Session	FIRST – Film Restoration and Conservation Strategies Towards Standards and Recommended Practices in Archival Films, Digitisation for Conservation, Restoration and Access
Presentation Coordinated By	Nicola Mazzanti FIRST Project Manager Paul Read FIRST Consultant
Abstract . .Presentation .. Download Powerpoint Presentation.. ..Download Paper. .Speaker Bios

ABSTRACT

Thanks to a complex and articulated program of activities, meetings, workshops and research, scheduled over a two-year period, the Project FIRST aims at creating the conditions to produce sets of standards or "recommended practices" for the digitalisation of archival materials for the some specific uses - mainly for conservation, restoration, access/low resolution distribution.After having met the support of other important partners as IFTA (International Federation of Television Archives) represented in the Consortium through RTBF (Radio Television Belge de la Communauté Française de Belgique) and ORF (Osterreichischer Rundfunk, Austria), INA (Institut National de l'Audiovisuel, France), BELGACOM (the leading Belgian telecommunication company), EMF (European Multimedia Forum, UK), ACE submitted the project to the IST Programme in April 2002. The project was finally approved and started its work in July 2002, under the title of FIRST - Film Restoration and Conservation Strategies.

There are five main areas of interest for the FIRST project: 1) Archival film digitisation, 2) Restoration by digital processes for different uses, 3) Storage technologies and policies of digitized archives, 4) Cataloguing and retrieval of digitised film archives with focus on on-line management and retrieval, 5) Strategies for distribution and access of digitised archive material with focus on on-line management and delivery.

This report will concern the final Guidelines and Recommended Practices at the conclusion of this 2-year study.

_______________________________

PRESENTATION
"Film Archives on the Threshold of a Digital Era",
Highlights from the FIRST PROJECT's FINAL REPORT
by Nicola Mazzanti & Paul Read

In the following, we highlight some of the conclusions of the Project FIRST, as contained in its FINAL Report, published under the title "European Film Heritage on the Threshold of the Digital Era" by the Film Archive of Belgium.

Although the Report refers to the European Heritage, where some of the conclusions refer specifically to the European situation in 2004, most of the conclusions - we believe - have a wider impact, and can be referred to most countries, and most film archives and collections.

FIRST: WHAT IS IT?

First of all, a short reminder of what the Project FIRST is, or actually was, since it came to an end on 30th June 2004.

FIRST is an acronym for the project "FIlm Restoration and conservation STrategies, an "Accompanying Measure" of the EU Programme IST. It was not designed for the research and creation of new technology, but aimed at the evaluation and discussion of strategies and polices to assist the digitisation of European FILM collections.

FIRST was the first EU project to focus its attention on the digitisation of film: features, shorts and documentaries, as well as broadcasters' collections, which include, but are not primarily, film. FIRST was concerned with the digitisation of film collections for re-use in a wide variety of distribution and delivery methods: including content research, Broadband Internet, DVDs, D-Cinema, and Broadcast. As with all EU-funded projects, FIRST had a precise lifetime of 24 months, and developed its activities around a number of events, conferences, meetings, and workshops in Europe and abroad.

FIRST was originally devised and proposed by ACE, the Association of European Film Archives, and the Royal Film Archive of Belgium, its primary promoters and organisers. Other members of the consortium were Belgacom, the Belgian telecom company, the European Multimedia Forum, and three broadcast archives, representing the International Federation of TV Archives: the French speaking Belgian television, RTBF, the Austrian State television, ORF, and the French Institut National de l'Audiovisuel, INA.

The project's general aims were:

• Gathering and diffusing information.
• Facilitating exchanges among different actors in the field.
• Improving awareness on the issues at stake.
• Assessing the "State of the Art".
• Outlining future scenarios and strategies.
• Identifying "bottlenecks" and blocking factors in digitization.
• Issuing guidelines.
• Issuing recommendations for future research.

The main outcome are two basic reports, published roughly one year apart, in 2003, a "State of the Art Report", describing the situation in this domain as per early 2003, and, and the Final Report, containing conclusions, guidelines and recommendations for future research in June 2004. The Final Report appears as a booklet containing an Executive Summary in a printed form, and a CD which includes the full text of the Final Report, plus the State of the Art Report in it's original form, published one year earlier. Clearly some of the information contained in the 2003 Reports has become outdated, but we believe they still have an interest, and the final report describes the changes that occurred over the year. The Final Report can be obtained in printed or electronic form through the project's website, which has operating for over two years, at www.film-first.org.

These reports, and the activities of the project, are addressed at a wide range of institutions, bodies and individuals interested in this field including, but not limited to, the following:

• Film archives, both private and public.
• The world of film production.
• Film and video distribution chains.
• Copyright and film catalogue holders.
• The IT sector, including providers of hardware and software solutions.
• The digital storage industry.
• Advanced European research centres.
• Telecommunications and the ISP sector.
• The educational sector.
• Academics and researchers.
• Providers of digitisation and restoration services.
• Broadcasters and their archives.
• The multimedia industry.
• National governments and EU Institutions.
• Funding organizations.

