A group of DIY archivists has undertaken an ambitious project to preserve 1,800 rare books by using budget Nikon cameras to capture 902,000 images. This effort has been driven by a passion for preserving cultural heritage and has involved extensive manual work. The team has trained a neural network on Photoshop edits to process 526,000 scans, significantly reducing the time and effort required for digitization. This initiative highlights the potential of combining low-cost hardware with machine learning to achieve large-scale archival projects.

The project began around 2015 when a group of Pakistani friends decided to digitize a large number of out-of-print Urdu books. Many of these books are lithographs, and the team has worked tirelessly to ensure their preservation. The effort has involved a significant number of camera clicks, with one camera alone reaching 576,000 shutter counts. This manual approach has been complemented by the use of machine learning to automate parts of the digitization process, making the project more efficient.

The team has processed a total of 526,000 scans using a neural network trained on Photoshop edits. This has allowed them to handle the vast number of images generated by the 902,000 camera clicks more efficiently. The use of budget Nikon cameras has been crucial in keeping the project financially viable. This approach has demonstrated that even with limited resources, it is possible to achieve significant results in the field of digital preservation.

In India, where there is a growing interest in digital archiving, such initiatives could have broader implications. The involvement of regulators like TRAI could influence how such projects are supported or scaled. However, the focus remains on the technical and cultural aspects of the project rather than immediate policy changes. The success of this project may inspire similar efforts in other regions, emphasizing the importance of grassroots initiatives in preserving historical documents.

The project has faced challenges, including the physical wear and tear on the cameras used and the need for continuous manual oversight. However, the integration of machine learning has helped mitigate some of these issues. As the project moves forward, the team will likely explore ways to further automate the process, potentially reducing the reliance on manual labor. This case study serves as a testament to the power of combining low-cost technology with innovative approaches to achieve meaningful outcomes in digital preservation.