Digital historical reconstruction: case studies of an interdisciplinary task Eng. Arch. Han Vandevyvere, CAAD, Design & Building Methodology research group, ASRO, K.U.Leuven (B) Prof. Dr. Eng. Arch. Herman Neuckermans, CAAD, Design & Building Methodology research group, ASRO, K.U.Leuven (B) Prof. Dr. Eng. Arch. Krista De Jonge, Architectural History and Conservation research group, ASRO, K.U.Leuven (B)
Address: K.U.Leuven, Department of Architecture, Urban Design and Planning Kasteelpark Arenberg 1 B-3001 Leuven Belgium
E-mail:
[email protected] [email protected] [email protected]
Tel: ++ 32 16 32 13 61 Fax: ++ 32 16 32 19 84
Correspondent: Han Vandevyvere
1
Digital historical reconstruction: case studies of an interdisciplinary task Abstract The present paper considers the interdis ciplinary process of creating a digital historical building reconstruction. On basis of the study of a few completed reconstruction cases to which the authors have participated, a review is made of the opportunities, the difficulties and the sometimes unexpected outcomes of such a project. Theoretical considerations have been developed during the project work, building on the acquired experience. These considerations are now reflected in the formulation of a general methodological framework for digital his torical reconstruction projects. As a result from our experiences, we can argument that the use of digital media in historical reconstruction does not merely mean applying a new tool to perform traditional historical approaches. Digital modelling indeed introduces a new competence into historical research, apart from having a strong representative power as such. It has the potential to provoke new insights in the subject at study, and thus to contribute significantly to the understanding of a building’s history.
Key words: digital historical reconstruction, multimedia, CAAD, building history
2
1. Introduction In 1998, the CAAD, design & building methodology research group at the architecture department of K.U.Leuven got the commission to create a multimedia presentation on the construction history of the gothic town hall of Leuven (figure 1). The digital presentation was to be part of a large exhibition on the building and its history, organized at the occasion of the 550th anniversary of its construction.
Figure 1: the town hall of Leuven in an early photograph of 1857
The creation of this multimedia presentation turned out to become a fascinating interdisciplinary journey. Continuing along the same lines, other reconstruction projects were launched within the group, in collaboration with the architectural history group. In what follows, some of these reconstruction projects are looked at in detail, and discussed from a more fundamental point of view.
2. The trigger project: Leuven town hall reconstruction The task of representing the building history of the Leuven town hall, and in particular of its ‘front house’, could be situated at one end of a scale regarding the level of historic certainty. Indeed, the front building as it stands today has undergone few changes since its early times and extensive 15th century archive resources document extraordinarily well its construction process. By consequence, reconstructing the building process could here be done almost to perfection.
3
On the other hand, there was considerable mystery on the pre-construction phase. Furthermore, the digital reconstruction included the evolution of the surrounding city-centre, which as a whole was less well documented. Altogether this fostered a very complete and diversified research challenge for both historians and CAAD members of the project team. The project as a whole took more than a year to complete.
The work scheme can be summarized as follows.
First, a database was created for the management of all available source material. Considered for inclusion were data on the town hall complex, the neighbouring gothic St.-Peter’s church, important buildings around such as the meat hall and the clothing hall (the later university hall), and the urban tissue of the city centre in more general terms. This source material could be divided into archive documents (in particular the material bills), literature, iconography, building surveys and the actual buildings as they stand. It was decided that the multimedia presentation would focus on a few carefully selected time sections. These coincided with key moments in the construction history. By conclusion the hall and its surroundings would be virtually represented in 1400, 1450, 1500, 1680, 1870 and 1998. Subsequently, for each of these time sections, working drawings were composed by hand to serve as a first synthesis document (figure 2). They consisted of an annotated plan and perspective drawing, regrouping graphically a maximum of elements from the database. As far as hypotheses on certain elements or buildings were necessary, building up the reconstruction model was done in close consultation with the historians of the team.
