|
|
|
This paper addresses two of the three key themes of infog99 [1], namely: the provision of user access to digitally delivered screen products, and related issues for screen education in the digital environment. It describes research and development work undertaken during 1998 and 1999 by a University of Glasgow consortium, led by the UK's Performing Arts Data Service (PADS), on a publicly funded pilot project concerned with the Networking of Moving Images for University Teaching and Research. The BFI/BUFVC/JISC Imagination/Universities Networking
of Moving Images Project The British Universities Film and Video Council
(BUFVC)
The Joint Information Systems Committee The Arts and Humanities Data Service (AHDS) The remit of the AHDS is to collect, catalogue, manage, preserve and encourage the re-use of digital resources which result from or support humanities research and teaching. These resources make up collections which are either acquired under agreement from individual depositors, project teams, and institutions, or made available to AHDS users through co-operating digital archives and information services. The Performing Arts Data Service (PADS)
Its remit can be summarised thus:
In the early phase of its existence PADS has focussed on the following priorities:
Hence, a pilot project dealing with moving image delivery and dissemination over computer networks falls squarely within the natural area of endeavour of a fledgling service which is attempting to build from nothing to a critical mass of data arranged across a series of collections and involving a range of data types. I hope to illustrate in this article the close coincidence between the particular aims of the project and the wider remit of PADS, so that our involvement in the pilot project can be seen as a natural extension of PADS' core activities. Project Development At this preparatory stage, the project group took some
basic decisions which have underpinned the whole experience.
In particular, they agreed that the pilot should seek to
provide temporary access to some thirty hours of moving
image material sourced principally from the BFI's National
Film and Television Archive (NFTVA). It was also decided to
encourage pilot sites to augment this material with a
limited amount of moving image material sourced locally, and
on which pilot sites could be expected not to need to seek
clearance of rights. Three academic disciplines, with
particular educational requirements for the use of moving
images in teaching and research, and offering user-bases in
arts, humanities and science, were targeted: Film Studies,
Social History and Medicine, and specialist subject advisory
groups were assembled during Summer 1997 to advise on the
selection of suitable content in each of these fields. After
selection, the BFI began, in Autumn 1997, the process of
clearing the rights on the NFTVA material for the purposes
of the pilot. A call for proposals went out to the Higher
Education community, eight bids to serve as pilot sites were
received and four were short-listed, before two - coming
from different parts of the UK and offering divergent
approaches which would thereby broaden the test-bed - were
selected. The Glasgow/PADS consortium was one; the other was
the University of Glamorgan on behalf of the then
newly-formed South
Wales Metropolitan Area Network (MAN), which, like its
counterparts in other parts of the UK, including in
particular Scotland, provides co-ordinated connectivity
between Higher Education institutions in a particular region
over high-speed, state-of-the-art computer networks. (It was
encouraging to see that for once the Celtic fringe had
beaten off all other pretenders and dispelled the prospect
of metro-centricity in the conduct of the project!). As will
become clear, Glasgow/PADS was seen as offering a high
quality, and possibly therefore more expensive, model, while
the South Wales MAN proposed a lower-cost solution using NT
server technology running AVI files into Microsoft NetShow
and NetTheatre applications. Alongside selection of the two
pilot sites, a Project Manager was recruited on a
consultancy basis and located at his home institution of the
University of East Anglia, while a Project Steering Group
was established with representation from the BFI, the BUFVC,
the Project Manager and the two pilot sites. The Manchester
Visualisation Centre (a part of the University of
Manchester's Computing Service) was appointed as the
digitising agency for the project. |
[1]This article is a fuller version of the paper presented in Keynote Session 2 at the infog99 Conference, Treasury Theatre, East Melbourne, Thursday 15 July 1999.
