EuroMed2024 Workshop 3

CoVHer Multiplier Event on Hypothetical 3D Reconstructions and Documentation in Cultural Heritage

CoVHer stands for “Computer-based Visualization of Architectural Cultural Heritage” and is an Erasmus Plus Project (2021-1-IT02-KA220-HED-000031190, CoVHer Project), which began in February 2021 and will end in January 2025.

The use of 3D models for the virtual reconstruction of past architectures, whether unbuilt or lost, has become widespread among both scientific and amateur communities. Despite this growth, the field still lacks standardized practices for processing and evaluating these virtual reconstructions.

Key gaps include the absence of a shared glossary for specific terms, including a shared classification of digital representation methods and 3D modelling techniques, as well as a scholarly approved, ready-to-use methodology for computer-aided hypothetical 3D reconstruction and documentation.

This Multiplier Event will address several unresolved issues in the field, particularly:

Documentation of uncertainty in hypothetical reconstructions
Publication of 3D models as scientific products
Quality assurance and assessment of 3D models
Accessibility and reusability of results

In this Multiplier Event, we aim to delve deeper into these critical points by presenting and discussing some results developed in the context of the CoVHer project.

Specifically, we will illustrate the CoVHer MOOC course that will be published in January 2025; we will present the methodology to assess and visualize the uncertainty of hypothetical virtual reconstructions, as well as CoVHer developed to share 3D models as open scientific research outputs. Among the aims of this Event we will explore the potential of creating a digital archive of 3D models of virtual reconstructions, such as the CoVHer 3D Repository, and we will aim to promote the use of a shared terminology and methodology for the documentation of the reconstruction process and the assessment and quantification of the uncertainty in virtual hypothetical reconstructions.

Other scholars are also invited to the event to discuss ongoing or past European projects and international initiatives related to these topics, and the issues raised by the European Union concerning the preservation and sharing of European cultural heritage.

Join us to contribute to and discuss about the advancement of standards and methodologies in the field of 3D virtual reconstruction of architectural heritage.

Organisers

Accepted Abstracts and Authors

Alma Mater Studiorum – University of Bologna
Italy

Federico Fallavollita, Riccardo Foschi

Federico is a associate professor at the Department of Architecture of University of Bologna. In 2005 he graduated with honors in architecture at Sapienza University of Rome. In 2008 he obtained a PhD in Sciences of Representation and Survey at the Department: History, Design and Restoration of Architecture at Sapienza University of Rome with a thesis entitled: The ruled surfaces and developable surfaces, a reading through the virtual lab. He deals with the issues of representation and survey of architecture. In particular he is interested in renewal of descriptive geometry through the new informatics tools. He has participated in many seminars and international conferences presenting several memoirs, including Construction of the three principal axes of quadric ruled surfaces (intervento come relatore nel The 15th International Conference on Geometry and Graphics, 2012); and has published several articles including: The extension of the Problem of Apollonius in space and L’Ecole Polytechnique (in Ikhnos 2008). He is the author of a few chapters of the book Descriptive Geometry by Riccardo Migliari, including: Construction on the circle (vol. I) and The ruled surfaces (vol. II). Currently he is involved in the research team led by Marco Gaiani on implementation of computer tools for the study and visualization of architecture at the DA (Department of Architecture) in Bologna.

Empowering Digital Heritage: CoVHer MOOC on 3D Reconstruction of Lost and Unbuilt Architectural Heritage
The CoVHer MOOC of the Computer-based Visualization of Architectural Cultural Heritage Ersamus + Project (https://covher.eu/ ), grounded in research, provided comprehensive training on digitally reconstructing architectural cultural heritage (CH) that has been lost or was never built. Aimed at students, scholars, and the public, this course sought to enhance digital skills and deepen understanding of scientifically accurate virtual reconstructions, bridging research with accessible learning.
The course was structured into several modules, each focusing on key aspects of 3D modelling and digital visualization. Topics included an introduction to hypothetical reconstructions, the scientific foundations of model creation, and best practices for constructing accurate and accessible 3D models. Through a combination of lectures, quizzes, and assignments, participants learned about different modelling techniques, historical source documentation, and the importance of communicating uncertainty within digital representations.
One of the course’s unique aspects was its emphasis on interoperability and accessibility, providing guidelines for publishing models on digital platforms and ensuring they met scientific standards. In its final weeks, the MOOC covered documentation, sharing, and reuse of 3D models, including insights on augmented and virtual reality applications for these reconstructions.
Through the CoVHer MOOC, a research-driven initiative, participants gained a solid foundation in creating, evaluating, and understanding the role of 3D models in preserving and communicating cultural heritage. The course presented the research findings in a clear and accessible way, underscoring the importance of digital skills in this field and enhancing public appreciation of lost or unrealized European architectural heritage.

University of Exeter
The United Kingdom

Fabrizio Nevola & Luca Brunke

Professor Fabrizio Nevola is the Head of Art History and Visual Culture and the Deputy Head of the Department for Languages, Cultures and Visual Studies. He was the Director of the Centre for Early Modern Studies (2016-20), and is an elected member of University Senate. He specialises in the urban, cultural and architectural history of Early Modern Italy, including a special interest in street life, and is involved in innovative work using digital art history approaches for research and public engagement.
Fabrizio did his undergraduate degree in Modern History and Italian at University College, Oxford and MA (History of Architecture) and PhD (History of Art) at the Courtauld Institute of Art. He has held research fellowships at the University of Warwick, the Canadian Centre for Architecture (Montreal), the Medici Archive Project (Florence), and Harvard University’s Villa I Tatti (Florence). In 2022, he was Visiting Professor at Ca’ Foscari, University of Venice, Centre for Digital and Public Humanities.
Fabrizio’s first book, Siena: Constructing the Renaissance City (Yale University Press, 2007) was awarded the Royal Institute of British Architects, Sir Nikolaus Pevsner International Book Award for Architecture. His most resent monograph is Street Life in Renaissance Italy (Yale University Press, 2020): it was shortlisted for the Renaissance Studies Biennial book prize (2022). Among various edited volumes, he most recently edited Hidden Cities. Urban Space, Geolocated Apps and Public History in Early Modern Europe (Routledge, 2022).

