Human Interventions and Technological Rescues: Digital Restoration of the Moving Image
This piece was written for MIAS 220: Archeology of the Media, taught by Professor Michael Friend for the Moving Image Archive Studies Program at UCLA during the Winter of 2013.
“Before you become too entranced with gorgeous gadgets and mesmerizing video displays, let me remind you that information is not knowledge, knowledge is not wisdom, and wisdom is not foresight. Each grows out of the other, and we need them all.” — Arthur C. Clarke
As we shift into the digital film universe, it seems that film restoration and its associated processes are shifting just as quickly. While photochemical restoration itself is decreasing, it has not entirely disappeared. It has become a hybrid technique, working with a variety of digital methods for the final outcome. But this will not last. The world of the photochemical is slowly getting phased out and this hybridization will eventually be massaged into an entirely digital course of treatment. Even if last year’s battle cry was “nitrate won’t wait,” it has become a solemn “35 is no longer alive.” However, one thing has become awkwardly clear as we exist in this purgatory-like state, waiting to let the chips fall as they may: without the proper background of photochemical training and the meticulous and critical intervention of skilled human professionals, digital restoration itself will not live up to its full potential and will be entirely unsuccessful. I would argue that the very structure and make-up of the digital restoration process has been generated by the transition from, dedication to, and integration of its photochemical roots.
In the following pages, I will explore the ways in which human beings with solid backgrounds in photochemical restoration have made a difference in this high-tech world. I will also show how these same individuals and companies have managed to create a new role within the archiving and preservation/restoration worlds: they serve as liaisons between complex machinations and chemical mystery. Whether they are designing mathematical algorithms to resolve fractured frames at MTI, running scan tests to examine remnant color information on a print that has suffered considerable dye loss like at Disney or simply using customized software to assist in developing missing pieces of a sequence like at Technicolor, these people have all managed to become living, breathing conduits of information between the worlds of science and technology. They are the bridge that maintains an even keel between film handling and computing.
While the worlds of digital and photochemical may seem light years apart, they are not all that different. In fact, they are quite neighborly. Their connective tissue is us: the users, the humans, and the operators. As the people who work within the internal “guts” of each of these worlds, we are the ones who best know how to make them fly. Without us, they are inoperable. Digital or analogue, film elements need human beings to preserve them, restore them, construct them, make them go. And so we do.
The very practice of restoring a film bears a great similarity to the physical make-up of its photochemical materials, the primary example being that both things possess a multi-layered structure where each coating is built upon the one underneath. As a physical object, film consists of a series of different layers, each assisting in the capturing of the image, retaining it within its gelatinous content, and having enough reactive elements present to allow for that image to appear after a combination of light exposure and chemical development. In the world of film restoration, the series of activities that take place after the initial research phase create the “emulsion” of this process. From selecting best usable elements to choosing the best scanner for the job and determining the color timing, each of these decisions is placed upon the other, thickening the restoration; stemming from the strong research base that was lain out at the beginning. Keeping this analogy in mind, this can be a fragile procedure if any of the layers are “defective” or breakdown. Should any of these aspects be faulty, the final product will not be satisfactory. If the digital restoration process is not accomplished in a meticulous manner, paying careful attention to both the original authorial intent of the filmmaker and the concept of moving image aesthetics on the whole, what will be achieved will be, in a sense, a defective work just as much as if the emulsion on the film were to be scratched or damaged in some way. In order to make sure that the product turns out best, digital film restoration’s “emulsive” layer must be kept intact. Thus it is a balance between human and technological means. But, as film preservationist Ross Lipman writes, “restoration is not a precise science and the standards for which we strive are at a fundamental level not truly standardized—nor even standardizable. When I say this, I by no means decry such things as photographic aim densities, process control, or theatrical audio levels—measurable benchmarks that enable our work…The final text on any work is the work itself.” (Lipman 2009)
The Cast and Crew
In this investigation into what it means to digitally restore a film, I met and/or spoke with several professionals in the field. I was in contact with Grover Crisp, the Senior Vice President of Asset Management, Film Restoration & Digital Mastering at Sony Pictures, I spoke with Andrea Kalas, the Vice President of Archives at Paramount Pictures over the phone, and I met personally with Jayson Wall (Manager, Library Mastering at The Walt Disney Company), Tom Burton (Director of Restoration Services at Technicolor Creative Services, Hollywood), and Larry Chernoff (CEO at MTI Films). Each of these people has a long and extensive background in the visual media field, and has left an indelible mark on the digital restorations that they have worked on. Although their current focus is within the digital realm, it hasn’t always been. Every conversation and all of my research revealed that each person I spoke with who currently produces digitally restored content began in the world of the photochemical and has been there for most of their career. When the landscape began to change, they moved forward with it, anticipating new technologies and setting themselves ahead of the curve. However, it was these experiences and the learned procedures and practices that assisted them in being able to stay the course and build a healthy, dynamic and demanding atmosphere for the current work being done. Due to knowing what the tools had been, they knew what they needed to be and what they could be. Additionally, being highly educated on the many idiosyncratic ins-and-outs of photochemical materials and their design gave them a specialized knowledge going into the digital realm that those without much film-handling experience may not have had.