4
Figure 2: Hand drafted documents for synthesising spatial information: annotated plan and perspective (exhibition catalogue variants without text). The situation as in 1450. N° 2 in the right hand figure is the town hall. (Drawings by Han Vandevyvere)
Upon completion of the data collection and the working drawings, different members of the CAAD-group started to create partial 3D computer models using AutoCAD. Next , composite models were assembled in 3Dstudio. Static renderings and animation sequences for different end goals were then generated from these global models. Final exhibition output could now be prepared in different forms (figures 3, 4, 5): •
Static rendered images for display on exhibition panels ;
•
Two physical models of the ‘front house’ produced from the digital model. The maquettes were created by extracting an stl-file from the AutoCAD model and processing it in a rapid prototyping machine. This work was done twice. A nylon model was produced by selective sintering at the mechanical engineering department of K.U.Leuven; and a resin model was made by Materialise, a spin-off company of the university, using stereolithography. Both models were built in several parts, first of all because the complete models at reasonable scale were too big to fit in the stereolithographic device, and second, because each model should have detachable parts to show its interior layout and roof structure;
5
•
An interactive website that tells the story of the building process, progressing from time section to time section. Major visual components were static rendered images and animations. The whole was completed by text ual explanation in a user-friendly layout.
•
An ‘eye-catcher’ video of about 8 minutes, providing a multimedia historical overview. Here, additional computer animations were combined with other visual source material and carefully selected contemporary music fragments. ‘Real’ video images showed, among others, a reenacting of the extraction of stone from the quarry, its transportation by cart and boat and its sculpting. Wartime photographs and footage showed how the building narrowly escaped destruction on several occasions during the First and Second World Wars. Final video assembly was done by the specialized audiovisual services of the university.
As there was also a certain knowledge about formerly unrealized plans for the town hall and the St.Peters’ church, the reconstruction hypotheses for these non-built scenario’s were modeled as well and presented to the public in a transparent, ghostly red appearance. Such a ‘visual key’ could immediately make clear these images represented something that had been thought of, but never built.
Figure 3: static images of the city centre in 1425 and in 1480. In that period, the town hall was finished while the new gothic St.Peters’ church, built over the existing Romanesque church, was nearing completion. A completion that finally never took place. (Renderings by Kris Nuyts)
6
Figure 4: hypothetic model of the first project by building master Sulpitius van Vorst, inspired by the contemporary proposal for the town hall of Brussels. Red color and transparency underline the hypothetic character of the simulation. (Rendering by Kris Nuyts) Model of the actual front house, designed by Matheus De Layens, rendered transparently to show the underlying roof structure. (Rendering by Han Vandevyvere)
Figure 5: stereolithographic model of the hall: assembled and partially laid out. (Model by PMA at K.U.Leuven)
It is worth noting that the early, hand drafted working drawings were considered for inclusion in the exposition catalogue – be it in a ‘cleaned up’ version. The complete exhibition materials resulted thus to be a mix of analogous and digital information, although with a heavy focus on the digital media.
3. Discussion The use of CAAD and multimedia techniques for the Leuven project had, besides the rather well-known assets of such an approach, some specific outcomes.
7
Using dynamic computer graphics made it possible to effectively illustrate the peculiar character of gothic construction techniques to both a professional and a laypersons public. One example illustrates this very well. Today, virtually everyone asked to do so will spontaneously analyze the front façade of the Leuven town hall in terms of a series of bays with windows. The contemporary gothic approach was, however, different. It considered the façade as being built up of three layers of pillars, each one going from half a window to half the next window (figure 6). This meant in essence a structural logic. By consequence, 15th century tendering for contractors went by pillars and not by windows, a fact experts had clearly observed from the archive sources.
Figure 6: pillar principle. It is the structuring element for the famous front façade. (Model by Han Vandevyvere)
As we know for every pillar which contractor got the contract and in which order they were executed, the building process could be dynamically represented with an uncommon precision. The video producers took up that detail, and let the construction rhythm of the exhibition animation coincide with the rhythm of the accompanying music, a style figure that was capable of passing the message without words.
