From ruins to reconstruction: Harnessing text-to-image AI for restoring historical architectures

Data Availability: The datasets created and/or analyzed during the current study are not publicly available, but are available from the corresponding author upon reasonable request.

Acknowledgements: This research has previously been presented at the 2nd International Summit on Civil, Structural and Environmental Engineering (ISCSEE2024) held in Florence, Italy, on March 18-20, 2024. Extended version of the research has been submitted to Challenge Journal of Structural Mechanics and has been peer-reviewed prior to the publication.

The preservation of cultural heritage has become increasingly important in the face of conflicts and natural disasters that threaten historical sites worldwide. This study explores the application of artificial intelligence (AI), specifically text-to-image generation technologies, in reconstructing heritage sites damaged by these adversities. Utilising detailed textual descriptions and historical records, this study employed AI to produce accurate visual representations of damaged heritage sites, bridging the gap between traditional documentation and modern digital reconstruction methods. This approach not only enhances the architectural design process across various disciplines but also initiates a paradigm shift towards more dynamic, intuitive, and efficient heritage conservation practices. The methodology integrates data collection, iterative AI-generated image production, expert review, and comparative analysis against historical data to evaluate reconstruction accuracy and authenticity. By integrating AI with traditional preservation practices, this study advocates a balanced approach to conserving cultural legacies, ensuring their preservation and revitalisation for future generations. Preliminary findings suggest that AI-generated imagery holds significant promise for enhancing digital heritage preservation by offering novel approaches for visualising and understanding historical sites. These findings also highlight the need to address ethical, technical, and collaborative challenges to enhance the precision, reliability, and applicability of AI technologies in the field of cultural heritage. This study contributes to digital humanities and archaeological conservation, demonstrating AI's potential to support and complement traditional heritage preservation methods and suggests a pathway for substantial methodological evolution in the field.

The authors received no financial support for the research, author-ship, and/or publication of this manuscript.