Authorial Intent and the Digital Aesthetic
The film archiving and preservation community contains a wide variety of professionals with divergent opinions on a multiplicity of topics. When it comes to film restoration, however, there is an almost universal agreement on two issues: first, there is no steadfast “rule” that applies to every restoration: each project requires different tools, “fixes” and methods. Like snowflakes or human fingerprints, no two restorations are the same. The second point of confluence is that the end goal of moving image restoration is to produce a piece that is as close to what would have been seen originally as possible. But that is where the problem lies: what is “original”?
Academic arguments on the subject abound, with a multiplicity of learned scholars focusing their energies on theories of restoration ethics and concepts of “originality,” perspectives ranging from the more traditional art restoration/conservative stance of Paolo Cherchi Usai to Julia Wallmüller’s more modern and flexible approaches. Whatever the case may be, the debates rage on. These arguments, of course, have only become more complex with the introduction of digital. Because of the tools and their ever-changing abilities to change the materials, our stance on originality must become more defined. As Paul Read and Mark-Paul Meyer write,
To know what to aim for, the concept of the original has to become very concrete, because with the new technologies for the first time a film can really be repaired in its smallest elements, image and sound. If we consider ‘original’ in its strict sense, image and sound quality could become ‘as new.’ But then one should ask if it’s desirable to strive for a perfect rejuvenation of image and sound…for researchers and future restorers this can create a great problem. It is absolutely imperative that a digital restoration can be analyzed and done again from the start. (Read 2000)
Julia Wallmüller writes, “Paolo Cherchi Usai…insists that the original state is a hypothetical condition, since the restorer cannot know exactly what an original looked like.” (Wallmuller 2007) In a sense, Usai’s statement is correct. Unless we are dealing with recent visual materials and their restoration, we do not (and can not) know what the original moving image looked like, especially upon projection. We can have an idea and a sense, but we were not in that theater, nor were we in the lab where it was produced. We have received these original materials at a very different space in their timeline and our own. This fact does not mean that we are not responsible for their well-being, however. If we have access to adequate resources to assist us in making our best effort at reconstructing the piece into what we feel it would have been like, we have a certain obligation as part of the preservation community to do our best to return it to what we feel it would have been or as close to it. As Andrew Oren of Fotokem remarked during the question and answer period of the March, 2013 meeting of the Society of Motion Picture and Television Engineers (SMPTE), “It’s the well-kept secret of our business. [The moving image restoration community] does all the research we can do, but then we just take our best guess.” (Oran 2013) While the wording of this may seem casual, Oren’s intent was far from it. Due to the fact that we will never be able to know the exact nature, we do highly educated estimations. And now that digital has come into the picture, it is entirely possible that the look of the materials and what is being shown is unlike anything that has ever been seen due to the mathematic and technical precision of the tools. Whether or not this results in what others might derogatively refer to as the creation of “new” materials due to digital technology, by taking on this career there is a commitment to seeing these cultural materials move forward into the future as best we can.
The question then lies before us: if the image we are seeing is not the same as what was originally screened, is this what the film’s author intended? If the digital restoration process is giving us a new collection of moving image products under the same names, are they, in fact, restorations since they are shifting from one state (analog) to the other (digital)? While many would grumble and argue and say, “Of course! The resolution is higher! The image is clearer! It is clean and beautiful now! It looks better!” I do not think that the argument can be or should be solved by being placed on a better/worse than continuum. Qualitative terms delve into subjectivity even amidst the most highly skilled technicians. What we should be doing is gauging this discourse using an entirely new lexicon. Digitally restored moving images will never look like photochemical work because they are not photochemical work. I believe that once we have established this we can move forward and be able to appreciate each area for its own worth and on its own terms.
Tools and Toys
Through the use of new scanning equipment and a variety of automated software programs our moving images are slowly being restored to a state considered to be as close to the “original” as possible. We have tools that help to reduce film grain and conceal scratches on the image by looking at frames before and after and estimating pixel placement without adding any new data in addition to software with modules for grain modification, dust/dirt cleanup and scratch assistance which is also attuned to flicker correction, chemical stain removal and other “blemish” work.
It is significant to note that while automatic mathematical algorithms designed specifically for the moving image are doing much of this digital restoration work, none of it should be done without the human element. A lack of human intervention within any of these tools can create digital artifacts, remove too much grain, or catalyze a variety of other “digital dangers.”