The work within the project was carried out by 5 workgroups, formed by the partners with the participation of other bodies and experts. These workgroups focused on Digitisation, Restoration, Data Storage, Metadata, and Access-and-Delivery.

As can be seen from the choice of partners, the Project was designed to take a very comprehensive approach. It recognized that the vast European film collections are held by both public and private bodies, that the cost and complexity of the process of digitising will be equally vast, and that the range of applications and uses for the film materials after digitisation is already extensive, and is extending.

FINAL REPORTS: SOME HIGHLIGHTS

The first action of the Project was an analysis of the digitisation process, to better understand the various steps, the technical choices, the bottlenecks and the opportunities available, and tried to keep into account the different uses and users of the digitised content.

The resulting picture is of an extremely complex chain, both from the perspective of technical solutions (some solutions exist and some still need to be developed), and strategies (which are largely so far undefined).

Diagram 1 shows a simplified version of this Digitisation Chain: yellow indicates processes in the analogue domain, and light blue shows the digital processes. The blue arrows indicate where processes are highly interrelated, and are dependent on basic strategic choices, and on the uses, and the users of the final digitized content.

Diagram 1

As the diagram shows, an important part of the process must take place long before the content is actually transferred in the digital domain, and some of the most critical deeply interconnected decisions influence the whole process. These strategic decisions, for example the mode of distribution and delivery, will decide the resolution and technical characteristics of the digitised content, or the selection of the content to be digitised, and these are positioned all along the chain. In consequence any strategy of digitisation must take all the steps of the process into consideration. This contributes to the complexity of the analysis, and leads to a multiplicity of critical decisions.

Although the FIRST Project was not able to precisely quantify the costs involved in the digitisation of large collections of film materials, the Final Report contains some general cost and cost structure considerations.

• The digitisation process is a high-cost investment
• Ongoing costs are high: requiring qualified staff, and due to migration strategies, technical obsolescence, etc.
• Cost structure is very rigid, in the sense that ongoing costs must be sustained with a degree of consistency: migration, maintenance, upgrades of software and hardware must be performed regularly, when needed, with the risk of loosing the whole or a part of the digitised collections.
• Cost structure is extremely complex: costs depend on many complex variables and on their interaction (uses of the content, quality for digitisation, type of content, categories of users, etc.)
• Digitized content does NOT replace the analogue film original: the costs for digital are in addition to activities in the analogue domain (the film cannot be discarded just because its content is digitized for today's access)
• Film collections' more 'traditional' activities, requiring traditional film handling and management skills, need to be strengthened, not reduced.

In conclusion, there is a large consensus of opinion that due to the high costs involved, any digitisation strategy must be designed with a medium term perspective. Short-term policies are doomed to be more expensive because material is more likely to undergo repeated digitisation as soon as a new standard, or a new market leader, arises.

Other conclusions contained in the Final Report include:

• The high costs of digitisation impose a far-sighted approach, looking at least in the medium term, the risk being that film materials need to be digitised again as quality demands increase.
• Digitisation is NOT a preservation strategy, at least not yet: film remains the safest carrier for high quality, high value film content.
• Digital preservation as a technology IS needed, but so far principally to preserve the digitised content (due to the high added value produced during digitisation), rather than as an alternative to proper conservation of the film originals. It is recognized that this could change with the loss of film stocks from the market.
• There is a crisis in the cinema and film may longer be an effective access medium - particularly for archive film - and imposes a need to research new solutions and approaches.
• New views on long-term film conservation have been presented by a number of organisations, and it is nowadays clear that film, if conserved under proper storage conditions, has a life expectancy that exceeds that of any digital medium, and at a significantly lower cost.
• The advent of new digital delivery modes & channels (DVD and HD-DVDs, Broadband Internet, D-and E-Cinema, HDTV, iTV, VOD…….etc) offer unprecedented opportunities for film collections to provide access to their holdings. All are worth exploring, particularly in a context where the traditional theatrical screening model is not responding to a growing demand anymore.

In other words, the FIRST Final Report stresses the fact that Digitisation can become an effective strategy to improve Access, provided that new access and delivery models are researched and implemented. It is clear and inevitable, that any access strategy for the time being, needs to be 'hybrid', i.e. to include both analogue and digital distribution.

