Preserving 3D Scans: A Journey

Authors

• Lode Scheers (meemoo)
• Nastasia Vanderperren (meemoo)

Abstract

The world of digital heritage is rapidly evolving, with 3D scanning becoming increasingly common use to preserve valuable heritage artifacts. Are we adequately preparing for the long term? Our research into the preservation of 3D scans tackles that crucial question, by investigating how cultural heritage institutions preserve these valuable 3D assets. A recent international survey and literature review revealed a concerning gap: while 3D digitisation speeds ahead, preservation often lags behind. Many organisations rely on fragile solutions like simple backups or online platforms, hardly ideal for ensuring future access and understanding the data.

Our study emphasizes that robust preservation requires more than just saving the 3D file itself. It highlights the importance of selecting appropriate archival formats, favouring open and text-based options like X3D, STL, and OBJ, alongside preserving original raw scanner data, a vital safeguard against future software incompatibilities.

Comprehensive metadata is key to preserve these 3D models. This goes beyond descriptive and technical metadata, but includes paradata and provenance metadata. Paradata is detailed information of the creation of the model, such as scanner settings, processing techniques and decisions. Provenance metadata describes the history and origins of a 3D model, ensuring authenticity and transparency. We researched specialised 3D metadata models like CRMdig and the Smithsonian 3D Metadata Model which provide the ability to store these essential metadata.

Ultimately, this research journey underscores a critical need: to bridge the gap between rapid digitisation and thoughtful preservation. By prioritising raw data, embracing open formats, and implementing robust metadata strategies, we can safeguard these digital treasures and the stories they tell for generations to come.

Introduction

This all started as an internal research project…

The goal was to renew our 3D preservation guideline¹, since the last update was made back in 2016. In 2021-2023, meemoo also started 3D digitising Flemish masterpieces². These developments presented an opportunity to start the work needed.

We started gathering insights with an international survey and conducted a literature study to know how other organisations were preserving 3D scans. We received twenty responses, mainly from Europe, from different types of institutions in the cultural heritage field. Findings were compiled in our final report: A State of the Art in Preserving 3D³. This short paper gives a summary of that report.

Preservation Policy

Preservation is often an afterthought. Institutions tend to only think about preservation after a digitisation project is already finished. A finding that is also reflected in our survey. Only two of the respondents had a preservation plan in place. However, there are many risks involved in preserving 3D objects, which are further magnified if you do not have a preservation policy, such as:

• 3D files are complex, evolve quickly and there is a lack of documentation. Furthermore, there is no consensus on the use of standards, both for files and for metadata.
• There are no open-source formats for raw files. (Raw) files are usually not backwards compatible causing institutions to rely on outdated software or vendors, potentially losing access to files in the future.
• Information and context are lost because not all essential metadata is retained. Preserving the full story requires rich, well-structured metadata for both physical objects and their digital files.

The European Commission has been clear on this issue as well: plan before you scan⁴.

Files And File Formats

Numerous file formats exist for representing 3D models, each with its own capabilities and limitations. That is why choosing the right format can be a complex process. The survey and literature study revealed that there is no consensus regarding standard (archive) formats. However, there is consensus on what type of files should be stored:

• The original raw scanner data to allow quality control or for future processing.
• Derivative formats that are tailored to specific use cases, like visualisations, printing, AR/VR, …
• A high-quality derivative in an open and standardised format that is used as an archive copy.

It is recommended to opt for text-based and standardised formats that can be identified by PRONOM and can be validated. This gives you independence from specific software programs and offers greater flexibility.

Archive Formats

Based on the literature review and the survey, we took a closer look at six potential archive formats:

• X3D (°1997): an open-source file format and ISO standard for representing 3D-models maintained by the Web3D Consortium. X3D files are viewable directly in the browser. It is based on XML and can be validated by an XML Schema. X3D boasts a long list of features including colour, textures, animations and embedded metadata. The file format is less adopted because of its complexity.
• STL (°1987): remains one of the most stable and widely used formats. It is a royalty-free, patented format that offers encoding in ASCII text. It is limited for long-term archival purposes due to its lack of structured metadata, textures and colour support. Also, the format is no longer maintained and only a summary of the specification on some hobby websites is available.  Despite these limitations it is the current de facto standard for 3D printing.
• OBJ and MTL (°1990): one of the most widely used open and ASCII text-based formats for 3D models that contain texture. Like STL, it is a simple format that lacks support for structured metadata, is no longer maintained and has a specification that is only findable on hobby websites. OBJ remains a popular choice for archiving models, as evidenced by 70% of our survey respondents.
• DAE (°2004): an open-source file format and ISO standard for representing 3D models maintained by the Khronos Group. It was originally developed as an exchange format. Like X3D, it is based on XML and can be validated by its XML Schema. It has a long list of features including colour, textures, animations and structured metadata. A disadvantage of DAE is its complexity, making it difficult to implement in software by developers.
• PLY (°1994): a straightforward and widely supported format for storing simple 3D models. It is developed as an exchange format and is known for its simplicity, human-readability and flexibility. PLY files have restricted support for structured metadata and a lack of native texture support. The format is no longer actively maintained.
• E57 (°2011): A royalty-free file format developed and standardised by ASTM, mainly adopted for 3D laser and LiDAR scans. Because of its combination of binary data with XML-based metadata sections, it is known for its compactness. E57 has extensive features such as colour, textures, and structured metadata. Inconsistent implementations across different tools can pose interoperability challenges.

For meemoo’s case, the preservation of 3D scans, X3D and DAE seem powerful and future-proof options.

Metadata

For preservation, it is important to preserve extensive metadata in addition to the files. Specifically for 3D, given their complexity, it is crucial to preserve provenance metadata and paradata, next to descriptive, technical and preservation metadata.

Paradata details the methodology and decisions made during 3D model creation, explaining why specific choices were made, such as scanner selection, processing techniques, or file format choices. It’s defined in The London Charter for the Computer-Based Visualisations of Cultural Heritage (2006)

Provenance metadata tracks an object's journey, documenting its history including origins, agents, and sources, used to ensure authenticity and transparency of both the digital and physical versions.

Paradata and provenance metadata are both vital for preserving the integrity and usability of 3D models. Together, they provide a complete record of an object’s history and the digitisation process itself. In practice, this information is often not documented. A limited number of organisations indicated that they preserve 3D specific metadata.

On the other hand, while it is important to collect as much metadata as possible during the digitisation process, like technical outputs from the 3D scanner, post-processing steps, and a description from the workflow, it is not always possible to obtain these. Vendors often regard these data as a trade secret.

Metadata Standards

The importance of metadata is widely acknowledged, but neither literature nor the survey point to a metadata standard. Our survey shows that organisations use the typical preservation standards, such as PREMIS and METS, but these models lack the ability to store paradata. While specialised 3D metadata models exist, they haven’t yet found widespread adoption in the field.

The full list of metadata schemas is available in our report. In this paper, we briefly highlight those that support paradata:

• The Smithsonian 3D Metadata Model emerged from our literature study as the most used model for describing 3D cultural heritage objects, although our survey did not confirm this. It comprehensively documents digitisation events, detailing both the resulting models and their source data, including provenance, intellectual property rights, technical processes, and location.
• The CARARE Metadata Model aims to integrate archaeological and architectural heritage data into Europeana. It is built on existing standards like CIDOC-CRM and extends the Europeana Data Model (EDM) and introduces possibilities for documenting 3D workflows. Compared to Smithsonian, it offers less extensive support for metadata about the digitisation event.
• CRMdig is an ontology designed to capture metadata related to the production of digitised products, like 3D models and its digital representations. It is an extension of CIDOC-CRM and comparable to the Smithsonian 3D Metadata Model.

For meemoo we are still investigating which model best suits our needs.

Conclusion

Our (ongoing) journey into the world of 3D preservation reveals a critical gap: while 3D digitisation is advancing rapidly, sustainable long-term preservation lags behind. Only few organisations have formal policies or shared standards for formats, metadata, or workflows.

Emerging best practices emphasize the preservation of raw data, the use of open archive formats (such as OBJ, X3D, and DAE), and the importance of paradata and provenance metadata.

But challenges remain. When outsourcing digitisation, organisations are dependent on what metadata, files and file formats the commercial partner wants to deliver. And due to the complexity and a lack of standardisation, many ad hoc practices and workflows are created.

Our study aims to consolidate the available solutions into clear guidelines for 3D preservation.

References

Acknowledgments

We acknowledge the use of The Gemma 3 model from Google to assist with translating text from Dutch to English. The Gemma 3 model was used locally through the LM Studio software with the instruction to "Translate this text to English". The output was modified further to better represent our tone and to correct errors where necessary.

Mirthe Schoofs, Marketing and Communication Officer at meemoo, proofread this short paper.