|
|
The pilot sites effectively began work on the project at the beginning of 1998 and initially the project was to have run to the end of that year. However, unforeseen delays of various kinds quickly showed such a timetable to be unrealistic, with the result that the project only came to an end in July 1999. Indeed the final Project Report has recently (September 1999) been finalised for consideration by JISC's CEI, when discussions begin on future initiatives in the field of digital moving image delivery for the UK Higher Education sector [2]. The PADS pilot Realising the importance of documenting and reporting results, PADS interpreted the chief Aim of the project in slightly more detail as follows: to investigate, research, demonstrate, document and report on the most effective management of digitised moving image materials, along with associated metadata implemented to accepted cataloguing standards, so as to facilitate their efficient and user-friendly delivery over fast computer networks in support of high quality teaching and research. |
[2]
Imagination/Universities Network Pilot for online delivery
of moving images (a joint project of the British Film
Institute, British Universities Film & Video Council,
Joint Information Systems Committee), Final report
(August 1999). |
|
Clearly, we were keen to propose PADS as a pilot site because we could identify real synergies between the project's objectives and the special remit of PADS concerning the development of on-line digital data collections for the performing arts. We reasoned that we were already grappling with the challenges of accessioning, mounting, describing and disseminating multimedia objects and instances of time-based media and that, in the interests of interoperability, we should have a stake in this project. We had already, at the time the project was announced, done a great deal of detailed research into technical solutions for time-based media and the wide range of data types encountered across the performing arts. We had specified and installed what we considered to be a highly appropriate and flexible technical architecture. We had also done extensive investigation into user needs through a series of preliminary workshops [3] and had played a leading role in national discussions about the representation of metadata attached to moving image resources, focussing specifically upon Dublin Core attributes [4]. Moreover, we had already decided to try and investigate for ourselves, for the ultimate good of our service, many of the same kinds of issues envisaged by the pilot. We were, and remain, interested in collaborating with others in the UK, as well as tapping into overseas experiences, to progress the idea of a larger-scale moving image delivery service for UK Higher Education. Content Key activities
Technical specification The sub-net on which the Media Server is installed is a test network not used for regular traffic and thus has no impact on conventional users; its use has been donated by the University's Computing Service whose staff have provided significant network support throughout the pilot. The normal connection between this sub-net and the parent building sub-net is by means of a 100Mbps switch, though additional or alternative hardware has been installed for each testing stage of the pilot. Initial testing of the server infrastructure used a half-hour television current affairs programme which had been digitised by the Manchester Visualisation Centre at a number of different bit-rates: MPEG1 at 1.25Mbps, 2.5Mbps and 5Mbps, and MPEG2 at 2.5Mbps and 5Mbps, as well as a sample of locally sourced material from Glasgow University's Media Services Unit coded at three different bit-rates: 1.4Mbps, 2Mbps and 2.5Mbps. Subsequent tests used a range of material encoded by Manchester at 2.5Mbps. In order to provide an interesting technical contrast with our own SGI infrastructure, PADS arranged for the medical material to be mounted separately on a server maintained elsewhere on campus by the University's Department of Computing Science, which is separately conducting an ambitious multimedia research initiative known as the Revelation Project, which has the aim of persuading users in diverse academic fields to store complex data content with them while its educational uses are under development. The Revelation Project uses Sun server technology with a Sun Media Centre for the streaming of video. On the client side, the principal recipients of streamed video have been a number of Pentium II PC desktop workstations, with CPUs ranging from 200MHz to 266MHz , all with 64MB RAM and either Matrox Millenium or ATI graphics cards (with 4MB RAM), 10/100Mbps Ethernet cards, 17 inch monitors and running Windows 95. A PC laptop with an Intel 166MMX processor, 32MB RAM, 2MB Trident graphics on a 12.1 inch screen , and a 10/100Mbps Ethernet PCMCIA card was also used for testing, as were two 90MHz PowerMacs running Mac OS 7.5 and two elderly SGI Indy machines with 133Mhz processors, both with 96 MB RAM and 24 bit Indy Graphics boards. This array of technology provided a broadly based platform for experimentation and allowed well informed conclusions to be drawn with regard to relative performance. The majority of the PCs are located in the Learning Resources Room of the Department of Theatre, Film & Television Studies, a modern, well equipped computing and videotape viewing laboratory for students which has the advantage of offering excellent access by virtue of its proximity to the PADS offices. When at full complement, it has between sixteen and eighteen fairly high-end PC workstations and is in heavy demand from undergraduate and postgraduate students of screen and theatre studies. Video streaming was tested over various types of local network technology in the interests of comparing performance and assessing the most reliable configuration so that a stable environment could be presented to users at the evaluation stage. We demonstrated the delivery of moving images from the server on the following:
