Let’s Emulate the Sound of Colours!

Collaboration

I like to thank warmly all the persons
who have been most crucial for my own professional development.

Very little of the available technology and tools have been designed and manufactured especially for audio-visual archiving purposes. On the contrary, most of the technology and tools, that we are using on our daily archival work, has actually been designed and developed for today’s production and postproduction. We as archivists, conservators and restorers are continuously adapting this technology and tools to fulfil our specific preservation and access needs. Therefore we only rarely have for example a carbon arc projection of a toned and stencilled nitrate print, but usually we will enjoy a modern digital projection, that emulates as good as possible that specific historic look trough a file.

Emulation can, of course, also be seen in a pedagogic sense: a master teaching to his disciples. And this is important indeed! I entered in this field in late October 1986. During these 30 years I had the pleasure to meet many persons, which have been fundamental for my own career. I will mention here the most influential ones. The work of the teams I had the privilege to lead has always been a collaborative work. And this is true also for my own work! Collaboration on different levels: between institutions, between individuals, between companies; between institutions and/or individuals and/or companies.

Analogue and Digital

Back at the end of the 1980s and in the early 1990s, when I started considering digital methods for both film restoration and audio-visual conservation, I have mainly been considered as a fool – or at best a candid dreamer. Many years after, before the end of 2013, when I closed down our photochemical laboratory, I have often been considered as a person who missed to switch to the new digital world, because sometimes I still preferred to apply the “good old” analogue, photochemical methods. So goes the world… I fact I always tried to use the potentiality offered by both worlds and to mix them together, in order to achieve the best possible result, to have the most historically accurate presentation of a masterpiece or a document in the modern screening context. The previous last job my photochemical lab did was the preservation of amateur films shot on Kodachrome film stock. We were used to use for this the Fujifilm daylight 64 ASA camera negative as an intermediate film stock, which was no longer available, because the analogue colour reproduction is the best possible. We used the very last reels of our fridge. The Joint Technical Symposium’s audience is a special one, because, as it’s name says, this is a symposium of technical interested persons that crosses over specialisations.

The people from the sound area moved first and quicker from analogue to digital, also because the relatively small files made it possible earlier. Yet sound people were in general less carrier fetichists as film people. I have the impression that the move from analogue mechanical recording to analogue magnetic recording, to digital magnetic recording, to file-based recording was more natural and less polemic in the sound fields.

The broadcast people shifted to digital just after the sound people, in order to take advantage of the digital possibilities for production and postproduction, sometimes forgetting the new challenges for the preservation, and almost always forgetting that restoration and enhancement are two completely distinct activities. I am not saying that one is superior than the other, yet I am saying that they are fundamentally different.

The film people are still in the middle of paranoiac contradictions. It seems to me that they use the most of their energy, time and money by continuing to restore again and again a few dozen of films they call canon. Some of the film people thinks that also digital-born films should be preserved onto film stock. Yet in today’s worst/best scenario we can record onto film only ¹/₆₄ of the image quality there is actually in the file. That’s less than 2 % or, otherwise said, more than 98 % of the data are just good for… the rubbish bin. This situation comes from the fact that the sensors are constantly improving, but not the re-recorders onto film, as the marked for this specific type of equipment is gone.

If analogue storage is the answer to all audio-visual preservation needs, why do we have to struggle that much with analogue conservation and restoration issues?

Standardisation

To follow standards makes an effective collaboration possible. I wish to illustrate this with three examples regarding file formats, that may be relevant to our field.

ProRes if often hated by archivists, because it’s a proprietary format by Apple. Yet this it is a de facto standard in postproduction. SMPTE (Society of Motion Picture & Television Engineers) wishes to standardise it. I don’t know for sure if “only” Apple ProRes 422 HQ is meant to be standardised, or also other flavours like 4444 XQ. I hope this time they will do a good job, avoiding the the mistake they did with the standardisation of the CineForm or VC-5 codec. Essential information is still keep secret by GoPro, and not all metadata described in the standard do actually match the metadata generated by… GoPro’s own products. An unfortunate situation indeed.

A group of scholars from the University of Basel in Switzerland has the goal to define a TIFF flavour coming with all the technical metadata that are relevant for preservation. Adobe having refused the name TIFF/A, as an analogue to PDF/A, the group has chosen the name the new file format TI/A (Tagged Image for Archival). ISO (International Organisation for Standardisation) is the body where this new recommendation should be submitted, possibly amended, accepted and finally published.

The IETF (Internet Engineering Task Force) has established a workgroup called Cellar (Codec Encoding for LossLess Archiving and Realtime transmission). Cellar is currently standardising the lossless video codec FFV1, the lossless audio codec FLAC and the extensible media container Matroska, which is based on EBML (Extensible Binary Meta Language), a binary XML format. Once adopted, these standards could also be submitted through ISO and SMPTE for additional validations.

