Audiovisual creation and preservation: part 2

Paul Heslin, Digital Collection Infrastructure Support Officer/Film Preservation Officer at the National Film and Sound Archive of Australia (NFSA) has generously contributed the following blog post. Introduction by Cambridge Policy and Planning Fellow, Somaya.

Introduction

As Digital Preservation is such a wide-ranging field, people working in this field can’t be an absolute expert on absolutely everything. It’s important to have areas of expertise and to connect and collaborate with others who can share their knowledge and experience.

While I have a background in audio, broadcast radio, multimedia and some video editing, moving image preservation is not my area of speciality. It is for this reason I invited Paul Heslin to compose a follow-up to my Audiovisual creation and preservation blog post. Paul Heslin is a Digital Archivist at the NFSA, currently preoccupied with migrating the digital collection to a new generation of LTO tapes.

I am incredibly indebted to Paul and the input from his colleagues and managers (some of whom are also my former colleagues, from when I worked at the NFSA).


Background to moving image preservation

A core concern for all archives is the ongoing accessibility of their collections. In this regard film archives have traditionally been spoilt: a film print does not require any intermediate machinery for assessment, and conceptually a projector is not a complicated device (at least in regards to presenting the visual qualities of the film). Film material can be expected to last hundreds of years if kept in appropriate vault conditions; other moving image formats are not so lucky. Many flavours of videotape are predicted to be extinct within a decade, due to loss of machinery or expertise, and born-digital moving image items can arrive at the archive in any possible format. This situation necessitates digitisation and migration to formats which can be trusted to continue to be suitable. But not only suitable!

Optimistically, the digital preservation of these formats carries the promise of these items maintaining their integrity perpetually. Unlike analogue preservation, there is no assumption of degradation over time, however there are other challenges to consider. The equipment requirements for playing back a digital audiovisual file can be complicated, especially as the vast majority of such files are compressed using encoding/decoding systems called codecs. There can be very interesting results when these systems go wrong!

Example of Bad Compression (in Paris). Copyright Paul Heslin

Example of Bad Compression (in Paris). Copyright Paul Heslin

Codecs

Codecs can be used in an archival context for much the same reason as the commercial world. Data storage is expensive and money saved can certainly be spent elsewhere. However, a key difference is that archives require truly lossless compression. So, it is important here to distinguish between lossless codecs which are mathematically lossless and those which are visually lossless. The later claims to encode in a way which is visually indistinguishable from an original source file, but it still dispenses with ‘superfluous’ data. This is not appropriate for archival usage, as this data loss cannot be recovered, and accumulated migration will ultimately result in visual and aural imperfections.

Another issue for archivists is that many codecs are proprietary or commercially owned: Apple’s ProRes format is a good example. While it is ubiquitously used within the production industry, it is an especially troubling example given signs that Apple will not be providing support into the future, especially for non-Mac platforms. This is not a huge issue for production companies who will have moved on to new projects and codecs, but for archives collecting these materials this presents a real problem. For this reason there is interest in dependable open standards which exist outside the commercial sphere.

FFV1

One of the more interesting developments in this area has been the emergence of the FFV1 codec. FFV1 started life in the early 2000s as a lossless codec associated with the FFMPEG free software project and has since gained some traction as a potential audiovisual preservation codec for the future. The advantages of the codec are:

  • It is non-proprietary, unlike the many other popular codecs currently in use.
  • It makes use of truly lossless compression, so archives can store more material in less space without compromising quality.
  • FFV1 files are ALWAYS losslessly compressed, which avoids accidents that can result from using formats which can either encode losslessly or lossily (like the popular JPEG-2000 archival format).
  • It internally holds checksums for each frame, allowing archivists to check that everything is as it should be. Frame checksums are especially useful in identifying where error has specifically occurred.
  • Benchmark tests indicate that conversion speeds are quicker than JPEG-2000. This makes a difference for archives dealing with large collections and limited computing resources.

The final, and possibly most exciting, attribute of FFV1 is that it is developing out of the needs of the archival community, rather than relying on specifications designed for industry use. Updates from the original developer, Michael Niedermayer, have introduced beneficial features for archival use and so far the codec has been implemented in different capacities by the The National Archives in the UK, the Austrian National Archives, and the Irish Film Institute, as well as being featured in the FIAF Journal Of Film Preservation.

Audiovisual creation and preservation

Following on from the well received Filling the digital preservation gap(s) post, Somaya has followed this up by reflecting on an in-house workshop she recently attended entitled, ‘Video Production: Shoot, Edit and Upload’, which has prompted these thoughts and some practical advice on analogue and digital audiovisual preservation.


My photographer colleague, Maciej, and I attended a video editing course at Cambridge University. I was there to learn about what video file formats staff at the University are creating and where these are being stored and made available, with a view to future preservation of this type of digital content. It is important we know what types of content the university is creating, so we know what we will have to preserve now and in the future.

While I have an audio background (having started out splicing reel-to-reel tapes), for the past 20 years I have predominantly worked in the digital domain. I am not an analogue audiovisual specialist, particularly not film and video. However, I have previously worked for an Australian national broadcaster (in the radio division) and the National Film and Sound Archive of Australia (developing a strategy for acquiring and preserving multi-platform content, such as Apps and interactive audiovisual works etc.)

