An unsupervised method to increase film restoration’s sustainability for small cinematheques

Abstract

Digital film restoration plays a crucial role in preserving audiovisual heritage, offering a unique potential to enhance image quality and accessibility. However, current approaches face significant challenges in terms of technological and, even more critically, economic sustainability. This study proposes an automated, unsupervised method to optimize film restoration workflows, reducing costs and resource consumption, particularly for small cinematheques with limited budgets.

Firstly, we analyze the meaning of the term “sustainability”, form its first appearance in the “Brundtland Report” to some new declinations that appears in recent years able to match different field of studies. Secondly, we analyze the limitations of traditional restoration software, such as Da Vinci Resolve, Phoenix Finish or Diamant, which rely heavily on operator-driven, task-specific processes that are resource-intensive. In contrast, unsupervised restoration technologies leverage automation to improve efficiency and scalability, streamlining tasks such as scratch removal and color grading. These operations, traditionally performed “by eye” and “by hand” based on the operator’s subjective judgment, can be significantly expedited and standardized.

The article evaluates the sustainability of the use of Spatial Color Algorithms as unsupervised kick-off for color film restoration, from economic, social, and technological perspectives. Economically, automated workflows can help reduce operational costs by optimizing time and human resources. Socially, the democratization of restoration technologies encourages greater community involvement in the selection and restoration of films, while training initiatives equip young technicians with adaptable skills to meet future innovations. Technologically, open-source software is a critical tool for ensuring accessibility and resilience in film restoration processes.

To illustrate these points, we present two case studies. The first is the film “Funerali delle vittime di Piazza Loggia di Brescia” (The Funerals of the bombing of Piazza Della Loggia in Brescia), shot by Mario Bertoli in 1974 and currently preserved in the archive of the “Casa della Memoria” in Brescia. This historically significant document allows for a comparison of traditional and digital restoration techniques in terms of cost and quality. The second case study focuses on “La Lunga Calza Verde”, a 1960 animated short film directed by Roberto Gavioli and based on a script by Cesare Zavattini. Created for the centenary of Italy’s unification, it was entirely restored using the proposed automated digital process.

These examples demonstrate the feasibility and advantages of an unsupervised approach and highlight how it can enhance the sustainability of film restoration for small cinematheques.