In speaking with Tom Burton at Technicolor, we looked at the different items that are used for digital restorations and what and why each would be sought out. We looked at a variety of different scanning devices and discussed what would be appropriate for each and why. While I found the Scanity to be incredibly fascinating due to its ability to scan shrunken film and its speed, the first real scanner that we looked at was the Spirit DataCine and it was one of the older variety. The Spirit works on continuously moving film and is what is known as a charge-coupled device (CCD). CCDs have a special kind of chip, originally developed in the 1960s, with an array of points that captures photons and converts them to electrons. CCDs render images into pixels but they are also what are known to be linear array scanners, meaning that their essential vision system will pick up and point out the surface issues on the film. Burton notes that the Spirit Datacine is “used primarily for HD resolution material. This is more of an old-school scanner…you would work on them in real time and then go directly to videotape if you were color correcting.” (Burton 2013)
The Spirit
The next scanners we discussed were the Northlight and Northlight 2. According to Jayson Wall, Disney used the Northlight when they worked on 20,000 Leagues Under the Sea because, as he said, “It’s a great scanner. The clarity. The sharpness which you get out of the scans compared to other scans, to a Spirit, or to a Scanity.” Although he said that other scanners today might give different (or better) results, he said that the Northlight definitely provided Disney with what they were looking for with 20,000. (Wall 2013) Andrea Kalas at Paramount, upon being asked about her preferred scanners at Paramount also listed the Northlights (Kalas 2013). While her reasoning ran along the same lines as Jayson Wall’s, after discussing the technological capabilities of the machines, there was certainly more than simply the clarity of the product that lent value to the Northlights as a restoration tool. As these machines have been updated, they have progressed in information extraction and accelerated, both highly valuable assets in this industry.
Northlight Film Scanner
What Burton had to say about the Northlight and its updated sister scanner, the Northlight 2, was quite fascinating. While both scanners have many of the same qualities, Northlight 2 is simply faster. About six times faster, according to the literature. While they are both pin-registered and able to support 16mm, Super16mm, 35mm or 65mm film, the Northlight 2 has the distinct advantage of a heightened design for the digital scanning work that is currently being done. As they say, time is money and now moreso than ever. Not only does the Northlight 2 have a scanning speed registering a 4K scan of 35mm film at 0.8 seconds/frame compared to the Northlight 1 which came in at 4.7 seconds/frame, but the Northlight 2’s initial scan resolution is markedly higher, listed as 8K according to its vendors. (Filmlight, Ltd 2013) While that was impressive, that was not what really was mind-blowing. What Burton explained about this CCD machine that set it apart from the Spirit was its infrared scanning capabilities.
The Northlight will take the film and instead of simply scanning the image for the normal RGB (red, green, blue) colorspace, it will do an additional pass with an infrared scan. This separate infrared channel will pick up only the items that, as Tom notes, might be “a physical defect either on the surface of the film or dug into the film. So anything that violates the horizontal plane of the film, that infrared channel sees that defect and it creates a map of that.” (Burton 2013) According to Tom, while there are various kinds of infrared channels, the one on Technicolor’s Northlight 2 has 1-bit-depth of information: black or white. Whatever the infrared pass “sees” as surface flaws becomes translated into clear or white dots and/or lines on a black surface, mapping out the issues. This information (or metadata, if you will) will then get brought into an image restoration program that will look at the file and automatically repair what has been marked.
Through what is commonly referred to as temporal analysis, these programs use visual data contained in the frames before and after the damaged pieces to “read” and “repair.” If these other frames do not have the “flaws” that the IR channel had picked up, then the software automatically assumes that the marked/mapped sections are, in fact, errors and will automatically grab similar data from previous or subsequent frames and place it in the defective area. But, as mentioned earlier, this is where it gets tricky. What if the infrared has marked something that is not a defect?
Tom and I spoke about this issue at length. Beyond mentioning the programs that can read this IR channel (he mentioned both MTI’s CORRECT and Pixel Farm’s PF Clean as being the ones used at Technicolor), and the fact that the IR channel option is only an option with color film (with black-and-white, it reads all of either the black or white channel as “defect,” so it’s not a very helpful tool) he emphasized that temporal analysis certainly has its flaws. In Burton’s words,
One of the downsides of that kind of analysis is that it’s kind of binary. It’s on/off, yes/no, is this a defect or isn’t it. Sometimes it could see some of those defects as something that isn’t really a defect and take out something you don’t want it to take out…[but] sometimes the information in that adjoining frame isn’t exactly what was in this frame. For example, cameras in motion: no two frames are exactly the same. For example, if it’s trying to repair a dot here and it takes something from this frame over here and puts it in…you may not see the dirt anymore, but you may see something else that is not exactly right. (Burton 2013)
What Tom is describing is the birth of the digital artifact, the enemy of digital restoration engineers far and wide. It is one of the primary examples why humans are still incredibly relevant to the field. There is simply no way that automatic engineering can “fix” all the damage or smooth out all the ruffles of a print with problems. Machines and software are great, but human care is still a necessity.