AV Media

A range of analogue and digital carriers. Image credit: Somaya Langley

Since my arrival, both Cambridge University Library and Bodleian Libraries, Oxford have been very keen to discuss their audiovisual collections and I’m led to believe there may be some significant film collections held in Cambridge University Library (although, I’ve yet to see them in person). As many people have been asking about audiovisual, I thought I would briefly share some information (from an Australiasian perspective).

A ten-year deadline for audiovisual digitisation

In 2015, the National Film and Sound Archive of Australia launched a strategy paper called Deadline 2025: collections at risk which outlines why there is a ten-year deadline to digitise analogue (or digital tape-based) audiovisual material. This is due to the fragility of the carriers (the reels, tapes etc.), playback equipment having been discontinued – a considerable proportion of equipment purchased is secondhand and bought via eBay or similar services – as well as the specialist skills also disappearing. The knowledge of analogue audiovisual held by engineers of this era is considerable. These engineers have started to retire, and while there is some succession planning, there is not nearly enough to retain the in-depth, wide-ranging and highly technical skill-sets and knowledge of engineers trained last century.

Obsolete physical carriers

Why is it that audio and video content requires extra attention? There is a considerable amount of specialist knowledge that is required to understand how carriers are best handled. In the same way that conservation staff know how to repair delicate hundreds of years old paper or paintings, similar knowledge is required to handle audiovisual carriers such as magnetic tape (cassettes, reel-to-reel tapes) or optical media (CDs, DVDs etc.) Not having the proper knowledge of how to wind tapes, when a tape requires ‘baking’ or holding a CD in a certain way can result in damage to the carrier. Further information on handling carriers can be found here: http://www.iasa-web.org/tc05/handling-storage-audio-video-carriers. If you’re struggling to identify an audiovisual or digital carrier, then Mediapedia (a resource initiated by Douglas Elford at the National Library of Australia) is a great starting point.

Earlier this year, along with former State Library of New South Wales colleagues in Sydney, Scott Wajon and Damien Cassidy, we produced an Obsolete Physical Carriers Report based on a survey of audiovisual and digital carriers held in nine Australian libraries for the National and State Libraries Australasia (NSLA). This outlined the scope of the problem of ‘at-risk’ content held on analogue and digital carriers (and that this content needs to be transferred within the next decade). Of note is the short lifespan of ‘burnt’ (as opposed to professionally mastered) CDs and DVDs.

Audio preservation standards

In 2004, the International Association of Sound and Audiovisual Archives (IASA) first published the audio preservation standard: Guidelines on the Production and Preservation of Digital Audio Objects. I have been lucky to have worked with the editor (Kevin Bradley from the National Library of Australia) and several of the main contributors (including Matthew Davies) in some of my previous roles. This sets a standard for the quality.

Other standards publications IASA has produced can be found here: http://www.iasa-web.org/iasa-publications

Video preservation standards

Since approximately 2010, IASA has been working towards publishing a similar standard for video preservation. While this has yet to be released, it is likely to be soon (hopefully 2017?).

In lieu of a world-wide standard for video

As audiovisual institutions around the world are digitising their film and video collections, they are developing their own internal guidelines and procedures regarding ‘preservation quality’ video, however best-practice has started to form with many choosing to use:

  • Lossless Motion JPEG 2000, inside an MXF OP1a wrapper

There is also interest in another CODEC as a possible video preservation standard, which is being discussed by various audiovisual preservation specialists as a possible alternative:

  • Lossless FFV1 (FF Video Codec 1)

For content that has been captured at a lower quality in the first place (e.g. video created with consumer rather than professional equipment), another format various collecting institutions may consider is:

  • Uncompressed AVI

Why is video tricky?

For the most part, video is more complex than audio for several reasons including:

  • A video file format may not be what it seems – there is both a container (aka wrapper) holding inside it the video file (e.g. Quicktime MOV file containing content encoded as H.264).
  • Video codecs can also produce files that are lossy (compressed with a loss of information) or lossless (compressed, but where data is not lost as part of the encoding process).

The tool, MediaInfo, can provide information about both the container and the encoded file for a wide range of file formats.

Of course, there are many things to consider and parameters to configure – hence needing film and video digitisation specialists and specialist equipment to produce preservation quality digitised video.

From the US, the Federal Agencies Digitization Guide Initiative (FADGI) are also a great resource for information about audiovisual digitisation.

Consumer-produced audiovisual content

While I would recommend that consumers capture and produce as high-quality audiovisual content as their equipment allows (minimum of 24bit, 48kHz WAV files for audio and uncompressed AVI for video), I’m aware those using mobile devices aren’t necessarily going to do this. So, in addition to ensuring, where possible, preservation quality audiovisual content is created now and in the future, we will also have to take into account significant content being created on non-professional consumer-grade equipment and the potential proprietary file formats produced.

What can you do?

If you’re creating audio and or video content:

  • set your settings on your device to the highest quality it will allow (however you will need to take into account the amount of storage this will require)
  • try to avoid proprietary and less common file formats and CODECs
  • be aware that, especially for video content, your file is a little more complex than you might have expected: it’s a ‘file’ inside a ‘wrapper’, so it’s almost like two files, one inside the other…

How big?

Another consideration are the file sizes of digitised and born-digital film and video content which has implications for how to ‘wrangle’ files as well as considerable storage needed … however this is best left for a future blog post.

We will discuss more about born-digital audiovisual content and considerations as the DPOC project progresses.