Besides its instructive representation power, computer modeling led in the Leuven case also to a research spin-off. While building the 3D AutoCAD model of the front house from older survey plans, repetition in a series of measures started to appear. Analysis of the measure and number proportions led to the
8
discovery of a modular system in the dimensions of the building (figure 7). The probability of such an intentional geometric system not only stemmed from the numerical and graphical analysis that was performed, but also from the fact that the base module for the whole design appeared to equal two Leuven rods, with a Leuven rod being the measurement unit in use in the city during the 15th century. One foot in Leuven stood for 0.2855 m, and 20 feet composed one rod (5.71 m).
Figure 7: cubic grid underlaying the design composition of the Leuven town hall (Reconstruction scheme by Han Vandevyvere)
After analysing the Leuven town hall, a series of other town halls was researched in a similar way. Several buildings appeared to be conceived following the same design logic , and each of their building plans was laid out in the measurement units used locally. This eventually led to the formulation of a major hypothesis on the systematic geometry of Flemish town hall architecture in general. By conclusion, representation by CAAD techniques not only enhances knowledge acquisition for others, but the act of modeling may do that for the modeler as well. This is in fact similar to the observation that by drawing an object, one analyses it far better than by taking a picture of it: re -creation involves a much stronger contact and thus understanding with the design of an object than superficial re-presentation. Combined with the necessary background knowledge, this can lead to new insights.
9
In a more general way, the exhibition project fostered enough experience to set up a methodological framework that works also for other digital reconstructions. This will be discussed in detail further on.
4. Other digital reconstruction cases The Leuven project was furthered in several other reconstruction projects. They are individually characterized by a different degree of hypothesis building and by a minor or major contribution of computer techniques, i.e. digital reconstruction, in the total research output.
In the case of the historical research of two gothic houses in Leuven, there was a strong emphasis on the archaeological investigation in situ. This led, among others, to the discovery of a re-used and displaced roof structure that once covered the prestigious reception hall of Italian banker Bartholomeus de Madea’s private residence. The owner of this ‘palazzo’ comp lex, of which the two houses then formed a part, has entered the history books, not at least for provoking a financial scandal in the mid of the 15th century. Here again, the typological research of the analyzed building form led to new insights on the history of other important buildings of a similar typology. The reception hall theme was moreover particularly well-documented in contemporary Flemish primitive paintings (figure 9). This provided a stable basis for formulating a more general design paradigm on gothic halls in the Low Countries. The interior of the hall that once stood at the place of one of the houses under investigation, could thus be digitally simulated with a reasonable degree of certainty (figure 8).
Figure 8: evocation of the reception hall that stood at the location of a gothic house in Leuven; truss structure as it was discovered on another place in the transformed building. (Renderings and photograph by Han Vandevyvere)
10
Figure 9: contemporary iconography for typological research: detail of a 1438 painting by the Master of Flémalle. The similarity of the building form with the results of the independently performed archaeological reconstruction is striking.
A very different reconstruction project is currently done for the former hunting residence of Mary of Hungary, at Mariemont, some 50 km south of Brussels . The goal here is to make a full digital reconstruction of a building from which only some transformed foundation walls subsist. However, two major sources of information allow us to make a very feasable reconstruction of the castle as it was originally built in the mid of the 16th century. First, we dispose of the original surveys of the construction translated into words, summing up all building parts and their dimensions. The most important accounts based on these descriptions were issued between 1547 and 1550. Second, there is a considerable amount of iconographic source material. However these etches and paintings represent the castle in its 17th -century shape, whereas the reconstruction aims at re-creating the original design (figure 10).
11
Figure 10: interior reconstruction of the Mariemont castle based on original surveys summing up all building parts and their dimensions (Rendering by Simon Meersmans)
The building as it was erected by Mary of Hungary was a daringly modern renaissance villa. It stood on the top of a slope, in a water bassin, and was provided with a terraced roof (from where one could look over the hunting grounds). Already in 1554 it was destroyed by invading troops. It was rebuilt soon, this time with a pitched roof, and in this configuration it appears in the paintings. Most probably the choice for another roof type resulted from earlier infiltration problems with the lead-covered roof. For being so revolutionary at its time, the reconstruction effort concentrates on the original building, even if it existed only in that shape for less than a decade. Of all the digital representation formats for the reconstruction of the first design, the model showing every building part with its corresponding degree of certainty merits a closer look (figure 11). In this representation, all building elements are made partially transparent and receive a color code expressing the degree of hypothesis on that particular element. The scale evolves gradually from green to red. Green means certainty, yellow indicates a substantial degree of certainty, orange means an important degree of uncertainty and red stands for mere hypothesis. The building model can thus be represented showing only one accuracy level, or showing the composite image with the four accuracy levels displayed together. The latter gives an inmediate visual appreciation of what we know today about the building.