CHALLENGES FOR FILM ARCHIVES

The FIRST Final Report highlighted the most relevant challenges for the Film Collections and Archives community, and what is required to 'switch' into the Digital domain for Access:

• New approaches and strategies for Access and Preservation
• Re-assessing the collections under new Access perspectives
• Business modelling must be developed - whenever applicable
• New models, new ideas and solutions are needed to address copyright issues to allow Access
• Archives need to think or re-think their functions, their educational activities, access policies, conservation strategies, acquisition priorities, etc.
• Any strategy for digitisation must be multi-purpose, to ameliorate the high costs
• Training is a key issue: staff in the Film Archives also need to learn new skills, and new words within new concepts:
  -- DAM, DRM
  -- Data: storage management, preservation
  -- Format migration / Media migration
  -- Digital delivery modes
  -- D-Cinema

In parallel to these challenges, Film Archives need to continue and reinforce their more traditional activities, which acquire a new meaning, a new function, and a new value in the new context. From this picture, it is clear that Digitisation will not change the basic priorities and problems of training and education, human resources, and financial resources. Digital isn't a shortcut to anything, or a final solution, rather it is a new start, where issues and costs do not replace the old ones, but are now added to the old ones.

Agenda for the Future

FIRST was requested to draft an Agenda for Research, to be presented to Film Collections, their supporting bodies, national governments, and EU Institutions.

However the FIRST Final Report considers that similar Agendas are needed to:
" Consider the changing scenario, looking at least into the medium term of 5-10 years
" Comprehensively consider the whole chain as an interactive process.
" Provide actions for governments and supporting bodies
" Provide actions for individual Institutions and their Associations
" Foster trans-national co-ordination
" Stimulate education and training
" Research strategic planning, and new policies particularly on Access
" Stimulate and direct technological research

THE TECHNOLOGICAL LANDSCAPE

The Risk of Vanishing Forever

There exists a paradox.

All existing film-based moving pictures are endangered and could be gone in 50 to 100 years, if badly stored. Yet at the same time, provided the optimum storage conditions are provided, i.e. very low temperatures. Even acetate film on the brink of vinegar syndrome decay (at the point of autocatalytic vinegar decay, an A-D level of 1.5) will last 50 years or more if frozen, longer than any digital format can be expected to last. To a large extent the difference between these two positions is principally cost. (The longevity of film in different storage conditions can be demonstrated by using Prescalc, a calculator that can be down-loaded from the Image Permanent Institute, Rochester, NY, website), and FIRST highlighted the Preserve Then Show documentation from the Danish Film Institute that described a recent approach to this management issue.

Although the substitution of nitrate film with safety film in the 1950's slowed the decay, later substrates have proved to have a shorter life than originally expected, although polyester has so far seemed far more stable. Most colour film dyes are far less stable than silver even in dark storage, and the very act of projecting film will inevitably fade the dyes. Finally all film is physically damaged to some extent by every handling step.

Clearly decay and fading are emotive issues. From a business perspective once a film has reached the last of its commercial exploitation outlets it stops generating profit and becomes a financial liability. Films are costly to store, especially for optimum life, and even more expensive to restore to their original state, yet it seems that digitization is at this time only adequate for extending access, and is not yet viable as a preservation process.

The FIRST partners are of the opinion that no collection should consider discarding any film original or master unless the condition of the image is beyond what could be eventually preserved or restored (and what can be eventually restored digitally increases daily). This philosophy will be seen to be fundamental to FIRST's recommendations.

User quality requirements and opposing views

FIRST has highlighted several important issues that have polarized, and in many cases have generated opposing, technical view-points. The most contentious are:

1. that digitizing film requires very high sampling rates, i.e. resolutions and bit depths, in some cases higher than are realistic or possible today, to ensure that all the image data available is recorded as digital data,
2. that these high specifications will be needed to provide restorations or access versions for an increasingly high quality cinema experience.
3. that no digital image file should be lossy compressed in order to create a digital preservation master,
4. no satisfactory long life storage medium exists today for these large digital files.

It is quite clear that research and development is needed to address each of these.

Critical issues of concern

The following issues are the key areas of concern.

1. Digitization requirements for film images.
The modern cinema production, cinema exhibition and TV production industry is in the process of setting the resolutions and sampling processes of scanners and telecine units available for digitising film. Sampling can be roughly separated into resolution (in terms of number of pixels/horizontal line or area) and bit depth.

There have been suggestions that 35mm film needs as much as 10K, 20bit, scanning to acquire all the information on film, but this is primarily opinion, and not based on research. It can also be pragmatically observed that many early films do not require the same high resolutions needed to capture the image contents of modern film stocks.
Considerable information exists from the television industry about bit depth for TV and screen displays, but only practical tests exist for the bit depths needed for extensive colour and contrast re-mapping (such as faded dye restorations). No recent intrinsic research seems to exist that examines these needs, and opinions vary.

Telecine units and scanners differ in their technology; the term scanner is used for those that record un-modified R, G & B components of an image. Telecine units usually involve more elaborate encoding to enhance the output, improve scanning rates or fle sizes….in effect they compress. The visual differences in the final results may be less than might be expected; some telecine units produce better image quality than some scanners at similar resolutions, because so many other factors are involved.