Just ask Larry Chernoff and Kevin Manbeck of MTI. When Sony approached them for the restoration of David Lean’s Lawrence of Arabia (1962) it was due to the severe amount of damage that had been found on the elements. There were enough issues that average systems were just not appropriate. MTI was called on to design custom algorithms as part of the restoration software to assist in fixing the damage. As Grover Crisp writes in his notes for the film,
Throughout the film there were sections, which we thought were mostly in the 2nd unit footage, where the camera rolls had suffered an exposure that resulted in the emulsion drying and cracking, resulting in hundreds of small vertical fissures in each frame. Although this problem had always existed, looking a bit like white columns of light or streaks on release prints, the problem had clearly gotten worse over the decades. These cracks are thin, irregular lines 1 to 3 pixels wide separated by as little as 1 pixel in a 4K frame. We found this particular problem to be evident in almost every reel of the film, so it was not relegated to the 2nd unit footage. Each defect was unique and one solution would not work for all. (Crisp 2012)
Now one might think that a certain amount of temporal analysis would work in this circumstance, but to return to the platitudes discussed a bit earlier, no two restorations are the same. While films like The Invisible Man (James Whale, 1933), Sunset Blvd (Billy Wilder, 1951) and others may have been able to rely more heavily on the standard provisions of DRS techniques, this was not possible. Lawrence was simply too damaged. Chernoff explained that in order to solve the cracks in the emulsion and the bits of “light” MTI’s CTO Kevin Manbeck analyzed the problem and came up with
A mathematical algorithm that was able to look at all the pixels, determine where the cracking occurred, and then fill those cracked pixels with adjacent pixels…it was about 70% effective, so the balance of it had to be done with a little bit of smoke and mirrors. This meant an artist (namely myself, in this case) working side-by-side with the programmer, Kevin, and determining when things were starting to become visible again and whether the algorithm could be fine-tuned for that area or whether we had to do something else, like painting pixels…It was really a combination of math and artistry to solve all the problems.” (Chernoff 2013)
Lawrence itself was scanned on two Imagica XE scanners, using the original picture negative. This was after a great deal of preparation of the materials, answer printing to a silent 70mm print (used to revise the color timing), and the creation of a new 65mm interpositive for protection purposes. For a world that is going digital, the restoration of Lawrence of Arabia is a perfect example of the hybridity of our professionals and their work. As Grover wrote about Lawrence, “The plan was to repair the damage to the film that couldn’t be addressed in [the previous 1988 photochemical restoration] simply because the technology to do so did not exist at the time. The digital tools available today allow us to repair film damage in a precise manner so as to render the film as close to its original look and sound as possible.” (Crisp 2012)
When asked about the use of the Imagica scanners, Grover wrote me via email that “since this is a large format film (65mm) there are only a few scanners that can actually handle that size and format.” He also mentioned that there are only two locations in Los Angeles that work with materials fitting this size/format description so he decided to select the one which whom they had the best relationship with and the one that dealt with actual filmic elements, Fotokem. In regards to these same scanners, Andrew Oran of Fotokem has stated in the past that, “Digitizing 65mm classics at FotoKem occurs on one of our two matched Imagica XE 65mm scanners, nicknamed ‘Big Foot’ and ‘Yeti’ due to their sizable footprint. Armed with 11K sensors, they are the world’s only film scanners able to sample 65mm negative at 8K resolution ‘perf to perf’…While the sampling of the negative element occurs at 8K (wide) resolution, digital mastering – which moves us from preservation into the realm of restoration – occurs at either 4K, 2K or HD resolution.” (International 70mm Publishers, Netherlands 2010) Imagica XE scanners are CCD and pin-registered, and equipped to handle a wide variety of film formats. The original scan of Lawrence was done at 8K because, according to Grover, that was “the equivalent resolution of the 65mm negative.” (Crisp 2012) However, it is impossible to work in 8K due to bandwidth so the 8K scan was brought down to 4K.
This extremely high-resolution scan served as the big reveal for Lawrence. While the film had looked nice before, the amount of detail that was now visible due to digital tools raised the film to a new level and gave it a whole new set of conflicts. Larry Chernoff remarked,
In certain areas, I would say, the film was unwatchable because of the problems, particularly because Grover and his team decided they wanted to do it at the highest resolution, which was an 8K scan, so it revealed the problems that much more. There was no obscuring the problems, actually, because each pixel was exactly what was on the film if not better, in a sense. So when we worked in 4K, we really did have a perfect representation of every film frame and therefore a perfect representation of every problem. I would say the result after some scenes were unwatchable was that they were watchable. In some cases they were perfect, in other cases they were just slightly less than perfect but it really would take a discerning and knowledgeable eye, somebody who knew there was a problem to be able to discern there was a problem. We were very proud of the work we did on that.” (Chernoff 2013)
As Lawrence and the MTI experience demonstrate, the depth that a higher resolution scan can get to with restoration is measurable. Had this film been scanned at 4K and brought down to 2K or been scanned at 2K, the film’s injuries would not have been as noticeable. Would that have affected the way the average viewer saw the film? Perhaps not. But for restoration purposes this is essential. As we move forward and we lose actual film as a medium, we want to make absolutely certain that the restorations we are creating are the best we possibly can. This means going as high as possible so that an issue like emulsion that is lifting from the base is visible and can be fixed by a unique algorithm. While Chernoff, who also worked with Sony on their digital restoration of The Bridge on the River Kwai (David Lean, 1957), was on the fence about whether there were any digital techniques from that project that he felt would be able to carry over to more restoration work, he was positive on the Lawrence restoration and expressed that there were significant things that could be taken from the experience and applied to digital restoration on the whole.
