Following is an abstract of the presentation by Jörg Houpert:

The inspection and digitisation of 100,000 film reels - lessons learned, what’s next?

In recent years, only a few of the major film archives have digitised more than one percent of their holdings annually. And their analogue film holdings still keep growing. If this continues, there will be no access to today's analogue film heritage for the next hundred years. As a result, this also means that only a few selected analogue films will survive, while all other irreplaceable materials will slowly deteriorate on the shelves due to a lack of resources and attention.

The presentation shows how, especially in the case of large film collections, each individual film reel can be examined for conservation purposes within a few years. A resulting edge-to-edge digital viewing copy in almost 2k-resolution visualises the current state of preservation. The size of the video-files allows for easy network access. Analogue archive material is no longer required for viewing.

A detailed condition report is generated largely automatically. The report clearly shows the defects and imperfections: decomposition due to vinegar syndrome, dye fading, shrinkage, plasticiser leakage, mould damage. Time-based events such as bad splices, scratches, defective perforations, etc. are documented with high-res thumbnails. The condition report is also good to determine the required effort for physical preservation.

The machine-readable condition report data is mapped to the client’s collection management system to be used for search and statistical analysis. The film degradation figures can be fed into an open-source risk modelling for motion-picture film conservation, so that the film canister can be stored even more appropriately. This means drastically reducing energy costs and still better protecting the materials. A new method is presented that detects virtually identical scenes in hundreds of thousands of filmreels.

This will make it much easier in the future to find even small differences in multiple versions of film elements and compare them frame-by-frame. In researching film generations, the method also supports the material-based reconstruction of genealogy with all permutations, cuts, and additions. Thanks to AI technologies, in-depth content indexing is finally a reality. However, for many archives, it is important to retain data sovereignty and use the processes locally. Practical use cases will be presented.