12
Figure 11: Mariemont hypothesis representations: ‘certain’ model (l) and composed model (r) overlaying the four accuracy codes. (Renderings by Simon Meersmans)
At this moment, the Mariemont project is in a fine-tuning phase because the reconstruction will be used in a public exh ibition on its architect, Jacques Du Broeucq. Recapitulation from the model building will provide for final corrections, and eventually lead to new digital output in different kinds of multimedia format (figure 12).
Figure 12: Mariemont: photoshop collage of rendered model and suggestive settings (Simon Meersmans, Han Vandevyvere, Krista De Jonge)
5. Methodological framework As mentioned before, the Leuven town hall project triggered a series of digital reconstruction projects, allowing to gradually establish a more general methodology. We now briefly discuss this framework.
13
To effectively set up a reconstruction project, we think it makes sense to distinguish five work phases: 1.
Survey and archaeological investigation: this phase concerns the survey and the investigation of the site and the buildings or archaeological remains that are present on it. In most cases, this work is done or has been done by specialists not involved with the actual reconstruction project. However, going back to the source material in situ at the moment of starting the digital reconstruction still may prove to be fruitful, even if some studies or documents have already been produced. In particular, attention shall be given to distinguishing the different physical strata that facilitate the formulation of hypotheses on the site and building layouts at different moments in history.
2.
Historic investigation: the historic investigation combines source materials from survey, archaeological research, archives, literature and iconography about the site and its history, and summarizes the data, findings and conclusions in alphanumeric reports (i.e. the ‘written’ history).
3.
Creation of a ‘metafile’: the metafile is considered the nerve centre of the reconstruction exercise. It is a worksheet that mentions and orders all results from the survey and the historic research in a specific way, proper for formulating hypotheses that will underpin the computer modeling of the site and/or building throughout history. In particular, the metafile comp rises all possible references to iconographic sources, together with a rating for their reliability. These score valued references are of major importance when one has to create 3D geometric/graphic models.
4.
Building the 3D digital reconstruction model(s): in this phase the material from the metafile is transformed into one or more geometrical models. This work implies a continuous feedback from and towards phases 2 and 3. Data from phase 3 enable the initial setup of the spatial model. Where information for completing this model is lacking, additional historic research may give an answer. If that is not the case, a hypothesis will be formulated and a correspondent geometry modeled. The resulting degree of hypothesis is reflected in the structure and representation format of the spatial model.
5.
Creation of the multimedia interface: the spatial model and the metafile contain all the elements that can be used for the creation of any kind of multimedia output. For example, one can link
14
scanned images of iconongraphic or archive sources (referred to in the metafile) to clickable areas of a space model, in its turn accessible through an interactive website. Video, text and sound are selected from this combined base materials or are produced through specific software (e.g. 3D-studio animations on basis of a geometric model built in AutoCAD). Again, feedback from or to phase 2 can be necessary to compensate for information gaps in the metafile or the geometric model.
6. Further discussion As mentioned, the metafile is at the core of the reconstruction process. It must be built in a rigourous way and follow clear data representation conventions, so that a team can work on it, even after a long period of inactivity. A person picking up the data for a new phase of research should always be able to extract precise knowledge from the metafile. Assessed interpretations are as important as factual sources: e.g. the fact of having merely referred to a painting representing the building is worthless if we know that the painting is not completely reliable as a source (which is rather often the case with historic images). It is necessary to include the judgement on that painting under a precise formulation in the metafile, for later processing. Hypotheses formulated in the metafile are in words, or in the form of sketches and drawings intended to support the modeling process. Graphical schemes often can easily represent knowledge that hardly can be expressed by a written description.