2. Compression.
Compression, the process of reducing the total file size of a digital data file or video signal while still retaining an image acceptable for a particular purpose, is of wide concern to film archivists. Lossless compression (LLC) is more generally accepted (although it is not always reversible). Lossy compression (LC) is irreversible. Compression in one form or another is fundamental to all digital video and computer display formats, widely used, often without recognition, and in many forms. All video formats are compressed in the sense that they are reduced from component R, G and B records, and any digitization of an analogue image that does not record every detail of resolution and colour gradation may be considered compression. A number of commercial collections have accepted, or are considering, HDTV formats (even interlaced HD) for image "preservation" of film where there may be an eventual requirement for both high quality (cinema) and low quality (TV and DVD).

Within the 18 month duration of the FIRST project digital cinema display quality has improved dramatically, with the resolution of commercially available projectors rising from 1,280 pixels per horizontal line to 2,000 (2K), and 4,000 (4K) being demonstrated. In the same time the projector dynamic (brightness) range has risen from 1,000:1, to over 2,000:1 being demonstrated (to equal film well over 4,000:1 is needed). So far many of these high resolution displays operate from HD signals, 1920p/hl, 24 or 25solid frames per second and 8-10bits, and are heavily compressed. There is a view, especially in the USA, held by the feature film major producers, that higher resolutions than 4K will be possible and necessary for the largest screens, and to make a distinction between Hollywood and others (although manufacturers are not commenting). These will be supplied in even more heavily compressed formats to retain a real time image display, at least at first. Modern 35mm film stocks carry more image information than the early low resolution digital projectors were able to display. If a digital restoration or an access copy is made from a film image it should achieve at least the optimum limiting quality of the system. At present the limiting quality is the digital cinema display system, but this may not be the case within a few years.

In general a "good" human eye is thought to need at least 4,000 separate pixels across the width of a cinema screen viewed with a 90° viewing angle (the front seats) to avoid obvious artefacts.

3. Obsolescence.
The short and dramatic lives of digital video, computer and display hardware, software and even file formats is a principle concern for long term preservation of digital film records. Manufacturers of magnetic video-tape expressly avoid giving life expectancies. The potential life of film, even 100 year old film, properly stored below freezing, probably far exceeds most, perhaps all, digital formats. Of all the hardware/format suppliers only Sony has given half-hearted guarantees that it will be possible to extract the files from just one of its tape formats, DTF1 & 2, for the next 30 years. There are no guarantees whatsoever over the longevity of any hardware or software. Routine and regular data migration is accepted as essential (every five years). This is both costly and imprecise as it is impossible to look ahead five years in this ever-changing technology. Standards beyond broadcast TV hardly exist, and the current plethora of current technologies for HDTV, DVD and image file formats suggests that standards will become increasingly difficult to agree and define.

4. Realistic time-scales.
Today, using quite low resolutions still creates practical problems. Digitizations, of 2 or 4K, and bit depths, like 12 or 16 bit, used for restoration, are not created, transferred or processed at real time (this is only possible at SD and HD TV). A semi-automatic software image repair (a small scratch for example) may require several minutes of operator's decision, and up to 25secs a frame for the repair render. In many instances every frame of a feature will require rendering, and this may be carried out several times. Adding together all these less than real time scans, imports and exports between workstations, decision times, and a final film record, the total time needed may be, and usually is, up to 500 times the running time of the film. Clearly process times and bit stream rates will increase, but at today's capacity very few restorations are possible (or affordable).

5. Digital storage media for digital PRESERVATION of film images.
No current long term digital film data storage solution exists that fits the requirements film archives, modern cinema film makers, rights holders, or even cinema exhibitors.

These requirements include:

• Content, hardware and software longevity at least approaching that of film.
• Capable of holding the full image (and sound) information of an original film image.
• Fast (preferably real time) load or download.

There seem to be the following alternative approaches available, separately or in combination:

• Encourage research and development of an entirely new storage support in order to fulfil existing user's requirement for a passive long life medium that does not require frequent un-quantified migration.
• Analyse and discuss the existing archive and collections requirements and modify them in order to find a compromise which can be fulfilled by technology in a short term. If there is sufficient market potential any manufacturer will be interested (and the modern cinema industry has similar unfulfilled requirements)
• Make a thorough investigation on the use of low levels of lossy compression.

6. "Business models" for film collections.
Two schools of thought exist amongst film archives and collection managers.
One is that digitization should only be carried out at the optimal resolution for film, and, in effect digital preservation should wait until that is affordable, available and agreed. This, largely a view of some public archives, means that film may wait for ever and never be accessible.

The alternative view is that digitization of film is essential for accessing and distributing images, and that it must be done directly into the most marketable format. This, largely a view of some commercial and broadcaster collections, means that broadcast standard videos are made and the film collection put to one side (and in some cases lost, broken up or even destroyed)

Inevitably both views attract criticism.

