From Archiving to Activation: A Methodological Pipeline for Cultural Heritage Visualisation

From Archiving to Activation: A Methodological Pipeline for Cultural Heritage Visualisation Evgeny Kalachikhin Martin Gordon Sophie Tummescheit Bjoern Stockleben Creative Exchange Studio, Film University Babelsberg KONRAD WOLF, Germany While cultural institutions invest heavily in digitisation, a significant gap continues to exist between preservation-focused captures and the content required for creative industries [1, 2]. This "digital heritage paradox" [4] leaves valuable assets underutilised for new narratives [5]. We present a methodological pipeline that addresses this by treating 3D capture as a dual-purpose process. Our core contribution is using the rich photographic datasets from any multi-view 3D capture [6] not as a temporary byproduct for an archival model, but as a bespoke training set for generative AI. This dual data stream of visuals and geometry intelligently conditions pre-trained models via accessible tools: parameter-efficient finetuning (LoRA) [3] for material fidelity and conditional diffusion (ControlNet) [8] for spatial accuracy. A crucial expert review phase guarantees historical authenticity. We demonstrate this through two contrasting case studies - material and intangible heritage - culminating in a novel format for spatial narrative. Our material heritage case study, with the Film Museum Potsdam, sought to contextualise a static 3D scan of a costume from Fritz Lang’s 1924 film, Die Nibelungen, placing it into new creative narratives for various media (Fig. 1, Row 1). This goal also provides a powerful answer to a common industry question: what is the value of a 3D scan beyond offering a simple digital view, given that further creative applications typically require costly production workflows? Our pipeline initially did not produce the expected outcome due to the limitations of standard generative models, which failed to grasp the artifact’s unique material essence. Our key solution was to create a curated training set composed of both source photographs and synthetic renders from our dual-purpose capture. This allowed us to fine-tune a model [3] that accurately reproduced the costume’s specific properties. Achieving this material integrity transformed a static 3D model into a source for high-quality, cost-effective visualisations. The success of this pipeline is validated by our partnership with Botspot 3D Scan GmbH, a leading developer of professional scanning systems. The collaboration provides our research with high-end capture hardware in exchange for our contextualisation, which offers them a potential use-case for their clients. The partnership is now expanding to the digitisation of the next series of costumes from the museum’s collection. For intangible heritage, our project with the Institute of Urban Culture in Bielsko-Biala tackled a profound challenge: how to visualise a history that exists almost exclusively in textual accounts. The story of the laundry as a clandestine contact point for Auschwitz prisoners is documented in historical research [7], but by only four low-quality archival photographs. Our approach was twofold: first, we generated narrative scenes based on the analysis of textual archives and scholarly articles. Second, to ground these narratives in physical reality, we conducted an on-site photogrammetric capture of the building. This 3D model provided the essential architectural ground truth, allowing us to use tools like ControlNet [8] to ensure all visualised events adhere strictly to the site’s geometry. An iterative refinement loop with expert review was vital for correcting anachronisms and navigating the ethical complexities of visualising such a sensitive history (Fig. 1, Row 2). As proof of concept we developed the "Corridor of Memories" prototype (Fig. 1, Row 3) - a novel format for historical storytelling. This is a projection-based installation that uses the foundational 3D structure from the capture process as a "digital canvas." Onto this real-world geometry, we layer visual narratives combining original archival materials with our verified reconstructions, grounded in precise 3D geometry. This immersive environment allows the spectator to experience a multi-layered story grounded in historical and spatial reality. Our work presents a methodological pipeline that bridges contextualisation with visualisation by treating 3D capture as a dual-purpose process. Figure 1: The methodological pipeline: Row 1 (material heritage) - from archival documentation to 3D capture and AI-generated visualisation. Row 2 (intangible heritage) - from geometric site capture to historically accurate narrative scenes. Row 3 - the "Corridor of Memories" concept, transforming the 3D capture into an immersive spatial narrative. The visual data gathered for archival purposes becomes a bespoke training set for generative AI, providing a validated and practical pathway to activate cultural heritage archives for contemporary media. Crucially, this pipeline is democratised, requiring only consumer-grade hardware. Its validation moves beyond academia, confirmed by industrial partnerships and strategic discussions on new cultural engagement (e.g., developing the historical laundry into a cultural centre). This provides a promising response to the "digital heritage paradox", turning archives into active content creators and dissolving the distinction between preservation and creation. [1] F. I. Apollonio et al. A photogrammetry-based workflow for the accurate 3d construction and visualisation of museums assets. Remote Sensing, 13(3):486, 2021. [2] A.-M. Boutsi et al. An integrated approach to 3d web visualisation of cultural heritage heterogeneous datasets. Remote Sensing, 11(21): 2508, 2019. [3] E. J. Hu et al. LoRA: Low-rank adaptation of large language models. In ICLR, 2022. [4] W. Hupperetz. The digital heritage paradox. In CHNT, 2015. [5] J. Korro Bañuelos et al. The role of information management for the sustainable conservation of cultural heritage. Sustainability, 13(8): 4325, 2021. [6] K. jr. Pavelka and J. Pacina. Using of modern technologies for visualisation of cultural heritage. The Civil Engineering Journal, 4-2023: 41, 2023. [7] J. Proszyk. Pralnia w bielsku-białej jako przykład pomocy udzielanej wi˛e´zniom kl auschwitz-birkenau. In M. Panecki, editor, Formy zniewolenia na okupowanych przez Niemcy ziemiach polskich (19391945). 2023. [8] L. Zhang et al. Adding conditional control to text-to-image diffusion models. In ICCV, 2023.