It becomes clear that phases 1 to 5 ma ke up a recursive process; this is particularly the case for phases 2, 3 and 4. The discovery of a proportion system in the Leuven town hall design is a singular example of feedback from modeling to historic research, but in practice this kind of feedback will rather concern prosaic problems (such as, e.g., was the direction of the pitched roof ridge parallel or perpendicular to the street? Can the layout of the windows on painting X be taken for sure? Is the number of depicted floors right? And so on…).
When performing a digital reconstruction, historic research and modeling efforts should be done with a particular attention for contemporary building typologies. The high importance of typological
15
information stems from the simple fact that one can fall back on the architectural composition of similar buildings when hypotheses are needed on the particular building under investigation. When creating architectural details, this typological knowledge is of utmost importance. It guides the shaping, dimensioning and proportioning of building parts. In accordance herewith, the over-all value of a reconstruction model highly depends on the sensitivity of the modeler to ‘feel’ and correctly interprete the architectural language of the historic building style under consideration. If the modeler fails to do so, the digital reconstruction will be of a poor quality even if it is underpinned by a sound historic research and a professionally built metafile. This means that the modeler needs to combine the qualities of architectural sensitivity and CAAD-related skills.
7. Conclusion It becomes obvious that historical digital reconstruction is not only an interdisciplinary process, but also a process of which the output depends much on the ability of the different specialists in the reconstruction team to work tightly together. As such, e.g., a CAAD expert needs to be able to ‘feel’ historic reality and a historian needs to be able to assess the problems of a spatial translation of written or pictorial source material. If that is the case, digital historical reconstruction may foster results that are often typical of interdisciplinary research, i.e. the discovery of new insights through combining knowledge that was never combined until then. A public confronted with the final output of such a process will then not only enjoy the possibilities of knowledge transfer through digital media, but as well the plus-value of new, combined research content.
The advent of the digital media in historical reconstruction (Patricia Alkhoven,…) is not merely just the new tool to perform the traditional historical approaches. Our research experience proves that digital modelling introduces a new competence into the historical debate, sheds a new light on the subject at study, and thus contributes significantly to the understanding of our past building history.
16
Bibliography [1] De Jonge, K. (2005), Mariemont “château de chasse” de Marie de Hongrie, in: Revue de l’art, No.148, CNRS, Paris.
[2] Maesschalck, A., Vandevyvere, H. & Viaene, J. (2002), Een 15de-eeuws bankkantoor op de Vismarkt te Leuven. Een kunsthistorisch en bouwkundig onderzoek van twee gotische panden, in: Jaarboek van de geschied- en oudheidkundige kring voor Leuven en omgeving, No. 40, Leuven, pp. 93-271.
[3] Meersmans, S. (2003), Het domein en het jachtslot Mariemont ten tijde van Maria van Hongarije: een digitale reconstructie, unpublished masters’ thesis, ASRO, K.U.Leuven.
[4] Vandevyvere, H. (2001), Gothic Town Halls in and around Flanders, 1350-1550: A Geometric Analysis in: Nexus Network Journal, vol. 3, no. 3.
[5] Vandevyvere, H. (2002), Gothic Town Halls in and around Flanders, 1350-1550: A Geometric Analysis , in: Symmetry: Art and Science, The Quarterly of the International Society for the Interdisciplinary Study of Symmetry, Vol 2 (new series), No. 1-4, pp 365-378.
[6]Vandevyvere, H. (2000), Het stadhuis van Leuven: een geometrische analyse, in: Jaarboek van de geschied- en oudheidkundige kring voor Leuven en omgeving, No. 39, Leuven, pp.171-192.
[7] Viaene, J. & Maesschalck, A. (1979), Het stadhuis van Leuven, in: Arca Lovaniensis artes atque historiae reserans documenta: jaarboek 6 - Mensen en bouwkunst in Boergondisch Brabant, Vrienden stedelijke musea Leuven, Leuven.
17
