Stuart Gall, System Administrator at Greek project partner PPV, explained "No matter how film is stored, its quality deteriorates over time. Reinstating the image quality is a very time-consuming process that very few clients have the budget for. We think that PICASSO will provide a fast alternative."
UK-based Pandora is the lead partner of the project and an expert in the manipulation of images for broadcast. It has developed an initial software approach to the solution using algorithms that first analyse and then fix faults frame-by-frame, allowing real-time film restoration. The traditional method of restoring by hand took at least 30 seconds per frame, easily driving the costs of restoring a film to over a million dollars, which exceeds most budgets.
The innovative film restoration algorithms will also be able to tackle scratches and imperfections on new films. This will save the need for additional days shooting and will extend the potential market for PICASSO's system.
"Filming on location is very expensive. If the film is damaged in any way - in the camera or during development or editing - a client will do almost anything to avoid the need for a re-shoot of the scene" said Gall.
The simplest example of an algorithm for film restoration is one that looks at the digital level of every pixel for each picture element of a film. If there is damage, then the system can restore it by taking an average of the four neighbouring pixels. The actual PICASSO algorithms are much more complex as they copy original film, analyse problems and then produce high quality results.
PPV is conducting field trials that will include tests on material from the Greek National Film Archive. Once the trials are complete, work can begin on producing a marketable hardware solution.
Peter Stansfield, the project co-ordinator, described how huge the potential market is "due to the 'bandwidth explosion' demand for archived film material. The global revenue is likely to be in excess of £500 million over the next three years."