The objects in the museum basement

Why small collections need a digitisation strategy of their own.

When digital cultural heritage comes up, the subject is usually monuments: buildings, memorials, archaeological sites or landscapes. For work of that kind, high-precision methods such as terrestrial laser scanning, controlled photogrammetry and geodetic survey are as a rule the right choice.

In many cases, though, the gap originates in the storeroom itself. Museums keep a large part of their collections outside the permanent exhibition. A survey by the International Council of Museums (ICOM) found that many museums show only a small part of their holdings permanently. According to a more recent museological analysis, holdings of up to 90 per cent may sit in storage, depending on the institution.^[1]

The challenge is therefore not only to digitise large sites spectacularly. The challenge is to document a great many smaller objects — ceramics, tools, coins, textiles, figures, technical devices or regional everyday items — step by step, to make them findable, and to open them to a wide public.

What the figures say

The data situation is uneven. There are currently no reliable figures from which a worldwide total of museum artefacts still to be digitised could be derived. The numbers differ by country, by type of collection, by what "digitised" is taken to mean, and by whether inventory data, 2D reproductions, 3D models or content available online is meant.

The most frequently cited Europe-wide comparison comes from the 2014 ENUMERATE survey. It covered various types of cultural heritage institution, not museums alone. On average, 17 per cent of analogue holdings had been digitally reproduced at that point, while respondents estimated that 52 per cent still had to be digitised. The study points out explicitly that these are unweighted averages.^[2]

These figures must not be presented as the current state of Europe or of individual museums. What they do document is a structural starting position: digitisation is a long-term task whose scope is considerably larger than that of individual flagship projects.

More recent policy reports confirm that the task remains. In 2025 the European Commission found that while member states’ progress was measurable, the ambitious targets for 2025 and 2030 on digitising cultural heritage remained clearly out of reach. According to the 2020 cultural statistics, for example, 63 per cent of museums in Austria had a digital inventory and 44 per cent a collection available online. Among the institutions in a later, non-representative consultation these shares were higher.^[3]

No competition for survey work

For a great many tasks in heritage protection, in restoration, in inventory work with metric requirements or in scholarly analysis, a visually convincing 3D representation is not enough. Classical photogrammetry, laser scanning and documented measurement protocols are indispensable there.

Image-based methods such as 3D Gaussian splatting focus on something else. They turn a large number of photos, or a video, into a spatially displayable scene that can be viewed interactively. The method is interesting for the visualisation, interpretation and digital accessibility of cultural assets — with a clear quality check remaining essential.

Specialist publications name, for 3D Gaussian splatting on cultural objects, challenges arising from image alignment, unsteady shooting conditions, backgrounds, reflections and incomplete viewpoints. Comparative studies also stress that classical methods can retain advantages in geometric accuracy, depending on the object and the objective.^[4]

The appropriate positioning is therefore this:

DFX SplatCore complements existing digitisation methods where a local, visually high-quality and scalable 3D presentation of small objects is needed. It replaces neither metrological survey, nor conservation assessment, nor scholarly documentation with defined accuracy requirements.

The scalable workstation

For smaller, transportable and cleared objects, a compact local workstation can be a sensible way in. It should have a camera, even diffuse light, a suitable turntable, a neutral or textured background, and a computer with a capable graphics card.

DFX SplatCore is designed as a locally running Windows application. A system with at least 16 GB of VRAM is a realistic starting point for more demanding reconstruction workflows. Image data can therefore be processed inside the museum, the archive or at the commissioned service provider, without any transfer to a cloud service.

A low barrier to entry does not mean, however, that every data set automatically becomes usable. A camera may be within everyone’s reach today, but a reproducible shoot still needs a protocol. Even light, sufficient image overlap, constant camera settings, clean focus and appropriate handling of reflective, transparent or fragile objects are decisive for the quality of the result. Research on small archaeological objects shows the limits and quality differences between 3DGS, NeRF and classical photogrammetry particularly clearly with complex geometries and reflective surfaces.^[4]

From the find site to the storeroom

The part that is interesting economically and organisationally lies in the division of labour.

The shoot can also take place at a different site from the one that has the necessary computing hardware and 3D expertise. An excavation, a branch office, a regional museum or a volunteer partner can produce a standardised photo series. The unchanged original data is then handed to a central specialist office. There the images are reviewed and reconstructed, the metadata maintained, the results exported and finally released.

This model cannot replace a specialist’s expertise. What it does is distribute the work sensibly. The rare infrastructure — capable hardware, a 3D workflow, quality control and web editing — can be provided centrally. The shoot happens where the object is.

