it is the same as in most Christian buildings where the apse is the central focus of the ceremony. Consequently, we can create a hierarchy between the diverse spaces which can be summarized as follows: apse, central nave, lateral apses, collaterals, narthex, and southern annex.

This hierarchy is illustrated by the floor height of each space in al-Qusur’s church: the central apse is situated more than a meter above the southern extension and 80 cm above the narthex. For the purpose of adhering to this hierarchy and taking into account the floor’s altitude, the render adjusted the height of each space’s elevation.

The landscape

Backgrounds play a significant role in giving a sense of verisimilitude to a render. As accurate as a 3D render could be, if it is placed on a generic background this would immediately make it awkward. To address this issue, it was essential to choose a background that fits with what we know about the site. Unfortunately, little is yet known about Failaka’s vegetation during the medieval period. Some studies have been conducted on its vegetation during the Bronze Age and Hellenistic periods, but it is difficult to assert with absolute certainty that this did not evolve over the centuries. Palaeobotanical studies that may answer these questions are still on-going.

It was thus necessary to choose what vegetation to use and how to render the background, balancing between the need for accuracy and what scientific knowledge tells us. A balance was found by picking some of the plants that are still present nowadays and mixing them with others that have been identified as being present in the past and which are still fit for the current climate (Willcox, 1990). This resulted in the selection of a large number of palm trees, sebkha vegetation, and a few tamarix and jujube trees.

When recreating the environment of a building that no longer exists, one is confronted with the limits of images. Indeed, all of the pictures of the site taken during the studies were mainly for archaeological use and lacked depth. They show the elements of interest, sometimes with some context, but never the site as a whole. Moreover, even though some aerial photos exist, the landscape has changed with time and human occupation. It is now almost impossible to obtain a picture that could come close to what the landscape looked like a thousand years ago.

To meet these constraints, we experimented with the use of AI-generated images. Some free software allows for the creation of ‘new’ pictures by mixing old ones together, following a written requirement. Depending on the precision of the input, the resulting image can come close to what is required. Unfortunately, as this technology relies on existing pictures it cannot create something that no longer exists. The result always looks like a copied version of other landscapes, with some unreal touches (e.g. errors in the shadows, a blurry forefront, repetitive elements, and errors in complex objects like trees, etc.). For instance, when trying to recreate a flat desert island background, with sparse shrubland and sebkha (a word that the software did not understand), all of these software programs proposed views that were reworked aerial pictures of the Mojave Desert in Nevada (fig. 4). Moreover, there is a real ethical concern about this type of software as it uses pictures to mix and create new ones without the consent of their creator or owner. The aesthetical and ethical limitations of this type of background meant that in the end more traditional redesigned backgrounds were used.

 

Fig. 4. Top: Current aerial view of the Mojave Desert, source: aerial stock footage of a bird’s eye view of a flat plain in the Nevada Desert (Axiom Images, footage FG0001_000271). Bottom: AI-generated aerial view of a flat desert shrubland landscape—the similarity with the previous picture is striking.

Modeling, rendering and photo editing

The buildings and site were rendered using three software programs: Revit (fig. 5) and Sketchup for modeling and Artlantis for rendering. Modeling was carried out to scale on plans imported directly into the software, and textures were worked out from photographs (fig. 6). The 3D aspect of the modeling made it possible to quickly test architectural hypotheses before model finalization. The ability to move a camera according to the desired angle of view and to render orthographic or classical perspective views meant that as many illustrations as necessary could be produced, as well as video animations.

Fig. 5. 3D model of a building on Revit (Coppé, 2023).
Fig. 6. Top: A more accurate 3D model of the monumental church A1 made using Sketchup (Coppé 2023); Middle: The 3D rendering created by Artlantis (Coppé, 2023); Bottom: The final 3D image after the application of more detailed textures and photo editing with Photoshop (Coppé, 2023).

Once the 3D render had been produced by Artlantis this did not mean that the model was completely finished. Even after this step, the textures needed to be reworked, as this software uses seamless patterned textures (i.e. continuous and repetitive textures, with no apparent gaps or spaces between one part and the next) so they tend to be too repetitive when used over large surface areas. For this reason, it was also necessary to use Photoshop to add more complexity to the textures and make them more realistic. Moreover, the Photoshop process makes it possible to add context (e.g. trees, background, and silhouettes) directly from photographs, giving a sense of realism. The images were populated following the hypotheses about the population and type of animals present on Failaka at the time, which suggest the presence of sheep, goats and donkeys (Bonnéric 2022).

To help understand their purpose and scale, the interiors were also meticulously reconstructed and human outlines were added to them (fig. 7). According to the French mission’s excavation reports and publications (Bonnéric 2016) it seems that the interior walls and flooring were all completely coated.