Color Matters and Preservation
When these moving image materials are being digitally scanned, it is not just corporeal damage or “defects” that are found to be the problem. Over the years, there are many things that can happen to a film print. Careless film handling and the uneven labwork that has existed over the years has produced everything from the “simple” wear-and-tear of torn perforations and base and emulsion scratches, to chemical stains, dust, dirt or other pieces of the environment in the emulsion/printed into the film. While photochemical restoration found ways to cope with and correct these film injuries (wetgate printing, careful cleaning, replacement of the “defective” sequences and subsequent creation of new elements) there was a limit to what the photochemical world was able to do. Beyond what could be considered primarily human-catalyzed film damage (which includes impairment through use or poor development/element creation), time and chemistry have been some of the most substantial enemies of the moving image. Not only have they made it excessively hard to put films through a scanner due to film shrinkage, but the issue of color fading has been a high-level source of frustration.
Joe Lindner, preservation officer at the Academy of Motion Picture Arts and Sciences, has spent a great portion of his career working in photochemical preservation and restoration. While he named many examples of ways in which photochemical restoration and preservation can help a film survive and endure, he admits that one of the most problematic areas for photochemical restoration is color. It is where the tools of the digital environment have really made a difference. When I asked Lindner specifically how one might photochemically restore a film suffering from some kind of dye loss, he told me that while there had been attempts to resolve this in the past through photochemical means, it was an area of great conflict and has been overtaken by digital tools that are better able to handle these issues. (Lindner 2013) Color’s variations, chemistry and inconsistencies lent themselves far better to the algorithms of the high-tech world than what could be done in an analogue fashion.
On the other hand, when it comes to preservation, we are still actively using photochemical materials to engage in the conservation and protection of not only our images but their colors as well. While studios used to make YCM masters primarily for insurance purposes (meaning they weren’t always meticulous about checking the materials for integrity or accuracy), the environment has changed and become one where archival ethics and element maintenance is more of a concern. As of March 2013, it is likely that most of the film companies currently producing film stock will only be doing so for a few more years, as we make the transition to digital. But we still have yet to locate a digital preservation standard. This is a valid anxiety. From the standpoint of many people, the only entirely “tried and true” preservation material is film itself. And yet film has shown itself to be faulty as well. It deteriorates, it is a chemically based element, and it is not perfect. But it is a far more reliable standard than what we are being offered by way of any of the digitally based suggestions.
Tom Burton mentions that the use of YCM separation masters is not simply for older films and restorations. For many of their studio clients, Technicolor creates YCMs for new releases as well. These YCM separation masters are preservation elements that are intimately connected to both restoration and color. As black-and-white materials that carry the very essence of each of the separate colors, they create a collection of materials where the colors cannot fade because they are in black-and-white. Within color film, the dyes are organic and break down/fade with time. In these black-and-white elements, silver metal is utilized, which holds the color dye information and locks the data in. YCMs are clearly of high importance as there is long-term preservation to consider. Even the YCM model is one to consider as we move forward, due to its ability to hold information better than the original elements. Burton states,
We take the DI (digital intermediate) of the final film and its data. And [the client has] made their DCPs and they’ve made their release prints and whatever else. But now you have this data and you want to preserve this data. You can put the LTO cartridge away …and we all know that you’ll have to migrate that to something else pretty soon and it’s a big pain, we all know that. So what do you have as an alternative to that? And unintuitively, it’s like, ‘Well, let’s make film!’ The only thing we know today is that black-and-white film will last for hundreds of years…we know that if it’s taken care of and put away and we know that if its black-and-white, it doesn’t have color dyes to fade, so if we take care of that…this is, currently, the best, most predictable, long-term archive solution that we have. (Burton 2013)
Long term moving image preservation is crucial as is the necessity to find an option for retaining proper color registry. However, film itself is facing a time limit and an expiration date. This places more pressure on the digital development world to find an alternative for the tangibility of YCM separation masters and their reliability. While there is a great deal of suspicion around the various preservation solutions that are “not film,” there are equally as many media professionals using and espousing the glories and wonders of “not film” as a tool for the advancement of film-making, film restoration and the exploration of moving image work. We are making significant inroads into the digital realm for restoration and filmmaking every day. The faster we move forward, the more reliant we become on machines and technology, and the more dangerous it becomes that we have no digital standard for preservation. The push/pull tug-of-war that is happening as film leaves the battlefield will only be resolved with the passage of time and the discovery of one or more devices that do not have the same backwards-incompatibility issues that LTO-tapes have or other problematic features. The positive aspect of the digital landscape is that a variety of companies are working on solutions for this (it would be difficult not to see both economic and cultural returns in the development of such a product, were it successful) and the digital world is many times faster at development than the photochemically based one. The negative aspect? Now that we are so heavily digital, we are producing content that fast as well. It’s a Catch-22 and the outcome is still unclear. As time moves forward, we can only hope that a standard is reached and soon.
