How Melbourne's Visual Archive Got Swamped by Duplicate Images — and What Happened Next

Melbourne's network of public cultural institutions is sitting on a storage problem that has been quietly compounding since at least 2014. Duplicate images — identical or near-identical digital files catalogued separately, sometimes under different metadata, sometimes not catalogued at all — have bloated archives at institutions from the State Library Victoria on Swanston Street to the City of Melbourne's own digital collections unit. The reckoning, archivists and collections managers have said for years, was always coming.

It matters now because several of those institutions are mid-way through infrastructure upgrades. The State Library Victoria completed Stage 2 of its $88.1 million redevelopment in 2019, and the digitisation pipeline that accompanied that project accelerated file ingestion well beyond what existing de-duplication protocols could handle. Subsequent budget cycles, including pandemic-era cuts to the Victorian cultural portfolio in 2020–21, reduced the staffing headcount available to audit what had been ingested. The result: collections that are larger by volume but harder to search, harder to licence, and harder to maintain.

How the Duplication Problem Accumulated

The roots go back to the mid-2000s, when Victorian public institutions began digitising in earnest under programs aligned with the federal government's National Collaborative Research Infrastructure Strategy. Photographs, maps, architectural drawings, and ephemera were scanned across multiple workflows, often by separate contractors working to different file-naming conventions. The same image of, say, the Flinders Street Station façade might enter a collection three times: once from a donor's physical print, once from a microfilm copy, and once pulled from an earlier CD-ROM publication.

The problem compounded when institutions began sharing assets across platforms. Museums Victoria, which manages collections across Melbourne Museum in Carlton, Scienceworks in Spotswood, and the Immigration Museum on Flinders Street, integrated its catalogue into the national Collections Australia Network in the early 2010s. That integration flagged hundreds of cross-institutional duplicates, but the resources to act on the flags were not consistently allocated. A 2018 internal sector review — circulated among Victorian collecting institutions but not publicly released — reportedly recommended a shared de-duplication protocol and a joint image registry, neither of which was implemented in full before staff turnover and changing ministerial priorities intervened.

Commercial pressures sharpened the issue. When institutions began selling print-on-demand and licenced digital reproductions through platforms like Redbubble and their own online shops, duplicate listings created genuine consumer confusion and, in some cases, double-charging. The State Library's image sales service, operated through its commercial arm, had to manually reconcile catalogue entries on multiple occasions between 2020 and 2023.

What Replacement Efforts Have Looked Like on the Ground

The push toward systematic duplicate image replacement — the process of identifying a canonical version of a file, retiring redundant copies, and updating all metadata references to point to the single authoritative record — gathered pace after Creative Victoria signalled in its 2023–2025 strategy document that collections resilience was a funding priority. Several institutions applied for grants under the Cultural Infrastructure Fund, with at least two projects specifically addressing digital asset management.

The Australian Centre for the Moving Image on Federation Square has run a structured asset audit since late 2024, using open-source perceptual hashing tools to flag near-duplicate frames across its film and photography holdings. The NGV on St Kilda Road has piloted an AI-assisted cataloguing layer across its international prints and drawings collection, partly to surface redundant acquisitions before they are fully processed into the permanent record.

Neither project is complete. Sector observers note that the tools are now cheaper and more accessible than they were five years ago, but the human review step — deciding which version of a duplicate is canonical, and ensuring linked metadata is corrected — remains labour-intensive. A single major photographic collection can contain tens of thousands of flagged pairs requiring human sign-off.

For institutions still early in the process, the practical next step is establishing an image registry shared across at least the major State-funded collecting bodies. That idea has been on the table since the 2018 sector review. Whether the current round of Cultural Infrastructure Fund grants moves it from recommendation to reality will depend on what the Victorian Government allocates in the 2026–27 budget, expected in late October.