Interview with program director Hans Westerhof
Interview with Hans Westerhof
Manager sector Collections at Sound and Vision and program director Images for the Future
Sound and Vision is not just an archive where audiovisual material is stored. It’s also a production archive that is being used by hundreds of professionals every day. As a result, we have a high-quality infrastructure and all kinds of achievement agreements to live up to. A large part of our archives is made up of broadcasting archives that are mostly analogue. Meanwhile, the modern process is digital: digital material flows in on daily basis and has to be made available within a few hours. We do this for the public broadcasting services and that’s why we have to meet many strict requirements.
Last night’s news, for example, has already been described by an employee of Sound and Vision this morning. To do so, he logs in on iMMix (a catalogue and search interface for radio and television material) and can request the clip in question directly from the archives. So, a few hours after broadcasting, the news is already described in accordance with the fixed rules. The metadata is immediately available for reuse. Basically we deliver high resolution reproductions within 20 to 30 minutes. In that television archive 1,6 petabyte (over one and a half million gigabyte) of data has been stored up to now. And the archive will grow one petabyte per year thanks to Images for the Future.
The infrastructure for storage has been put out to Technicolor. Technicolor has tape libraries that wouldn’t fit into this room. How it works: just like every audiovisual archive, we build a catalogue. Sound and Vision’s catalogue is a beautiful flexible system called iMMix, with many fields for the description of the material and also for applying hierarchy to those fields. It has been custom made for audiovisual filing and for the work process of Sound and Vision.
We think that smaller archives can also profit from iMMix to regulate their digital filing and delivering. It is almost impossible for them to build themselves, because it’s so complex. Good ‘off the shelf’ solutions don’t really exist. At Sound and Vision a group of software developers have been building and managing the information systems for a couple of years. It would be nice if other archives could digitize their material in our slipstream. We would like to make our acquired knowledge and expertise available to them for this purpose. Archives can use our storage infrastructure. They will get their own iMMix-client with which they can manage and add metadata to their digitized material. Their own collections are maintained, but they manage them with our tools. The digitizing, the storage and the use of iMMix does cost money. Sound and Vision has built an enormous infrastructure over the past years that is worth millions. But before the project Images for the Future was begun, we were dealing with the same problems as the smaller archives: limited financial means.
From 150 databases to 1
Images for the Future enables iMMix to be expanded in such a manner that the transition from analogue tot digital can be made. Because, between tens of thousands of film containers and boxes something can get lost… usually that tape has been put back in the wrong place. That’s why the exact location of the depot in which the container can be found is put on iMMix. A lot of old material is still found in old databases and not yet in iMMix. Those 150 databases have to be converted to iMMix, a difficult task that is carried out by conversion specialists. Moreover, conflicting data form an inconvenient problem. For example: are we dealing with the same material if 90% of the characters of a title correspond to the title of another database? There are also databases made up of Word documents. They are difficult to convert, to say nothing of handwritten data or microfiches.
I think we’ll be working on the conversions to iMMix for another two years. But every time we’ve converted a database, we celebrate! iMMix really is the backbone of our system. We can’t go back to the Excel-lists of yore. Then things would go terribly wrong due to the sheer size of Images for the Future. Currently we are still working with Excel out of necessity, because not all the data have been put on iMMix yet. We intentionally let the people who have to work with the databases contribute to the new structure of iMMix by thinking along and supervising. They say: “If you registrate it this way now, I can use it again later”. There are international standards for metadata, like Dublin Core, but it is made up of only fifteen metadata fields. The metamodel that we use is based upon a more finely darned and hierarchic FRBR-model and has a multitude of fields.
Media documentalists are often interested in the programs they document. We don’t describe every single thing profoundly, but we do try to describe notable events. Due to the employees’ involvement in the subjects we don’t overlook much. At least fifty documentalists are active at Sound and Vision. In most cases the material gets an evaluation, on which we ground priority during conservation, digitization and metadata conversion.
The call for tenders: bringing the headache forward
Sound and Vision also works on film-on-film conservation. A lot of film can be found on an acetate carrier and acetate turns acid in course of time which renders the film useless. That’s why some depots smell like acid. And that is why we conserve important material, the heart of our collection, film on film. There are collections that have been so badly damaged, that they would become useless within the next ten years. New film carriers are polyester-based, so they can basically be preserved for hundreds of years. Because the conservation costs over 211 thousand Euros, this assignment calls for a European call for tender. The procurement takes a lot of administrative work and quite a bit of walking on eggshells. Everything has to be carefully thought out in advance. If you haven’t meticulously written things down in your guidelines, you’ll always get stuck with additional work. You can’t say to an offerer: “You’ll receive this many film containers, most of them carry 1 roll of film, but some of them carry 7…” You have to describe exactly what the offerer can expect. But how to go about that with ten thousand containers? Do you have to open each one of them? In the end you take random samples and estimate a percentage. Quality criteria have to be very well determined, because you can’t abandon them later. A selected tender is based upon price and quality criteria that have to be made known to all parties beforehand. We have decided to bring “the headache forward” by forcing ourselves to write out everything in the tender guidelines, so that price selection is the only thing left to do.
How much wow and flutter do you allow?
