|[Partners] [Contact] [French] [Top] [Previous] [Next]|
The continuing explosion in the number of broadcast, satellite and cable channels in operation, coupled with the current development of video on demand, video and DVD publishing, and multimedia services, are creating an extensive and exacting demand for programme material. This demand has made European broadcasters, archivists and film makers the custodians of a huge wealth of moving picture archive material. But the level of exploitation of moving picture archives is currently limited by the high cost and lengthy processing time required to restore archive material to meet the expectations of viewers. Efficient and cost-effective restoration is hence a key to large-scale exploitation of television archives, such exploitation being itself necessary to guarantee the continuous investment in preservation of the audio-visual heritage.
Documents stored as 35mm or 16mm film are very often affected by deterioration due to ageing and/or improper storage and handling conditions. When transferred to 4.2.2 digital video, the full-bandwidth broadcast-quality format, the quality of the copy is very often unsuitable for direct exploitation.
Programmes stored on video tapes are prone to the same problems, but there is the additional complication that the video storage formats have become obsolete. It is generally difficult to maintain the playing equipment in good working order and it is sometimes almost impossible to find replacement parts. In consequence, artefacts resulting from difficulties in playback often affect the programmes.
In both cases, it is however possible to address these impairments in the video domain, after playback. The ACTS AC072 AURORA project succeeded in proving that it was technically possible to raise the quality and efficiency of the digital restoration of archive programmes. However progress was still needed in the automation and efficiency of the restoration process, extending coverage of the wide range of common defects, improving quality of the results, and including HD video processing.
The BRAVA project aims at developing further the results of the AURORA project, to ensure the widest possible access to valuable archive material by significantly enhancing the efficiency of the video and film programme restoration process and preparing the restored material for dissemination via multi-resolution digital video broadcasting standards.
The main project focus will be to develop the prototype of a system to significantly reduce the cost of archive restoration, enabling archive holders and programme makers to open up and exploit a wealth of material not previously affordable.
A reduction in restoration time will be achieved through the development of algorithms which are capable of analysing the image and optimising the restoration process by controlling each of the individual stages used in the restoration. The prototype that will replace many of the current manual video restoration processes by providing real-time processing of a wide range of defects and will have automated features to provide restoration at or close to real time operation, hereby meeting the needs of professional archive restorers. This will be realised through the development of new techniques which will be evaluated on a hardware/software test bed. The project will improve picture quality by maintaining greater detail (bandwidth) of original film archives than is currently achieved and ensure that more of the fine detail in the original picture is processed during restoration, providing the required quality for broadcasting. Several hours of programme material will be restored at the Users' sites using the developed system, and a complete evaluation of the performance will be made on this occasion.
The key objectives for the project are :
The BRAVA system will be exploited in a broadcast environment for restoring TV archive programmes. Three different tasks will be made by operators : programme expertise, video restoration, audio restoration. For reasons of efficiency, these three different tasks will be, in principle executed by three different operators, but the system should allow for one operator to execute all of these operation at the same time in a unique environment.
As a general rule, each of these operations will take place linearly, and in real-time. This is strictly required for large programmes durations to be restored at affordable costs.
The operator will play the programme in real-time through analysis or processing equipment, and adjust on-the-fly the parameters. At any time, the operator will be able to stop the operations, come back to a specific point, change a setting or correct a problem, and resume operations.
The system will be built around a central media store. The programmes to be restored, or parts of them, will be stored and visible as files.
A lightweight tool for assisting the expertise, will assist an operator in making an assessment of the programme. This tool will run either directly from the files, or as a slave to a VTR. In both cases, it will run a real-time on-the-fly analysis of the parts of the programme that are played by the operator. The operator will be able to log down comments, take note of specific problems (a choice list will be provided), and navigate freely through the programme. The result of the programme evaluation will be a composite report, with the operator's report, completed by the automated analysis results. This report will be used for deciding on the opportunity and target quality of the restoration.
The central media store will provide Fibre Channel access to the different programmes, or parts of. These files will be visible as video and audio files. The different processes (loading, analysis, real-time restoration, video editing, audio restoration, off-loading), will be able to run independently, on a non-exclusive file access basis.
The video restoration will be split into two processes : the real-time video process will play from the files the video and associated audio into the external Real-Time Restoration Unit (RTRU), and the results will be stored again into results files.
A Non-Linear Editing (NLE) tool will be used to possibly re-edit the programme when real-time processing does not help (several images missing...), to control navigation through the programme, to apply processing plug-ins if necessary, to give access to a paint tool, to execute basic audio/video re-synchronisation.
The results of the analysis, completed by additional analysis done in real-time during the work, will be used to give access to functions such as "go to next/previous shot/problemX/problemY", or "apply setting to current shot".
A separate audio editing tool will have access to the audio files and video files, and will allow an in-depth re-synchronisation, will allow audio restoration in real-time through the use of plug-ins.
The system will provide for :