Abbreviated workflows for 3D reconstructions in art and architectural history
Over the past five years the Florence4D team have contributed to the debate around standards for research-based 3D models of lost or modified buildings from the past, and the art and material culture they contained (e.g. Nevola et al. in Getty Research Journal, 2022). Common ground is emerging in approaches to research-based or serious 3D models and the structured underpinning datasets that are an essential part of documenting uncertainty in reconstructions. However, such approaches are time intensive and are not always possible, even where a scientific output is a required end product. Visualisations/reconstructions are increasingly used as part of the ways in which museum and gallery curators communicate issues of historical context to visitors at permanent and temporary exhibitions, raising a question as to how these might better conform to our desired standards for scientific quality and reusability in art and architectural history.
This talk uses as a case sample a series of ongoing reconstructions produced by members of the Florence4D team for major exhibitions and related publications on Renaissance artists, where visualisations form part of a complex research-based argumentation proposed by curators. The question of uncertainty that is so central to how a visualisation responds to absences in the surviving evidence does not have equal valence for different constituencies of the research and curatorial teams, causing us to question: whose uncertainty? So then for example, in an altarpiece reconstruction, the curatorial team will be most concerned with ordering the surviving (and lost) panels correctly, while perhaps for the 3D modeller the main challenge is around proposing the reconstruction of the carpentry frame. Moving to a more complex full-scale building model – as we know – the challenges multiply as various parts of the modelling process frequently shed light on areas often overlooked in traditional analysis, while the placement of objects or the experiential representation of movement or light raise further series of variables. In the case example in question this led us to adopt a light 3D modelling approach where we limited the output requirements in order to avoid having to model all aspects of the building through the selection of optimal camera angles within the modelling environment that met the curatorial brief.
While time constraints on these reconstructions mean that a full implementation of research-based/serious 3D standards may not be possible, a simplified version of the workflow can ensure a level of methodological rigour that is often lacking in such visualisations. Such an approach in turn invites us to return to the workflow to consider how input data (including scholarly interpretations of that data) can be gathered and stored while attending primarily to the output visualisation. Systematic gathering and storage of the input data, combined with the curatorial analysis text would allow for possible future processes (including through use of AI chatbot assistants) to retrospectively reconnect underpinning data to the models.

Friedrich-Schiller-Universität Jena
Germany

Sander Münster

Dr. Sander Münster is junior professor for Digital Humanities (images/objects) at the Friedrich Schiller Universität Jena (DE). He received his PhD in educational technology from the Technische Universität Dresden (DE), where he studied history, education and business. Until 2019 he headed the Department for Media Design at the Media Center at the TU Dresden and the junior research group UrbanHistory4D and has been a Young Investigator at the Faculty of Education at the TU Dresden.
His research fields are mobile 3D interfaces, scientometrics and research methodologies in Digital Humanities and heritage. He has worked extensively in the areas of research dissemination and innovation support. The related positions he has held include: co-convenor of the workgroup for Digital Reconstruction of the German Association for Digital Heritage, spokesman of the German network on “3D reconstruction methods for architectural history” and head of the workgroup “Partnering with international organisations” in the Virtual Multimodal Museum EU network.
Since the beginning of 2024 he is an elected member of the CIPA Heritage Documentation executive board, co-chairing the Permanent Commission “Application of Recording, Documentation, and Information Management for Cultural Heritage”.

Virtual reconstructions in the age of generative AI
Virtual reconstruction of past architectures are still a very complex and research and labour intense approach. In our work we assess to which extent virtual reconstructions
could be created according to principles of the critical digital 3D model via technical pipelines and generative AI tools.
From a formal point of view, the reproducibility and objectivity of virtual reconstructions created by pipelines can be closer to scientific principles than ever before. They can be fully documented and uncertainty can be quantified in terms of probabilities and deviations between validator and proposer. However, the issues raised are of a deferent nature and include, for example, biased training data, parameter sets and numbers, and explainability.
A key issue is the truth of the content generated by the generative AI. As the proposer and validator architectures are still very similar to the process of traditional virtual reconstructions – as reconstructing and validating a 3D hypothesis – generative AI raises questions about the nature of scientific 3D reconstructions: While there is no doubt that the results are hypothetical and lack object specificity, the extent to which the results (such as inferring patterns from large examples to propose a single case) are a valid intellectual deduction and thus a scientific result is questionable.
What does this mean for the issues raised in the workshop? Regarding the documentation of uncertainty in hypothetical reconstructions, this could be well quantified and codified. Also, a grounding in sources can be formalised – limitations in this case are mainly due to the lack of explainability of AI results. Publication of 3D models as scientific products and accessibility and reusability of results: The current step is to separate the data (historical sources) and the modelling process (modelling pipelines): both could be reproduced and measured. Quality assurance and evaluation of 3D models is the big issue, especially for more speculative techniques.