Additionally, some comparisons were carried out, using MPEG2 decoding cards, between software and hardware MPEG decoding, and between MPEG1 and MPEG2 content coding. To this end, four different MPEG decoding cards for PC were purchased at prices between £150 and £1000 in order to determine the most advantageous and cost-effective configuration The final project report contains detailed results, analyses and recommendations arising from all these tests. Data and metadata The User Interface developed for the pilot found at the PADS web site, makes good use of a key feature of the PADS digital resource library, namely the browsing and searching facilities enabled by the Hyperwave database system. Users can browse the contents under various classified headings, for example:
User trials and demonstrations The most important large-scale demonstration came just before Christmas 1998 at the Networking Moving Images Conference organised by the BUFVC at Robinson College, University of Cambridge to monitor interim results. This was attended by most of those agencies, project leaders and educators in the UK who recognised the educational value of widespread digital moving image delivery. For this ground-breaking event, a special dispensation on bandwidth and connectivity had been negotiated with UKERNA, the networking authority, so that real-time streaming direct from the PADS servers in Scotland via the UK Higher Education network, SuperJANET to the conference hall in Cambridge could be experienced. Excellent picture and sound quality was achieved and this remains the most comprehensive and impressive live demonstration of the pilot undertaken by PADS. It was always a cornerstone of the project, one especially dear to the BFI and BUFVC, that there should be some use of the digitised materials in live teaching situations, permitting the receipt of feedback from student users in Higher Education. One problem of arranging this on any significant and consistent scale was that course content in university departments is usually determined well in advance of actual classroom delivery, making it difficult to fit teaching very closely to the available footage. We decided to deal with this problem by setting up workshops and seminar simulations that would bring the project and its materials to the attention of bodies of students, both new to university study and nearing the end of their undergraduate careers. We also thought it useful to share our development work as widely as possible with academics not directly involved with the project. These measures, we believed, would, in combination, produce the kind of user response we needed, and in this we were not disappointed. In November 1998, the Hitchcock film study resources were integrated into a lecture on film narrative and mise en scène to over one hundred first-year Film & TV students, who were additionally able to access the same materials for repeat viewing and close analysis under computing laboratory conditions over the next several days. Sixty-seven students completed our Evaluation Questionnaire with constructive responses. As a contrast, a much more experienced group of eleven volunteer fourth-year Honours students of Film and Television participated in a hands-on evaluation seminar which was externally moderated by a professional facilitator. We also held an interactive workshop, for twelve Film & Media Studies academics drawn from the neighbouring universities of Stirling, Strathclyde, and Glasgow Caledonian with a view to eliciting their responses as to quality of performance, how they would envisage using such a resource in their teaching in different institutions, and what features they would expect from any future scaled-up enterprise. We also arranged a live demonstration session, using a mixture of Ethernet and ATM, at Glasgow School of Art for staff of that institution and the nearby Royal Scottish Academy of Music and Drama; this gave us an opportunity to disseminate our materials over ClydeNet, the Central Scottish MAN. Finally, in March 1999, we ran a workshop attended by historians and sociologists from the Glasgow University's departments of Sociology, History, History of Medicine and the Humanities Advanced Technology and Information Institute. A good example of the interactive tasking handed out to participants is that used for the experienced fourth year Honours students. After initial questions about user interface, screen manipulation, metadata layout, and so on, the students were confronted with the following scenarios: 1. Hitchcock 2. Television 3. Searching 4. Browsing |
[3] See Celia
Duffy and Catherine Owen (Performing Arts Data Service),
"Scholars' information needs in a digital age: the view from
the performing arts", The new review of academic
librarianship (1998), 182-84. [4] See Celia Duffy, " Performing Arts Data Service: evaluation of resource discovery metadata for moving image resources": in Paul Miller and Daniel Greenstein, eds, Discovering online resources across the Humanities: a practical implementation of the Dublin core (Arts and Humanities Data Service and the UK Office for Library and Information Networking [UKOLN], 1997), 27-29. |
|
Findings: the users The questionnaire and discussion responses of staff and students were found to differ quite considerably, suggesting that there are varying attitudes to information technology depending on generation, occupational experience and purpose. There is a sense in which students regard the innovative method of delivery with some nonchalance, that is to say, real-time desktop and classroom delivery is nothing more than they have come to expect from a study environment that is increasingly digitally-based. They are, after all, now arriving at university possessing good familiarity already with IT, and many of them come from homes with a computer and internet connection. Students with lower levels of experience of learning methods - in other words, those with no institutional memory - can therefore be somewhat blasé about the digital process even in the face of their tutors' crusading enthusiasm. Many of them are attracted immediately to the flaws and constraints at the expense of the real progress achieved, and