From Joe Lindner’s perspective, for restoration, the onset of the digital age is not the worst thing that could have happened nor is it the media apocalypse that many pundits of film journalism would have everyone believing. As he explained to me, color work in film restoration improved greatly with the introduction of digital tools. He notes, “There were [some photochemical] techniques of printing color that was faded to fix them and some of them were essentially adapted to digital.” (Lindner 2013) Additionally, Lindner, who has been working in the field for quite some time, has seen the economics and equipment change and said that he and his peers have always been looking for a “better toolbox” to work from for restoration work. When new technologies began to appear (mentioning the appearance of Steven Spielberg’s Jurassic Park in 1993 as a watershed moment) those working in film preservation and restoration began to see that previous limitations and confines were about to expand and the industry people got quite excited. (Lindner 2013)
Leo Enticknap writes that as far as restoration is concerned, the downside of its transition to digital lies primarily within the ethical camp and not within what can actually be done to assist the elements in their aesthetic value. He writes, “the godsend is in the use of digital intermediate technology to carry out film restoration functions which could previously only be dealt with by photochemical methods. Until the late 1990s…elements with faded colour dyes could only be restored by photochemical duplication, specifically by manipulating the light source used to expose the destination stock. These and other defects can now be corrected using what is essentially the same technology that gave us the special effects in Jurassic Park and Terminator 2.” (Enticknap 2005) He goes on to discuss how the real apprehension in this arena should not lie in new equipment or software but in the individuals operating them. Much in the same way that the automated scanning and restoration software can lead to a variety of mistakes such as digital artifacts or the over application of grain-reduction, under-educated restoration professionals (or those not provided with enough contextual objects to work from/with) will create a media object that is not as true to original form or intent as it could be. As Enticknap writes, “[there is] the possibility – indeed, some would say the probability – that films will find themselves being ‘restored’ by individuals and organizations which either do not have the contextual and historical knowledge needed to evolve a model of the ‘original’ state they are aiming to restore…or are working more toward a commercial agenda than a cultural one.” (Enticknap 2005)
According to Giovanna Fossati, one of the biggest stumbling blocks for any color restoration is the lack of reference for the original colors. She writes,
In most cases, all original elements of a film have suffered the same kind of color deterioration and, as a result, a truthful benchmark for reconstructing the original colors no longer exists. As a consequence, film restorers often need to guess…what colors are to be restored. In fact, the same goes for other, if not all, aspects of restoration, where the target reference no longer exists or, if it exists, it is in such a deteriorated state that it can only be conjectured. The restorers’ work is based on their knowledge of the historical context from which the work to be restored originates, of the technology used to produce it, as well as the knowledge of the work itself and its maker(s). (Fossati 2009)
While Fossati’s point echoes Oran’s SMPTE remarks to a degree, it also reiterates Enticknap’s position in regards to professional training. In this period of technical transition, moving image facilities are in a quandary: there is a deep need for people who are qualified to do the restoration work and conduct the research in the most thorough manner possible. At the same time, there is a significant desire to have expert technicians for all the digital work that must be done. Will the onset of the digital environment affect the workforce? Will positions of preservationists and restoration professionals begin to shift more towards those with high-level backgrounds in computer technology and visual data? It is not impossible. It would, however, be unfortunate.
All Together Now: Culmination of the Process
A great example of the way in which these processes and ideas come together can be seen in the live-action restoration work that is being done at Disney. Many studios and archives are producing legitimately amazing digital restorations from heavily compromised materials. What Sony did with MTI in the restorations of both Bridge to the River Kwai and Lawrence of Arabia was nothing short of miraculous while Tom Burton and Technicolor’s assistance in rehabilitating Universal’s Invisible Man made the film not only watchable again but provided the piece with the necessary tech work to have full sequences that were formerly defective or lost. Within this final case study I will show how the work that Jayson Wall is doing at Disney displays the necessity for diligent and thorough research and education on all aspects of a given film for its proper restoration. As stated earlier, a film restoration consists of a product that gets as close to the original experience as possible. A good restoration work will construct a careful balance between heavy research, human intervention and digital assistance. From the discussions I have had with their staff and clips that I have seen, Disney is a good template to use to demonstrate this tender equilibrium.