Excavations carried out in 1988-1989 and 2007 by the French Mission in Kuwait revealed some stucco fragments and panels in the church which were mostly located in the southern nave and south chapel. However, these fragments have not yet been reproduced in the 3D model despite this evidence that the inside of the building was decorated (Bonnéric 2016). This is because their specific location, distribution, height on the wall, and total length are very hypothetical. Due to this lack of accuracy and the confusion that it would introduce into the reconstruction it was therefore deemed more correct not to incorporate them into the model. However, the proposed restitution is not definitive and will evolve in line with the continued scientific research into the church of al-Qusur.

Fig. 7. Render of the refectory’s interior (Coppé, 2023).

Conclusion

3D rendering is a powerful tool for popularizing archaeological discoveries, as it allows non-experts to easily comprehend a hypothetical past state of a site. However, the other side of the coin to this didactic aspect is that it fixes an image of buildings in the public’s mind so it could be difficult for them to accept any later changes to their appearance. This means that this technology must be used carefully and be evidence-based. Nevertheless, if it is used with care, following scientific methods, the results can open new fields for researchers.

3D rendering is an important tool for archaeologists to use, as it represents a continuation of their analysis work, allows for experimentation and also confronting different theories. The constant evolution of this software, which is becoming more and more realistic while also becoming more user-friendly, means that the scientific community cannot ignore it. Therefore, scientific methodology for using it should be created and widely shared among the scientific community. The use of artificial intelligence has to be taken into account in this methodology as this tool, that was in its infancy only a couple of years ago, is now reaching a point where it cannot be overlooked as it could represent a great opportunity for future modeling applications.

Bibliography

Bonnéric J., Al‑Qusur, a Christian Monastery on Failaka Island, Kuwait. An Archaeological and Historical Guide Book, Kuwait City, 2016.

Bonnéric J., “Why islands? Understanding the insular location of Late Sasanian and Early Islamic Christian monasteries in the Arab-Persian Gulf”, in: C. Durand, J. Marchand, B. Redon & P. Schneider (eds.), Networked spaces: the spatiality of networks in the Red Sea and Western Indian Ocean, Lyon, 2022, p. 193-210.

Bonnéric J., “Des églises monastiques de pèlerinage ?”, Bulletin d’études orientales, LXVIII,  2023, p. 95-119.

Bonnéric J., Herviaux G. & Rivoal M., “Al-Quṣūr’s Area A5: a possible cell amongst the findings north of the monumental church and its surroundings”, in J. Bonnéric (ed.), Al Qusur, a Christian settlement from the Early Islamic Period on Failaka, vol. 1. Excavations of the French Kuwaiti Archaeological Mission in Failaka (2011-2018), Kuwait City, (forthcoming).

Coppé E., “The quest to reconstruct al-Qusur in 3D (1/2): aims, previous work and issues”, Le carnet de la MAFKF. Recherches archéologiques franco-koweïtiennes de l’île de Faïlaka (Koweït), 10 octobre 2024. [En ligne] https://mafkf.hypotheses.org/2196

De Langhe K., “Early Christianity in Iraq and the Gulf: a view from the architectural remains”, In Actas del V Congreso Internacional de Arqueología del Oriente Próximo Antiguo, Vol. 1, 2008, p. 603–610.

Humbert J.,Mapping of al-Quṣūr Village”, Le carnet de la MAFKF. Recherches archéologiques franco-koweïtiennes de l’île de Faïlaka (Koweït), 17 avril 2020. [En ligne] https://mafkf.hypotheses.org/2019

Lassus J.,Liturgies nestoriennes médiévales et églises syriennes antiques”, In: Revue de l’histoire des religions, 137/2, 1950. p. 236-252.

Willcox G. “The plant remains from Hellenistic and Bronze Age levels at Failaka, Kuwait. A preliminary report”, in Y. Calvet & J. Gachet (eds.), Failaka, fouilles françaises 1986-1988, Lyon, 1990. p. 43-50. [En ligne] https://www.persee.fr/doc/mom_0766-0510_1990_rpm_18_1_1750

Etienne Coppé

Architecte du patrimoine à l’Ifpo Beyrouth depuis juillet 2022, Etienne Coppé a travaillé dans divers pays méditerranéens. Il participe aux activités de recherche portant sur l'architecture, l'archéologie et le patrimoine, à travers des travaux de relevés architecturaux, d’analyses architecturales et de restitutions. Il conduit également des travaux de conservation et mise en valeur du patrimoine archéologique.

More Posts


OpenEdition vous propose de citer ce billet de la manière suivante :
Etienne Coppé (14 octobre 2024). The quest to reconstruct al-Qusur in 3D (2/2): how did we create the model? Le carnet de la MAFKF. Consulté le 16 mars 2025 à l’adresse https://doi.org/10.58079/12hm7


Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

This site uses Akismet to reduce spam. Learn how your comment data is processed.