ask for facilities which might be seen as going beyond research aids into the pedagogically unacceptable realm of digital "spoon-feeding". Clearly there has to be a balance here, and this tends to be appreciated by the more experienced and confident students. The reactions of academic staff, even those in the relatively progressive field of Media Studies, are rather different. Such staff, with years of experience of unsatisfactory conditions attaching to the discovery, exhibition and re-use of moving image materials, are much more alive to the breakthrough that on-line digital delivery provides, more appreciative of the problems having to be addressed by pilot projects such as this, and more forthcoming about the potential benefits. They are also able to deploy their experience to comment upon priority areas for further development, pointing to, for example, the desirability of a national rights management strategy, the need for pilot studies to move forward to the cataloguing of a critical mass of available data, and better indexing and easy-to-use tools (though here they do tend to reveal a problematic wish to manipulate and re-purpose content - a process not likely to be welcomed by commercial interests!). Film Studies academics, accustomed to familiar constraints on availability and picture quality offered by the 16mm film hire and VHS sectors, as well as inadequate projection facilities in many university teaching spaces, were able to recognise that digital delivery is the support mechanism of the future for screen studies. General findings
Several user interface issues emerged:
Findings: technology On the client side, we found that the powerful PC workstations we used gave the best playback results, though we believe that Apple Macs and SGIs with faster clock speeds than ours would have performed equally well. Time spent carefully optimising configurations for MPEG playback paid huge dividends. There were noticeable differences in quality between supposedly identical machines which were eventually attributed to such factors as the brand of graphics driver used, the range of colour screen resolution and the presence or otherwise of DirectX support. The best pictures were smooth, clear and useable even at the lowest bit-rate tested (1.4Mbps), though some slight "pixelation" was evident on close inspection. The highest bit-rate of 5Mbps gave the best picture resolution but produced less-than-perfect lip-synch, while 2.5 Mbps gave the best balance between picture quality and sound synchronisation. Of the different graphics cards tested, the high-end product recommended by SGI for use with Mediabase, the Optibase Videoplex, gave an extremely good picture especially with SVHS output; two of the three cheaper cards tested also performed very well - to a level which would be more than adequate in a networked PC laboratory environment where volume and cost are factors. Generally speaking, hardware decoding on to an external Reference Monitor provided the best quality image and proved more robust at dealing with "glitches" on the network. Network performance under Contention-based Ethernet at 10Mbps - the method of connection in daily use by staff and students in the building where our trials took place - was uneven and cannot be claimed to have been a success. It varied a great deal depending upon time of day and day of the week, but multiple simultaneous streams were not really possible over this technology even at the lowest bit-rate. Dedicated switched Fast Ethernet was, however, a different matter, suggesting the need to clear bandwidth specifically for moving image delivery purposes. We managed to run up to fifteen active connections of 2.5Mbps at once (a total bandwidth of 37.5Mbps) during user demonstrations with few problems. 16 simultaneous streams at a mixture of bit rates, producing a total bandwidth of 74.86Mbps, was also successfully achieved over dedicated fast Ethernet. However, as many as eighteen simultaneous streams at 5Mbps pushed the bandwidth allocated to Mediabase (90Mbps) to its limit and further connections beyond that were refused; at this level of network load, a PC with hardware decoder gave performance identical to when it was the only workstation running. Results of IP over ATM and native ATM testing at MPEG2 were still under analysis as the project report was being finalised. Minimum specification Taking this point seriously we recommend in the final report that due consideration be given by the project partners to the problem of disproportionality in equipment provision between potential stake-holding institutions. This is almost certainly a matter of existing concern to the JISC in their desire to ensure wide uptake, but we consider it will become grounds for resentment if we find ourselves faced with a dual economy - of "haves" and "have-nots". We therefore propose the following minimum specification for an individual workstation capable of decoding and displaying content within the context of the Glasgow site's technical model. It assumes an appropriate networking infrastructure, and up to date drivers for components. PC Platform:
MPEG 2 software decoders are now on the market, which require Pentium II or greater class systems running at 300MHz or faster with 64MB RAM recommended. For Apple Macintosh, the recommended minimum specification is:
Macs below this specification will not work with Mediabase while a significantly more highly specified Mac is to be recommended. Networking infrastructure is as crucial to the quality of moving image delivery as the minimum specification workstation and this should be taken into account in financial planning. Unfortunately due to the range of technologies available and vastly differing implementations between campuses, there cannot easily be an equivalent minimum specification for campus networking. A wise policy, however, would be to plan a system that allows10Mbps to be streamed to each video workstation with good, or guaranteed, Quality of Service. Copyright The future |
|
Home | First release | Classics & re-runs | Reviews | Trailers | Cast & credits | Search | Comments