FIRST believes that a better business and preservation model might exist be between these two extremes. Low quality digital versions present the best access formats today, (leading to DVD, digital video and internet use, etc). Higher resolutions (being more costly and time consuming) are specifically appropriate for cinema restorations and may not be appropriate (or affordable, or practical from a time-scale viewpoint), for an entire collection, and should be concentrated on restoration for the cinema only. This bi-directional model needs to be balanced by storing/conserving the original film masters as optimally as possible to extend the life of the original images. This business models accepts that access formats are subject to change and fashion (as they have always been and that while DVD's may be today's browser access format of choice in a year or so this will inevitably change.

Some, notably commercial, collections that invested heavily in broadcast standard D1, D2 or Digibeta "preservations" have already discovered their short format, hardware and software lives, and now see they must carry out a "preservation" programme all over again.

Archives that quickly generated low cost browse versions (for example on DVD working from whatever ) are able to switch to tomorrow's browse format cheaply - still have the film to return to, and the cash to support programme. This apples to a number of small regional archives in Europe.

Digital to the Rescue?

The technical issue of the non-commercial 'afterlife' of a film comprise inter-linked issues which can be divided into the following categories: conservation, preservation, reconstruction and restoration, and access.

Conservation: Digital technology has not yet provided any solutions for the conservation and preservation of moving images.

• There is no universally agreed standard or method for preserving images in digital formats that matches the life expectancy, simplicity or the flexibility of film.
• The inevitable migration costs essential with current digital media are not yet quantifiable.
• In some circumstances digital technology could be (and has been in a number of well-documented cases) a threat to the original high quality film images, due to a misplaced confidence in the permanence of digital applications.

Preservation: Digital preservation sets out to preserve the "shape and substance" of images or sounds without preserving the format, or the support, or the original experience. Archivists are widely divided whether this acceptable…but would, in the end, agree that it is better to preserve what is possible, than lose a film image entirely.
Reconstruction and restoration. Analogue restoration of film is often extremely labour and time-consuming and therefore very costly, and only a few films get the full restoration treatment to the degree that they deserve or need. Digital technology is not expected to be cheaper, or by itself, save films and will definitely not increase the numbers fully restored. Technology companies claim that they have achieved efficient single-image compression schemes (spatial or inter-frame compressions) that are indistinguishable from uncompressed images, even to experts, and this is largely true. However these compressed images, in effect frozen in time, use, space and quality, cannot be worked on again to improve a restoration or extend quality to provide an new access version for an improved projection system. Probably only retaining the original film image as the master image allows this.

Access. There is no doubt that this is where digital technology has incontrovertible benefits, and in the future the range of access versions will include high resolution high quality digital cinema versions, to low resolution highly compressed browse versions for the internet.

SOME FIRST TECHNICAL GUIDELINES

FIRST's aim is to provide basic guidelines. They must be considered temporary, they are seen as being specific to this time and today's technology, and they will change as the technology changes and develops. The FIRST Partners also recognizes that, like the judgements of a Supreme Court, it could be expected that attitudes and philosophies will also change, and that therefore Guidelines will also change.

1. At this time no digital recording format for video or data can be considered to be as, or more secure than a film image element.

In consequence no collection should consider discarding any film original or master, unless the condition of the image is beyond what could be eventually preserved or restored.

2. There is no current consensus of opinion, or adequate research, on the correct resolution required far scanning film images.

• Therefore digitisation should be considered principally, at this time, as a means of providing access to the image in a suitable format for a selected display method. However in general it is recognized that 8 bit is adequate for a TV screen display (ITU 601). 10 bit is considered suitable for colour in higher brightness range digital projection systems (hence its use in HDTV formats)
• Where the distribution is for either an analogue video (of any format) or a digital disc or tape format, or where broadcast is planned, standard definition digital broadcast ITU 601 (recording on Digibeta) is recommended.
• Where distribution is planned as a film projected cinema display, clearly the original film format must be the first choice. This is not always a practical proposition and it is recognised that in many cases this would restrict access (for example few working 9.5mm projectors still exist), or would result in using poor quality restoration. High quality, high resolution digital film recorders only exist for 35mm film, so other original gauges and formats have to be transferred to an acceptable modern 35mm cinema format (including digital and cyan dye sound tracks) to be universally accessible. Even 16mm projection is no longer common.
• At present no higher resolutions (for 35mm film) than 2 or 4K are proposed by the modern film industry (despite pipe dreams by archivists hoping for higher resolutions) and it would be ineffective to seek standards beyond the technology available. The cinema industry widely uses both 2K (for digital Intermediate post production and effects) and 4K (effects only) pixels per line scanning for 35mm film and 1440 (so called 1.5K) for Super 16 and no higher resolution is correctly available. The highest film recording resolution of the Academy width 35mm frame (still universally used) is 4K pixels per line, although 2K is more common. The only digital cinema projectors of high enough quality now are between 1280 and 1920 pixels per line (with 4K proposed at an eventual standard) but so far the highest resolution digital provider format is 1920 HDTV signal (4K will be derived by pixel extrapolation, sometimes called "up-ressing").
• A 1920p/l HDTV digital format represents a useful temporary common source for both film and video restoration for access. There are many HDTV formats, and some may be used as storage formats for film images. The HD format 1080 24sfp (on uncompressed D6, or compressed D5) may be transferred to film frame by frame or downloaded to a standard 601 broadcast format for DVD, video or broadcast release. The format is less suitable for 35mm 4:3 Academy images as the resolution is reduced to 1440 pixels per line, adequate for 16mm film.
• Digital preservation master specifications. Should a collection manager feel that there is a need to produce a genuine digital preservation master because a film is decaying and a high quality restoration is not possible (for practical or costs reasons) FIRST recommends a specific route. All modern film stock manufacturers publish Modulation Transfer Function information and from these it is possible to estimate the resolution of a film and loosely equate that film resolution with a digital scanning resolution. This is not precise and, as can be demonstrated, an "over scan" factor helps to avoid both underestimating the film image quality resolution and avoiding aliasing artefacts. FIRST provides more details of this procedure in its final report.