When I spoke with Jayson about his restoration workflow, he really stressed research and knowing every angle and aspect of the materials that he works with. We discussed everything from the fact that he makes sure to color-time for carbon-arc light bulbs (they are a different temperature than modern xenon bulbs and a vintage 1950s print should therefore have a slightly “warmer” look) to noting how he builds a “blueprint” for the film from documentation like release patterns, edit lists, and print origin (national or international). Jayson’s entire restoration methodology was based on strict research, resource development and reference management. If direct references are unavailable from the precise film that is being worked on, Wall was very specific about knowing the film stocks that Disney works with and utilizes and what years, and was able to list a variety of other reference materials that they might use on the chance that the desired reference items were not accessible. For someone who handles digital restoration, a significant portion of our interaction was spent relating the digital landscape of Disney’s workflow to its photochemical genesis. One thing became quite clear as well: the Disney Archive itself is highly organized and well managed with a wide collection of materials for restoration projects such as those Wall had headed up. (Wall 2013)
Once a negative is selected for restoration, Disney uses a company called Pro-Tek for inspection. When I inquired as to why they would use an external vendor for this procedure and how Pro-Tek was selected, Jayson responded that it was in order to be as thorough as possible. When he works on a film restoration, he feels that it is part of his job to make sure that no stone is left unturned when it comes to the process. He said,
I had a relationship with Pro-Tek back when I worked at New Line, they were always our preferred vendor, and then I worked for Pro-Tek at one time so I know the ins and outs, the people who work there…And it’s nice to have a third party to inspect the negative instead of your lab, your facility, just to get a cross-section. The labs will tell you certain things about a negative, but I want to know the entire picture of the negative. Having worked in film, having the photochemical background that I do, I want to know all the problems with the negative that will help me two months down the road when I’m in the color bay. So I can say, ‘Oh, this is why the picture shifts, it’s because the perfs are torn at this frame,’ or whatnot. I’m always looking for all that information. I want all that metadata about a title before I go into scanning. So research is a huge huge part of the restorations.” (Wall 2013)
After the inspection, the negative is usually cleaned and prepped for scanning. This is the stage in which major decisions begin to be made based upon all of the research that has been done up until this point.
As previously stated, Wall has very specific terms and conditions for scanners and many come from his photochemical background. Many of the variables and choices for scanners he mentioned dealt with things like film stock and whether a scanner is pin registered or not pin registered. For Disney, a studio whose moving image work (both animated and live action) is highly color-centric, this raw digital information is extremely important, so the machines that transfer it should be chosen with care. As the earlier discussion about Lawrence of Arabia pointed out, high-resolution digital scans maintain the capacity to reveal a great deal more information and on a wider scope and range. While Disney may not be scanning everything at 8K for the restorations that Jayson and I spoke about, scanning at 4K can still expose all the inconsistencies in the physical film text that a restorationist would like to solve, dirt, dust or dye-fade related. For Jayson, his end game is to assure that the work looks as fresh (or as close to it) as the day it was released. This means lots of work in the color bay. The digital color world works in two different “colorspaces,” Rec 709 and P3. Rec709 covers a much smaller and more conservative area of the color spectrum and, while still a digital colorspace, is essentially more economical and designed primarily to work with standard and high-definition formats. The P3 colorspace is specifically oriented for digital work and offers more contrast and a larger color “gamut.” Thus, when I asked Jayson about final deliverables, it made sense that P3 was part of that equation. He told me that while it depends on the title,
On a 4K restoration, we will have our raw color-corrected scans in the P3 colorspace, which is the wider color gamma space versus rec709. P3 is where you do your D-cinema, where you have your DI workflow. It’s closer to film. We will have those two files, we will make a DCP from the P3 files (not from the rec709), we will also make clipster projects, which are really your files. We’re an all-file based studio, and have been for a while…maybe two years now. So we will make a master clipster project of all the different aspect ratios – 20,000 is a film that has three different aspect ratios – and then we have a servicing file called a pro-res, which services television and other platforms, and that will be done in rec709…we will also do film-outs on titles as well. We are covering all the bases on our restorations.” (Wall 2013)
Disney’s deliverables, much like their extensive project assessment and restoration, work towards preservation and restoration. Since there is no digital standard for preservation, they provide film outs. On the other hand, they have made the jump to a digital workspace within the company, are taking the transition seriously and have put systems like that involving clipster projects into full operation.
While the level of archival information varies from project to project and each studio and/or archive has its own varying amounts of historical assets and resources to draw on for restorations, it became quite clear during my interview with Jayson Wall that Disney is able to conduct a significant amount of strong research on its restorations due to the records that the studio kept. I queried Wall about this issue and he told me that Disney has a really strong archival sense. He spoke of the several different archives that Disney has to contain their work. It is no accident that they are able to pull original Technicolor timing sheets for assistance in getting the colors right according to how they would have looked at the time of production. He stated,
We save a lot of metadata…a lot of the documents are around. On 20,000 we researched how it was shown, what sections were shot on VistaVision, what sections were shot on CinemaScope, which lenses were used, a lot of that stuff exists! We have a lot of that material and when it comes into technology such as 20,000 having the Technicolor cards from 1954, and we can kinda sit there and we have a guy on our team who’s an old-school colorist who can read the cards and who can also read the tapes…we take all that information together. And by and large, we take what the director approved at the time. In that case, it was what he approved in the late 1980s. (Wall 2013)
As much research and documentation as Jayson mentions using here, this statement manages to return us to a previous point: professional training and the current digital transition. Jayson has access to individuals who are thoroughly trained in the use of the materials that Disney has within their archives. Much like Jayson himself, this “old-school colorist” is like many engineers in the field, and likely has specific knowledge that began in a photochemical environment. It is not alarmist to say that the assembly of analog experts is only going to get smaller with time. While Jayson himself is highly trained in photochemical work, he did not start or specialize in the kind of work that his colorist did. He relies on him for certain things in order to make the final product resemble the initial work as closely as possible. The question then remains: what will we do when we are left without this base of skilled professionals to draw from to assist in this work? What is being done to pass on that all-too-valuable knowledge?