That matters especially for fragile objects, objects under protective regulation, or objects that cannot readily be transported. Whether and under what conditions something is photographed, transmitted or published is determined by ownership rights, excavation conditions, conservation requirements and the decisions of the responsible institutions.

Everything hangs on the shoot

The shoot is the only link in the chain that cannot be fully repaired afterwards.

A missing rear view cannot be reliably derived from an incomplete series. Problems with blurred images, changing light, blown-out surfaces or files compressed by a messenger app can impair the quality of the entire workflow. Computing power applied later cannot replace missing image information.

For shoots in the field and in storage, a clear minimum protocol should therefore apply:

The object is circled completely in overlapping views. Additional raised views close gaps at the top surfaces and in recesses.

Within a series, the settings for light, focal length, focus and exposure stay as constant as possible.

In line with institutional rules, a scale bar and an object or find ID are documented.

Before packing or returning the object, a short check confirms sharpness and coverage.

The original files stay unchanged. They are filed in order by object ID and backed up redundantly at least once.

Publication and transfer take place only after rights have been clarified and under protection and release rules.

A scale bar makes size relations intuitively clear. To use a 3D model as a measuring instrument for metrological or forensic purposes, however, a calibrated shoot, control points and a documented evaluation are required.

Local data sovereignty

In the field of cultural assets, data paths are part of professional responsibility. Find sites, as yet unpublished finds, loans, ethnographic collections or sensitive object information can be subject to particular legal, ethical and institutional rules.

Local processing replaces no clarification of rights. What it does enable is technical control: original images, intermediate states and models can remain within your own infrastructure, or that of a commissioned service provider. DFX SplatCore is built for this local workflow.

Doing without cloud compulsion also matters for archiving. To build a robust digital memory, institutions are well advised to preserve not only an online view but also raw images, metadata, documented working steps and output formats that are as open as possible. DFX SplatCore supports PLY and GLB output. These should nevertheless be stored together with version information, metadata and a clear archiving strategy.

Interpretation begins with the embedding

A digital representation only becomes publicly effective once it is embedded intelligibly.

With DFX Orbit, a model can be extended into an interactive presentation with hotspots, image galleries, zoom, an audio guide, before-and-after views and guided camera paths. A marker can explain an incised line, an image gallery can show excavation or restoration photographs, and an audio guide can convey the object’s history.

Access can run through a museum website, a collection portal or an exhibition station. An export that runs on its own can also be used without a permanent internet connection — beside a display case, in a travelling exhibition or on an interpretation device.

Visibility complements preservation

Digital representations replace neither conservation, nor inventory work, nor research, nor restoration. What they can do is widen an object’s visibility and open up new uses.

An object in storage stays protected. At the same time it can feed digitally into education, research, exhibition preparation and public interpretation. Visitors can explore views that are hard to reach behind glass or in a conventional display case. Researchers and interested visitors gain an additional visual reference without the original having to be moved constantly.

European digital policy pursues exactly this wider frame: access, reuse, long-term preservation and the building of skills and infrastructure for digital cultural heritage. The Commission expressly calls on member states to accelerate their progress in digitisation and digital preservation.^[5]

A realistic first step

Not every collection needs a large-scale 3D initiative straight away. A sensible start can be a small pilot series with clear professional boundaries: some twenty objects of one type, with a documented shooting protocol, checked metadata and a defined interpretive aim.

That allows shooting times, limits with certain surfaces, file sizes, computing effort, quality and editorial work to be assessed for real. Decisions about scaling, priorities, staffing models and long-term archiving should only follow after that.

Erecting and surveying large monuments will continue to need specialist teams and precise measurement methods. The small objects in the basement need careful planning for their future too.

There is no compelling need to start with a laser scanner.

At the beginning stand a careful shoot, clear data paths and a technique that makes digital visibility practically achievable for many objects.

Sources

1. [1] icom — https://icom.museum/en/news/museum-storage-around-the-world/
2. [2] pro.europeana — https://pro.europeana.eu/page/results
3. [3] digital-strategy.ec.europa — https://digital-strategy.ec.europa.eu/de/news/eu-report-calls-member-states-accelerate-digitisation-cultural-heritage
4. [4] isprs-annals.copernicus — https://isprs-annals.copernicus.org/articles/X-M-2-2025/215/2025/isprs-annals-X-M-2-2025-215-2025.pdf
5. [5] digital-strategy.ec.europa — https://digital-strategy.ec.europa.eu/de/policies/cultural-heritage