3. The bit-depth of a film scan should depend upon the purpose for which a scan is being made (bearing in mind that no digital format is recommended as a preservation format at this time).

The higher the bit depth the more image alteration (in terms of colour grading and image repair) can be carried out without objectionable artefacts (such as contouring or other alias effects). Low bit depths increase the risk of eliminating some of the necessary film artefacts intrinsic to the film image (such as grain).

• Where a film image is scanned in order to retain a digital record for subsequent colour re-grading (to correct fading for example) a minimum of 14bit logarithmic is needed and 16bit is recommended for seriously faded colour print films.
• When a film image is already graded prior to recording the graded image (for example by a controller operating on a 14bit scanner such as a Spirit data telecine) an 8bit preservation file is considered the minimum. 10 bit would be better to allow for a considerable increase in the dynamic range of digital projectors in the future. This assumes that no further re-grading or restoration of this image will be required. If subsequent digital image repair is required 10 bit is the minimum requirement, but as the techniques improve we can expect this requirement to increase.
• A 601 broadcast scan will not allow major restoration regrading. Only limited image manipulation is possible at this 8 bit depth, and fade correctins are usually m not possible.

4. Digital noise reduction and "sharpening" processes should be used as little as possible in order to retain the film image character.

All grain reduction, sometimes called noise reduction methods, whether automatic or manually applied, will remove the grain character of the film image, producing a smooth unstructured un-film-like appearance. This also always reduce the sharpness of film originated images as the grain structure defines the edges of density. Recent research at Eastman Kodak has shown that digital enhancement of image sharpness, if handled carefully, may not be the ogre it has been portrayed as.

5. When a film to film digital restoration is being made the calibrated digital intermediate film processes offer the best results at present.

In this case the image may be graded with the restorer present to accept the image on a monitor at the scanning stage in the assurance that the film result eventually seen on the cinema screen will be similar, or the same. Scanners operating at 12 bit or less, with no image controller, reduce the range of corrections possible, and should be avoided. Some narrow gauges can only be scanned on scanners of this sort. In the future scanners may operate quickly and cheaply at 16 bit and this will allows scan to be carried out without any corrections which can then be done later. In general this is not economically viably now.

6. Prior to attempting any digital restoration, research to establish the technology originally used for both image and sound is essential, especially on extinct colour processes.

Photochemical restorations in the past have often relied on the limited range of modern film stocks on the market at the time, and it is admitted that many restorations do not simulate the original image accurately. Digital technology permits almost unlimited reconstruction and simulation and therefore there is less excuse for imprecision or guesswork. The literature on old processes has been poorly disseminated in the past but much of the information is available, but as restorers become more expert there will be a demand for this information.

7. The collection manager, restorer or archivist should direct all restorations carried out in service provider companies personally, recognising that few providers have done this before and very few have a specialist on their staff.

This applies to both analogue photochemical and digital restorations. Service providers, both film and digital laboratories vary in equipment and experience. In digital work the client should know and agree every stage and detail; resolutions, bit depths, equipment and software, digital formats used for intermediate storage, routes and techniques, and film stocks. The restorer will need to explain the exact nature of the job to be done and then agree the route and the final result preferably in writing. This will place an increasing burden on the restorer to understand the technology and to justify decisions that effect the image and sound quality.

8. No adequate solution exists for long term digital storage under the existing user's requirements.

Optical tape seems to represent a possible long-term solution if it is suitably developed. Further investigation should be made on this technology because their characteristics seem very well suited to the needs of film archive in term of longevity, capacity and speed. This technology (carefully analysed, developed, industrialised and if no drawbacks are found) if widely used by film collection holders (to achieve an low cost from an economy of scale), could be a possible future solution for long term digital storage of film images.

