New lease of life for archive film footage

Work to develop new methods of digitally restoring archive film footage could breathe new life into old recordings and improve on the quality of the originals.

The new approach aims to make the whole process cheaper, faster and more effective than current methods. The work could also dramatically improve public access to previously unavailable historic, artistic and cultural material.

Many historic events are captured on celluloid but its fragile nature means we are gradually losing vital aspects of our heritage. Video copies are just as vulnerable over time and they also degrade the quality of the original recording, particularly with multiple copying.

The 3-year initiative is being carried out at the University of Surrey, with funding from the Engineering and Physical Sciences Research Council (EPSRC).

Traditional approaches to restoring celluloid film mainly rely on complex techniques carried out by skilled operators. They are labour-intensive, time-consuming and very expensive. This has limited the amount of restoration work that has been undertaken to date. In addition, a lot of archive film is too fragile to be manually restored, even though it requires urgent attention.

The new project aims to tackle these problems by developing advanced image analysis and processing techniques suitable for automation by a computer or dedicated hardware. Operating at the pixel or sub-pixel level, these techniques include
motion estimation, statistical processing and application of principles of photographic
image registration. They could offer unprecedented precision and accuracy, and substantially reduce the need for human intervention.

Building on recent advances in film restoration, the researchers will probe well beyond the current state of the art. The project team is being led by Dr Theodore Vlachos of the University’s School of Electronics and Physical Sciences. “The techniques we are exploring may vastly increase access to films of major historical, cultural and artistic value”, says Dr Vlachos. “Ultimately, our work could benefit public service and commercial film archives, which are experiencing growing demand from new multimedia and broadcasting outlets”.

The work will target the key impairments of film flicker and unsteadiness, both of which interfere substantially with the viewing experience. To correct flicker, a new approach will be explored based on non-linear modelling of film exposure inconsistencies. To correct unsteadiness, the team will pioneer the use of higher-order motion models that are more realistic and maximise visual quality.