Open Source Software

Today there exist plenty of open source software which do not need to be an engineer to use them. These form a complete ecosystem for archival purposes. Here my favourites:

FFmpeg is a complete, cross-platform solution to record, convert and stream audio and video. In my opinion, every audio-visual archivist should be an FFmpeg literate.

QCTools offers an extremely wide range of strong quality control tools for video preservation, developed by the Bay Area Video Coalition and Dave Rice. I guess you all are familiar with MediaInfo. MediaArea is currently developing the new tool MediaConch. This consists of an implementation checker, policy checker, reporter and fixer that targets preservation-level audio-visual files (specifically Matroska, LPCM (Linear Pulse Code Modulation) and FFV1) for use in memory institutions.

The DCP has replaced the prints in the commercial film distribution for theatrical projection. DCP is a strange object, designed by the film industry to make film archivists’ life more difficult. However, the archive can also consider DCP in its own agenda. Never before an audio-visual archivist had a greater control over the way the audience has to see a historic film. The archivist can encode the right colours, the correct aspect ratio, an adequate projection speed… and the DCP will be screened that way in all cinema theatres worldwide. OpenDCP is one of the tools allowing this.

Open Source Hardware?

The minimal equipment needed so see the images of a silent 9.5 mm film in movement.
(This format is also called Pathé Baby.)

In the analogue world things may appear simple. Illustration 2 shows the minimal equipment needed to create the illusion of movement. In the digital world that simple equipment is not available so far.

When we built our first scanner, twelve years ago, we only modified one camera part of an optical step printer and kept all the other components – including the projector part, the lenses, the pre-wet device we developed and the mechanical movement – unchanged. The optical camera number 128 from Richard Craß in Berlin has been the first one we have converted from analogue to digital. Many did similar.

You can also build in your kitchen a relatively simple machine for cleaning magnetic tapes of all kind.

For archiving of big audio-visual data often LTO is chosen. The magnetic tape based solution LTO (Linear Tape-Open) with LTFS (Linear Tape File System) is indeed a good solution for data archiving in the real world. I do not think the archival world is needing an open source LTO player now.

Conservation

In short, the analogue conservation focusses on the chemistry of the base and the emulsion or the magnetic coating, and concludes that one should store cool and dry.

In very short, the digital conservation concludes that one should keep every single bit unchanged, and focusses on the container, codes and the so called raw data.

I would spend a few words on migration. From January to June 2014 we have migrated the digital archive of our company from LTO-4 to LTO-6. I decided to also change some file formats and to convert them en passant into other ones, more robust for the future. We started with roughly 1000 tapes and ended up with less than 300 tapes. Yes, we reduced the storage volume in the security cabinet by more than two thirds, and have now even more consistent data. We have migrated 5.7 PB of data and encountered one single error, which we could fix by using the second copy we had.

This kind of work must be done in an automated way, otherwise the so called human error became too important. Yet archivists need to be skilled in technical issues too, in order to be able to read critically the technical information provided by the industry and chose the solution that best fit the purposes.

Restoration

I have spoken many times about the ethics of conservation and restoration in my life (in 2010 I also organised a full conference on this very topic). As I am becoming older, I more and more agree with the definition: “Restoration consists in replicate the errors of the past.” We as a professional community should be able to do more in our restoration works than just replicate the errors of the past.

The first softwares for digital image restoration essentially selected the region where a problem occurs (called ROI for region of interest), and they defocus it a little, so the problem becomes less visible. In order to mask the fact that it’s less sharp, they increase a little the contrast, and the spectator has the illusion that the problem is solved.

More modern softwares do the work on the pixel level. They try to fix the actual problem that a pixel has, or that a small group of pixels have, and not longer to mask it. The newer programmes are more complex and need more powerful computers for efficiency. As restorers we can also use plenty of tools developed mainly by the game industry for the special effects. If we use them cum grano salis, then we can actually perform restoration work and not just do a camouflage of an existing problem. Of course, all these interventions have to be carefully documented, which is a difficult yet necessary task.

Recent Developments

Very important advancements have been made in the audio-visual and game industry.

The high dynamic range allows the archive to have a more accurate colour reproduction (for example for Dufaycolor, Kodachrome or Technicolor), and can be used to improve audio-visual archiving.

The high frame rate allows the archive to better encode the frame rates of the silent area and by multiplying frames and adding black frames, in order to emulate the projector shutter pales, gives a more accurate and also a more pleasant viewing experience. Theatrical access is improved dramatically. The forensic science developed new tools, allowing to read all the information out of a magnetic tape, and to interpret it by software.

The algorithms developed for DNA sequencing can be used to process the optical sound track from an archival scan, witch should be done frame by frame and edge to edge, with a little vertical over-scan. This opens the most doors for further restoration work and post-production for diffusion and access. This way both conservation and restoration are improved.

Summa summarum: Let’s emulate the sound of colour the best way the available technique allows us to do today, by strictly keeping open all the possibilities for our successors. They must be able to debate, modify and improve our own work.

2020-11-10