FIAT/IFTA World Conference 2026

Artificial intelligence has rapidly become an important tool in contemporary film restoration. Machine learning systems now assist with chroma reconstruction, density balancing, spatial inference, defect detection, and reference-guided restoration. Despite these advances, current AI methods still encounter significant limitations when dealing with the physical and statistical characteristics of photochemical film and scanned archival materials. This presentation examines the principal technical boundaries of present-day AI restoration systems and identifies the research directions most likely to overcome them. The first major limitation is the loss of high-frequency information: many generative models operate in compressed latent representations that suppress stochastic micro-signal, producing over-smoothed textures and synthetic noise patterns. The second limitation is temporal consistency: motion picture restoration requires stable behaviour of fine detail across frames, yet current systems often produce flickering textures and motion-dependent smoothing. The third limitation concerns acquisition-specific signal characteristics: film stock response curves, dye fading behaviour, scanner optics, and laboratory processing artifacts vary enormously across archival materials. The presentation then examines emerging approaches that may address these limitations. Pixel-space diffusion offers improved microstructure reconstruction but at extreme computational cost. Wavelet-based diffusion frameworks reorganise image information into frequency bands while preserving spatial localisation. Grain transport - treating film grain not as noise to be removed but as a structured stochastic signal to be probabilistically preserved - represents a further frontier. Drawing on the presenter's ongoing experimental work in machine-learning-assisted film restoration, the presentation argues that the most promising path forward lies in hybrid workflows that combine machine learning inference, classical frequency-domain processing, stochastic grain modelling, and expert human supervision.

Digital technologies have been revolutionizing audiovisual production, providing benefits in image and sound capture, mastering, and editing. The transition from analogue carriers to digital systems impacts operational costs and, crucially, long-term access. In an increasingly digital world, the dual pressure of born-digital production and the migration of analogue collections for dissemination and physical media conservation is intensifying the demand for digital preservation strategies. Given the challenges and complexity of digital preservation, the present study proposes a maturity diagnostic of the Centro Técnico Audiovisual (CTAv) through a hybrid methodological approach. It employs the DPC Rapid Assessment Model (RAM) to evaluate organizational and service capabilities, in conjunction with the NDSA Levels of Digital Preservation, focusing on the technical integrity and storage control of audiovisual assets, supplemented by field interviews with CTAv’s staff to evaluate the organisational and technological infrastructure, as well as the sustainable resources framework. The objective is to identify and analyse the principal challenges to the implementation of a sustainable digital preservation strategy within the CTAv, a public institution under the Secretariat of Audiovisual of the Brazilian Ministry of Culture that holds a significant national collection comprising over 24,000 items, including news reports, short films, documentaries, and animations. It constitutes an audiovisual heritage of high historical, documentary, and artistic value to Brazil. As the CTAv collections are immersed in the digital era of constant change, establishing a sustainable digital preservation framework is essential to safeguarding this heritage, ensuring it remains accessible for future generations.

LUPA-UFF is a university lab created in 2017 and dedicated to orphan films from the State of Rio de Janeiro (Freire, 2020). Within its collection, which consists mostly of home movies, the Esdras Baptista collection stands out as one of the most important with more than 1,000 film reels. Esdras Baptista's collection includes 16mm footage he shot while responsible for television coverage of the Rio de Janeiro government (until 1960) and the state of Guanabara (after 1960, when Brasília became the capital and Rio de Janeiro a city-state).
Although assembled by a cameraman who was a government employee (so, a personal collection), and not by a broadcast company or the government itself, the Esdras' collection of 16mm black and white reversal silent films, in its essence, does not differ from other Brazilian TV film collections from the 1960s. What sets it apart is the fact that the Esdras' collection has been preserved, digitized, and disseminated by a regional university archive such as LUPA, which represents a unique case in Brazil.
In addition to presenting this collection, the objective of this presentation is to discuss the practices of access and dissemination of the Esdras Baptista collection by LUPA-UFF, including its reuse and reinterpretation,through collaboration between students, professionals, and professors, within the scope of the undergraduate degree in Cinema and Audiovisual and the postgraduate program in Cinema at the Fluminense Federal University (UFF). This collection of 16mm films, originally produced for television, has become an important tool for teaching and research in the history of Brazil and, in particular, Rio de Janeiro. In addition, the Esdras Baptista collection at LUPA-UFF provides the possibility of access to and dissemination of television archives, which often face obstacles due to the lack of public policies for television audiovisual preservation and the policies of commercial companies (Tillmann, 2021).