9. Compression technology is improving rapidly and is essential to display large resolution digital files in real time.

However no lossy compression or any sort can be recommended for any film digital preservation format, since no subsequent alteration to the image is possible without image quality reduction, sometimes very severe indeed. Compresssion (up to 50:1 can be demonstrated as being indistinquishable from the uncompressed image in certain conditions) can be invaluable for display access versions only. This may not always be the case in the future. It is possible that compression systems could be devized that reduces the size of preservation formats without losing later re-grading, reconstruction or image repair capability.

10. Archives and collections should accept lossy compression for storage of access data files.

11. Browse formats may be considered as temporary versions convenient to the technology of today and cheapness would be a virtue.

DVD or hard disc storage may be useful today, but we should recognize that within a year or so new, better, faster, bigger, smaller, brighter, sharper systems will be here.

FIRST'S PROPOSALS FOR FUTURE TECHNOLOGY RESEARCH

The complex and continuous evolution of technology and markets, and the vast scale of costs and effort involved in digitising the European Film Heritage, convinced the FIRST Partners that Film Archives and Collections need to look at solutions for the medium-term, over a period of a decade.

It is also expected that the results of research proposed by FIRST will extend beyond the limits of the film archives. Many of the issues are important to the whole field of high end digital technology applications in the media - in filmmaking, broadcast television, the cinema and home entertainment. The film archives are no longer alone - their problems are the same as other sectors.

Standardisation

1. Standardisation is crucial to all the areas covered by FIRST, and it is vital that these actions are taken by concerned bodies and Institutions to define and implement standards and best practices.

Digitisation / Restoration

2. Further research and knowledge is needed for digitisation of film images (e.g. in terms of bit-depth and resolution at least) to provide for the different requirements for both access and preservation.

3. Effects of compression on moving images, and the selection of more efficient compression techniques must be researched.

4. Realistic time-scales for digitisation and restoration for access must be assessed. It is clear that by using current technology, digitisation of a significant percentage of existing European film collections is virtually impossible.

Storage

5. The storage of film images digitised at full film resolution is currently either not possible, or is economically and practically unsustainable. Research in this complex field is fundamental.

6. Technical obsolescence is a central issue for preservation, and an important concern for its economic impact on large digital collections.

7. Migration, as a response to obsolescence, must be correctly evaluated in terms of costs and time-scales.

8. Research for a long term, high capacity, storage solution is essential. Storage of moving images at film resolution poses unprecedented challenges to the digital storage industry, and it could open up entirely new perspectives, and markets.

In addition to the technology section of FIRST, other workgroups reviewed cataloguing, exploitation, access and distribution of film images, and the following were their proposed research projects:

Description and metadata

9. Standardisation, to guarantee a lasting consistency in data management systems, is a core issue.

10. Interoperability of systems is vital to the exchange of information, and thereby to evolving a sustainable model based on synergies between collections.

11. Automatic and semiautomatic tools for metadata creation need to be created.

12. Multilingual searching is a core issue in establishing a "truly European" market - a precondition for cultural distribution and economic exploitation of collections.

13. Content indexing, in all its forms, needs further research.

14. Version management is crucial to all film collections.

15. The management of migration of existing embedded metadata into unwrapped metadata must be supported and further researched..

Distribution strategies

16. Efficient Digital Rights Management (DRM) systems, content protection and Digital Asset Management (DAM) tools need to be further researched.

17. Further business modelling and cost/benefit analysis that involves potential audience research is needed.

18. R&D is needed in Digital Access and Distribution channels, specifically D- and E-Cinema, Internet applications, Interactive Television, and higher quality Home Video content.

This summary of the FIRST project covers just the technical aspects, and these very briefly. It is recommended that the full report be read for full details. The complete report can be obtained by emailing a completed application form, available on the FIRST web site, or email filmarchive@ledoux.be.

REFERENCES
The following is a short list of recent technical key documents that relate to this report:

GENERAL SUMMARY OF ARCHIVE FILM SCANNING AND RESTORATION ISSUES
FIRST - Film Restoration and Conservation Strategies - State of the Art Report. Limited Issue as a hard copy publication but available as a CD from Cinémathèque Royale de Belgique, Brussels, or via www.film-first.org. This intitial FIRST Report is also provided on the Full Report CD.

FILM LIFE EXPECTANCY.
Prescalc, Image Permanence Institute's Preservation Calculator downloadable software available from their website www.rit.edu/ipi, http://www.rit.edu/~661www1/.

Nissen, Larsen, Christensen & Johnsen, Eds,"Preserve then Show", Danish Film Institute, 2002, ISBN 87-87195-55-0). Several papers including:
Bigourdan, J-L, "Film Storage Studies - Recent Findings",
Bigourdan, J-L, "A Strategic Preservation plan for the DFI's motion picture film collections",

Bigourdan, J-L, "From the Nitrate Experience to New Film Preservation Strategies" (includes recent acetate research), in Smither & Surowiec, Eds. "This Film is Dangerous", FIAF 2002. ISBN 2-9600296-0-7.

