Obiettivo DIAMANT will develop a very large-scale co-operative repository for media objects (uncompressed film and video) based on a high performance file system, a media database and a distributed object model for media and I/O objects. This base technology can be a solution for application areas with similar demands on large-scale data manipulation, e.g. remote sensing, (3D) medical imaging. DIAMANT will build up a scalable digital film manipulation system by means of commercial off-the-shelf computation hardware (PCs) and high-speed interconnect hardware (SCI). DIAMANT will optimise the overall system throughput by means of a resource scheduler. Optimal load balancing of various manipulation jobs is applied for maximal utilisation of the system resources (computation, network, mass storage and I/O devices).Objectives:Within the last years processor power has grown dramatically, which created together with the parallel implementation of algorithms new (commercial feasible) application domains. One example is the digital manipulation of 35 mm film. While parallelisation enables excellent computing throughput, data handling and distribution, I/O device integration and the human computer interaction are the most critical factors for a commercially attractive film manipulation system. For large data objects handling, e.g. film digitised in high resolution, network bandwidth become a bottleneck in the overall process. This problem description can also be generalised to remote sensing, medical (3D) imaging and other application areas. DIAMANT will develop a solution for high-speed manipulation of uncompressed film, video and HDTV using commercial off-the-shelf computation and state of the art interconnect hardware.Work description:The project aims to develop a distributed system (DIAMANT suite) with a single interface to the user, which they can use to manipulate and process all system objects. The details of storage management, film processing and I/O are taken care by means of a scalable computing cluster, an I/O cluster, a high speed interconnect network, and a media database for the handling of digital film, video and HDTV.Inside the pilot application for DIAMANT, (Semi) automatic digital film restoration, all data objects and computing resources will be controlled by the so called Automatic Restoration Agent (ARA). The ARA directs the operator through the restoration process. The functionality of a hierarchical file system in connection with an abstraction layer for digital film will enable the management of 1-10 TByte of data. Movies can be organised in a shot/scene structure and will be stored as sequence of uncompressed images. Optimisation of the system throughput will be provided by a resource scheduler. Load balancing of different jobs for maximal utilisation of the system resources (computation, network, mass storage and I/O devices) is applied. "Virtual" I/O devices will be integrated into the distributed system by I/O nodes. An object-oriented model will describe attributes and methods of the devices, providing a single and consistent device API. A user interface capable of controlling film manipulation methods and virtual I/O devices will be implemented. Usability tests will be used to further improve the system and tailor the DIAMANT suite to fit the end users needs.The hardware configuration will consist of a cluster of computation-, I/O-, and client-workstations (PC-s) interconnected by a commercial off-the-shelf high-speed network (SCI). Interconnect requirements of DIAMANT will be met by the development of high performance application interface libraries. They will provide an overall network throughput of 200MByte/s through standard APIs (e.g. MPI, Sockets, TDI).Milestones:The overall project goal is to build up a scalable digital film manipulation system, the DIAMANT suite. A first prototype system at PM12 allows functional and usability tests at end users site, a fully equipped system available at PM18 enables performance and data management evaluations. Costs for digital film manipulation will be reduced significantly because of minimised system operator effort and due to the underlying commercial off-the-shelf computation and high-speed interconnect hardware. Campo scientifico natural sciencescomputer and information sciencesdatabasesengineering and technologyenvironmental engineeringremote sensinghumanitiesartsmodern and contemporary artcinematographyengineering and technologymedical engineeringdiagnostic imaging Programma(i) FP5-IST - Programme for research, technological development and demonstration on a "User-friendly information society, 1998-2002" Argomento(i) 1.1.2.-4.2.1 - Concurrent systems Invito a presentare proposte Data not available Meccanismo di finanziamento CSC - Cost-sharing contracts Coordinatore JOANNEUM RESEARCH FORSCHUNGSGESELLSCHAFT MBH Contributo UE Nessun dato Indirizzo STEYRERGASSE 17 8010 GRAZ Austria Mostra sulla mappa Costo totale Nessun dato Partecipanti (7) Classifica in ordine alfabetico Classifica per Contributo UE Espandi tutto Riduci tutto DOLPHIN INTERCONNECT SOLUTIONS AS Norvegia Contributo UE Nessun dato Indirizzo OLAF HELSETSVEI 6 0621 OSLO Mostra sulla mappa Costo totale Nessun dato FH JOANNEUM GESELLSCHAFT M.B.H. Austria Contributo UE Nessun dato Indirizzo ALTE POSTSTRASSE 149 8020 GRAZ Mostra sulla mappa Costo totale Nessun dato HS-ART DIGITAL SERVICE GMBH Austria Contributo UE Nessun dato Indirizzo DIETRICHSTEINPLATZ 3 8010 GRAZ Mostra sulla mappa Costo totale Nessun dato LABORATOIRES NEYRAC FILMS Francia Contributo UE Nessun dato Indirizzo 133, RUE DU THEATRE 75015 PARIS Mostra sulla mappa Costo totale Nessun dato MEDIA CONSULT BUSCHER GMBH Germania Contributo UE Nessun dato Indirizzo STRESEMANNSTRASSE 5 76530 BADEN-BADEN Mostra sulla mappa Costo totale Nessun dato STICHTING NEDERLANDS FILMMUSEUM Paesi Bassi Contributo UE Nessun dato Indirizzo VONDELPARK 3 1071AA AMSTERDAM Mostra sulla mappa Costo totale Nessun dato UNIVERSITY OF SOUTHAMPTON Regno Unito Contributo UE Nessun dato Indirizzo Highfield SO17 1BJ SOUTHAMPTON Mostra sulla mappa Costo totale Nessun dato
