DIY archivists push budget Nikons to 902,000 clicks to save 1,800 rare books (www.tomshardware.com)

A group of Pakistani friends initiated the Ibteda Digital Library project in 2015 to digitize out-of-print Urdu books, utilizing budget-friendly Nikon cameras to take over 902,000 photos over a decade. Lacking formal resources, they resorted to a manual process that involved extensive photo editing to ensure archival-quality copies. Their work is particularly significant in light of current practices by AI companies that often digitize rare books without regard for preservation, raising questions about access to cultural knowledge. As their project faced challenges, including the unique complexities of Urdu script and variability in book layouts, one member developed a machine-learning approach that streamlined post-processing. By leveraging their manually edited images to train a neural network, they created a tailored solution that improved accuracy in capturing the nuances of different books. Although the process still requires initial manual calibration, it represents a potential path forward for similar digitization efforts, combining cultural preservation with innovative AI techniques. The project underscores the importance of responsible digitization in safeguarding literary heritage while exploring the intersection of technology and literature.