Sean Rippington is Digital Archives and Copyright Manager at the University of St Andrews

Photogrammetry – taking overlapping photographs of an object and converting them into 3D digital models – has become business as usual for the University of St Andrews Libraries and Museums. Driven by the push to provide more, different, and better types of online access to our collections for teaching and learning during the pandemic, we now have over 170 examples on our IIIF-based Collections site. Many of the models have already been used in our Exhibit teaching and engagement tool, developed to provide new, curated, almost tactile encounters with digital objects. Some of our favourites include:

As a reader of this blog, you’re probably already thinking – these are great, but how are you going to preserve them? Here’s our approach so far.

A good first step was to read the existing DPC guidance on general 3D data preservation, including: 

I think what becomes clear after exploring these is the sheer variety of different types of 3D content, and the emerging nature of the guidance. While that can be a bit overwhelming, the approaches described are based on familiar concepts and solutions. 

There is also a professional community growing around the specific issue of preservation of digital 3D data at There I found some very useful contacts for specific problems, and came across the most detailed guidance for preserving 3D objects in a library and museums context  - Metadata Requirements for 3D Data (Blundell, Jon, Clark, Jasmine L., DeVet, Katherine E., and Hardesty, Juliet L. 2020) – a preprint from the forthcoming 3D Data Creation to Curation: Community Standards for 3D Data Preservation (Moore, Jennifer, Adam Rountrey, and Hannah Scates Kettler. 2022. Association of College & Research Libraries.) 

From all of the above, we put together a series of principles to guide our approach to photogrammetry preservation: 

  • Keep tiffs of the source object (additionally keeping raw files might be desirable but has significant storage implications). Tiffs should be 8 bit and LZW compressed to reduce storage requirements while maintaining acceptable quality.

  • Keep point cloud data, ideally as a text file, or in some other open format. We may need to have a text file explaining how these relate to the tiffs, as file locations in the metadata may no longer work.  

  • Keep a record of any control points system documenting the model’s relationship to real-world measurements, ideally in some open format such as a text file.

  • Keep the .obj file of the 3d object – other formats may be suitable, but we already produce this in our workflow and it is listed as an 'acceptable' file format by LoC Note that this may require preserving a separate image texture file. 

  • Generate and keep a project report (in our case, a pdf generated by Metashape), documenting some of the technical settings and fully quantifying any errors  

  • Document hardware and software used, and workflow. This could be saved as a text file. Note that workflow is iterative and may change from object to object.  

  • Ideally all this information will be saved with the item record in our repository. 

  • .glb files will continue to be used as the access version of the object as they are small, quick to load, and work in the IIIF viewer

Of course all the usual principles for digital preservation – including backups, use of unique identifiers, generating and storing common preservation metadata – also apply. 

The obvious downside to this approach is that we are keeping rather a lot, which has consequent costs and environmental impacts. However, photogrammetry technology is constantly improving, and we’re anticipating that the emerging ability to re-render 3d objects from our existing data in new and better ways will outweigh the potential costs of having to rescan the objects in the future. 

Preserving the underlying data also allows users to better understand and explore the provenance of the 3D objects, making them more credible as items for academic citation. 

We’ve also found that the bulk of the data generated by size per object is in the tiffs (and raw files, if you keep those). While these are large files, they require about the same storage per object as a digitised rare book or large manuscript, so are not particularly demanding in the content of our existing digitisation work.  

We will of course keep our approach under review as our understanding of good practice matures.

I hope that serves as an interesting starting point for any other institutions looking to preserve outputs from photogrammetry – I'd be interested to hear more about similar work happening elsewhere. 


#1 Emma Hancox 2021-12-01 15:10
Hi Sean,

This is really useful for us at Bristol. As you might have seen, I wrote a blog about how we've started doing some work in this area. It's great to see how you've been approaching this and I'll have a look at the resources you've highlighted.

#2 Sean Rippington 2021-12-06 12:03
Hi Emma - good to hear from you! Yes I was interested to read your post too, it's good to see that we're working on similar things. If it's ok I'll be in touch in the new year to compare notes!

Best wishes,


Scroll to top