Bigourdan, J-L & Reilly J M, "Effectiveness of Storage Conditions in Controlling the Vinegar Syndrome: Preservation Strategies for Acetate Base Motion Picture Film Collections", in
"Image and Sound Archiving and Access, the challenges of the 3rd millenium" , Proc. Joint Technical Symposium Paris 2000 (published 2003). ISBNno2-01020-03-8:

EBU Tech 3289, "Preservation and Reuse of Film Material for Television" European Broadcasting Union, March / June 2001.

EBU Addendum to Tech 3289, "Preservation and Reuse of Film Material for Television" European Broadcasting Union, May 2004.

INFORMATION CONTENT OF FILM IMAGES
M. Rothaler, work carried out 1990, described in EBU Tech 3289, "Preservation and Reuse of Film Material for Television" European Broadcasting Union, March / June 2001.

Morton, R R A, Maurer, M A, & DuMont, Assessing the quality of motion picture systems from scene-to-digital data, SMPTE Journal, Vol 111, no 2, 2002.

Morton, R R A, Maurer, M A, & DuMont, C L, An Introduction to aliasing and sharpening in digital motion picture systems, SMPTE Motion Imaging Journal, Vol 112, nos 5 & 6, 2003.

Morton, R R A, DuMont, C L, & Maurer, M A, Relationships between pixel count, aliasing, and limiting resolution in digital motion picture systems, SMPTE Motion Imaging Journal, Vol 112, nos 7 & 8, 2003. SMPTE Motion Imaging Journal, Vol 112, nos 5 & 6, 2003.

DIGITAL PROJECTION
Svanberg, L. Ed., "EDCF Digital Projection Handbook", Focal Press, 2004 (release Aug 2004).

PHOTOCHEMICAL RESTORATION PRINCIPLES
Read, P & Meyer, M-P, "Restoration of Motion Picture Film", Butterworth, 2000

ALGORITHMS AND DIGITAL IMAGE REPAIR, GENERAL PRINCIPLES.
Kokaram, A. "Restoration of Moving Pictures ", Springer, 1999.

MODERN DIGITAL TERMINOLOGY AND DEFINITIONS
"The Quantel Guide to Digital Intermediate", Quantel, 2003.

"The Digital Fact Book", Quantel, 2003.

EUROPEAN BROADCASTERS SURVEY OF ARCHIVE COLLECTIONS, FILM AND VIDEO
Presto - Preservation Technology - Key Links Systems Specifications D3.2 Chapter 6 Film a pdf download from http://presto.joanneum.ac.at/projects.asp#D3.2.

 

___________________________

SPEAKER BIOS

Nicola Mazzanti

• Specialist of Film Restoration, Preservation, Archiving
• Currently, Independent professional Consultant and Project Manager of the EU Project FIRST – Film Restoration and Conservation Strategies
• Founder of film restoration laboratories L’Immagine Ritrovata and IR2 in Bologna, Italy.
• Founder of film festival Il Cinema Ritrovato, Bologna
• Professor of ‘Theory and Practice of Film Restoration’ (University of Udine, Italy – and University of Frankfurt, Germany)
• Member of the Advisory Committee, Venice Film Festival
• Member of the FIAF (International Federation of Film Archives) Technical Commission, as well as member of associations as AMIA and SMPTE.
• Author of several essays and publications on film history and film preservation and restoration.
  

Paul Read, M.Sc., FBKSTS.

• University College, London: B.Sc. chemistry, biology and mathematics, M.Sc.Chemical Engineering.
• Kodak Ltd (UK) 1961- 1973, applied research, teaching, lecturing, and commissioning motion picture film laboratories in Europe and Asia, and in broadcasters in Europe. (1965-7 with Eastman Kodak Co, USA).
• Kay Laboratories Ltd (later Metrocolor), London as Operations Director 1973-1979.
• Since 1979 independent technical consultant & project manager of new technologies in film, and film for TV, post-production, and restoration. Clients include filmmakers, manufacturers, collections, post-production facilities, legal and insurance, government departments, broadcasters and Universities in many countries. Legal guardian of the British Pathe collection, 1990-1. Council BKSTS 1976-80. Member of the Gamma Group, 1990-2000. FIAF Technical Commission, 1999 - . EBU P/TK as consultant, 2001- , FIRST (EU IST project) as consultant 2003 -. Specializing in digital intermediate feature film post-production management. Recent restoration clients include commercial companies and collections, and specialising in digital techniques.
• Author of numerous papers and including, with Mark-Paul Meyer, Restoration of Motion Picture Film, Butterworth Heinneman, Oxford, 2000.