We have written down exactly what we expect. Take perfo-acetate, the audiotape for films, for example. If you transfer it to a new carrier, you’ll always have to deal with a bit of quality loss. But how much loss is acceptable? That’s why we’ve tested samples with digital meters to see how much so called “wow and flutter” appears. We have determined that a maximum of 6 decibel of ‘contamination’ is allowed to appear between the master and the copy. That a deviation of for instance 2 pixels per frame on so many frames per program is acceptable for videos. These are the kind of details that have to be recorded in the tender guidelines.
There always seems to be additional work
Even if you write everything down well, you’ll always bump into problems. Our film rolls, for example, have often been glued into one big roll in different ways. This was never a problem for television broadcasts because you couldn’t see the quality difference it brings about. But it is a problem for the parties that have to conserve these film rolls. This problem has to be solved and causes additional work that has to be paid. This is a typical example of the kind of stuff that you can’t really determine in advance. Everything that isn’t formulated in the tender guidelines will be charged as additional work. Consequently the tender guidelines are composed very carefully.
Offering parties spend a lot of time writing the offer, it can even take months. They don’t get any compensation for it, so there’s a lot at stake. You’ll receive offers on razor’s edge. Everything that still has to be changed after obtaining tender is considered to be additional work. This is very understandable, because, the tender system forces them to submit offers at very sharp rates. Due to this, the process becomes expensive and exceptionally complicated. But it is also useful, because it forces you to think everything through and to know your material intricately well. Especially when it comes to orders close to the 211 thousand euro limit, one can wonder if it is useful to go through with the whole circus. Not only is it expensive, but it also causes delays that can last for months. But rules are rules. Fortunately most tenders of Sound and Vision are much higher, and the benefits are much greater than the costs.
At the moment we’re working on the fourth tender for the Education Media Platform. The first tender was for 55.000 hours of digibeta video, the second tender was for film-on-film conservation (acetate) and the third for film audio conservation (perfo-acetate).
Sound is conserved at a speed of about 100 hours a month, video at about 2000 hours per month (it already has a digital signal). For the most part the process is automated and the quality control is executed by software. The quality criteria for that have also been accurately formulated. We spent six months in consultation with Technicolor on precisely regulating the work process after tender.
These are the things you bump into with the digibeta tapes (the digital version of the professional standard Betacam): Every box has to be opened to see if there is a barcode on the tape that the machines can read. Apart from that we need time codes. Different fragments are turned into high resolution files that can be described after they have been inspected by documentalists. So, they can be durably stored in the digital archive (the tape library) after having been supplied with complete metadata.
If it doesn’t have a number, it’s lost
Audio digitizing is the simplest form of digitizing, but its metadata are an entirely different story. Problem cases have to be removed from the principle machinery immediately and have to be intercepted by a different system. Because we digitize on such a large scale, we want to automate as much as possible. At the moment the fault-sensitivity can be linked mostly to human error. Inaccurate overtyping of the number on a box or of the filename, for example. This can be intercepted by inputting the number twice, taking into account that no difference between the two may occur. This is very important, because the following applies to a large archive: “if you don’t have the right number, it is lost”.
The input of file names and numbers, quality control and the making of browser files are still too dependent on human labor. Specific software does exist that can automate these processes in the near future.
Copyright and orphaned works
“Copyright is often being treated with unnecessary difficulty.” Most of all, it’s a lot of work because the precise rights weren’t registered very well in the past. Once again this emphasizes the importance of exact formulation and registration of information about the material.
Sound and Vision has very few orphaned works. Most of the hundreds of thousands of hours of material have rightful claimants. These are often broadcasting companies, but outside producers may also have rights. Neither we nor the broadcasting companies have done a good job at digitally registrating these outside producers. But both of us can profit from finding out who they are. But these works are not orphaned. We’re working on a large scale rights administration.
It also depends on the fragment. With well-known Dutch people you have to find out if they own rights themselves. Suppose you’re dealing with old radio programs, then the rights are often owned by the broadcasting company. Do you have to examine every broadcast, or can you assume that probably all broadcasts from that season fall under the same regime? No rules are accurate all of the time, but a little bit of common sense does help. Nevertheless it is a lot of work to unravel it all.
Orphaned works have never made anyone rich
You have to choose a pragmatic approach to orphan works. And you can come a long way on intuition. Often you’ll know if it’s a fragment from a Paul Verhoeven film and then you’d better not use it. But if it’s material from the thirties and you really can’t discover who the rightful claimant is, and you’ve indicated that you’ve undertaken all steps to find out… what can happen if you use it anyway? In some cases the rightful claimant will turn up. You’ll give him reasonable compensation and will do as he says from then on. For great amounts, you can set up a fund that can anticipate these exceptions and can consequently help you estimate the risks. Wouldn’t it be a shame to leave certain material behind closed doors to all eternity because of this? Furthermore, there are often parties who want to show material in good faith, without wanting to make a buck. Orphaned works have never made anyone rich.
Sound and Vision has contracts with important rights organizations. If we want something new, we can always continue in good consultation with them.
Regulating copyright is kind of like a steam tanker: not a complex, but a slow process. It takes a long time to get the process started, but once it’s sailing all’s well. In a manner of speaking, you could go to a congress about orphaned works every week. You can keep on talking about it, but why not just do it? At Sound and Vision about ten people are working on it and it’s a lot of work: faxes, telephone calls, files filled with paperwork. Copyright isn’t sexy, but it has to be done. And if you do it, you can go a long way.
Interview: Lieke Heijmans & text: Marcel Oosterwijk