Electric Dreams
New technologies go with the moving image like freezers go with ice cream or wheels go with cars. Their designs not only work with each other but also allow the other to continue on their path to survival. The history of the cinema has had a variety of technological changes, each affecting its audiences and the moving image culture on the whole. What it has yet to do is leave a gaping hole where film once was. Film restoration is a unique process by which individuals attempt to re-engage historical moving image materials within an entirely new set of terms and conditions. While many terms are bandied about – reconstruction, authorship, workflow, automation – the one thing that has not changed over the years is the necessity for human involvement in the process of making history reappear on a screen. Corporations may want to lessen their expenses by increasing electronic involvement over manual labor, but, much like photochemical restoration, there are limits to that. There are only a certain amount of people willing to trust the digital processes currently available, computers make consistent errors in the “automatic” procedures, and the transitory state that we are in requires far more intimacy than we are currently giving it. We may be moving to more highly developed machines than Steenbecks, but the ways in which we are working with the machines and moving the information through needs to be more fleshed out. It’s not just the materials that we need to consider. It is those who created it and have been working with it for years. We are restoring film and documenting that process, but we should also be working to document the restorationists themselves and their valuable knowledge.
While a film restoration will never be able to give a viewer the precise experience of what it was like to see Clara Bow in Wings in a theater at the premiere, what it manages to do is give the audience a unique and special experience of materials that may have disappeared forever without proper attention and care. Digital restorations are not the same as the original works. They are not made of organic material like nitrate nor will they have the same inconsistency that the silver halide crystal development process gives film. But these materials have a different identity: these are digital media objects. They are not photochemical and they are not trying to be. They are not there to try to “trick the audience” or pull a fast one. What they are is the end result of the desire to save important cultural product from disappearing forever into the void by accessing current visual technologies and exploiting their capabilities. Through research, scanning, delicate attention and transferring to final product deliverables, a once-suffering film can receive life again. While it may be a different visual in certain respects, the reason the filmmaker created the object was to tell a story and the story will continue to be told for that much longer instead of reaching termination. This, in itself, is a reason to believe in digital restoration.
Works Cited
Burton, Tom, interview by Ariel Schudson. Digital Restoration at Technicolor (February 22, 2013).
Chernoff, Larry, interview by Ariel Schudson. Digital Restoration at MTI (February 25, 2013).
Crisp, Grover. “LAWRENCE OF ARABIA: 4K Digital Restoration Notes .” Notes, 2012.
Enticknap, Leo. Moving Image Technology: From Zoetrope to Digital. London: Wallflower Press, 2005.
Filmlight, Ltd. “Filmlight Ltd Datasheets.” Filmlight Ltd. 2013. http://www.filmlight.ltd.uk/products/northlight/overview_nl.php (accessed March 8, 2013).
Fossati, Giovanna. From Grain to Pixel: The Archival Life of Film in Transition. Amsterdam: Amsterdam University Press, 2009.
International 70mm Publishers, Netherlands. Digital & 65mm: History and Development of 70mm. Arnhem: International 70mm Publishers, 2010.
Kalas, Andrea, interview by Ariel Schudson. Digital Restoration at Paramount Studios (February 14, 2013).
Lindner, Joe, interview by Ariel Schudson. Photochemical Restoration Interview with Joe Lindner (February 12, 2013).
Lipman, Ross. “The Gray Zone: A Restorationist’s Travel Guide.” The Moving Image (University of Minnesota Press) 9, no. 2 (2009): 1-29.
Oran, Andrew, interview by SMPTE. Today’s Use of Black and White and It’s Potential Future (March 7, 2013).
Read, Paul and Meyer, Mark-Paul. Restoration of Motion Picture Film. Oxford: Butterworth-Heinemann, 2000.
Wall, Jayson, interview by Ariel Schudson. Digital Restoration at Walt Disney Studios (February 15, 2013).
Wallmuller, Julia. “Criteria for the Use of Digital Technology in Moving Image Restoration.” The Moving Image (University of Minnesota Press) 7, no. 1 (2007): 78-91.
Progress Not Perfection: The Work of Ariel Schudson 