Maria Falconetti in Carl Theodor Dreyer's The Passion of Joan of Arc (1928). This is how the frame appears in the current Eureka!/Masters of Cinema Blu-ray disc edition of the film. Below, the image before restoration.
I first heard from James White, the British film restoration maestro, a few years back, around the time the below-mentioned ITV Blu-ray of Powell and Pressburger's Black Narcissus came out in Britain. White and I reconnected around last year's Christmas holidays, just as two very disparate projects in which he had a restoration hand were coming on the scene: the above-depicted version of the Dreyer masterpiece, and Arrow Video's iteration of Lucio Fulci's notorious Zombi, known in Britain as Zombie Flesh Eaters. We figured the time was right to catch up on some of the issues concerning film restoration in the brave and still relatively new but increasingly pervasive world of digital, and thus an e-mail exchange began. It took some time, but I think the results are informative, and contain news that is both heartening and distressing. If you can, you should read in conjunction with Nick Wrigley's essay "Crimes Against The Grain" in the December 2012 issue of Sight & Sound. Wrigley was one the founders of the Masters of Cinema imprint and worked as a restoration supervisor with White on the Dreyer project. The interview with White follows:
1) Film restoration seems to have pretty definitively shifted from a photochemical process to a digital and electronic one. When was the tipping point for this, and what does it mean in terms of the economics of film restoration? Not to mention the distribution of restored motion pictures?
It's certainly true that over the past decade or so, film restoration has moved from a largely photochemical process to one relying almost completely on digital technology. When I began working in this field in New York back in the mid-1990s, digital restoration was still in its infancy and most of the major projects such as those being overseen by Photoplay (Napoleon) or Harris & Katz (Spartacus, Vertigo) were being produced almost solely through traditional photochemical processes. These days, however, if someone embarks on a new restoration, they will almost always begin by scanning the best existing film materials and working in digital throughout the entire process.
My time at the BFI reflects this shift in approach. When I started working there in 2002, film restoration was still very much the preserve of the archive lab, with digital technology having little to no involvement. Restorations from that era represented the best that could have been achieved at the time, but in recent years the BFI has seen fit to revisit many of their key titles using digital tools, having seen the advantages these bring. A title like Blackmail for instance, restored last year by the BFI as part of their silent Hitchcock project, now looks amazingly improved in its digital incarnation in comparison to the prior restoration completed in the 1990s.
I don’t think there’s been a single tipping point that made this change happen, other than restoration simply mirroring the film industry at large, and its shift to using digital tools for pretty much everything under the sun. The simplest reason though, is that the tools just got better. Most importantly, the software became sympathetic to the needs of archive films and offered solutions to problems that traditional restoration hadn’t ever been able to deal with sufficiently. Issues that had always been difficult to impossible to fix -things like image stability, density issues, deep emulsion scratches, warping, registration issues, deterioration brought on by heat or moisture, and damaged or even missing frames – these could now be improved upon in a way that just couldn't be done by traditional means.
Having access to these new tools hasn't made the work cheaper, though. If anything, restoring a film has become more expensive simply because audience expectations are so much higher now. Fortunately there are so many more outlets to distribute these titles, so the economics are there to support the work. Mind you, there's definitely a limit to what most distributors can afford. A top-tier restoration like The Red Shoes or Lawrence of Arabia can demonstrate beautifully what digital restoration is capable of - these films now truly take your breath away - but the vast majority of film titles don't enjoy anything near to the restoration budgets these projects had. So while restoring the majority of film titles to a 2K/4K level might be beyond the means of some distributors, restoring a film to a very high standard in HD is often within their means. I should mention that going this route doesn't support a film's archival basis - restricting your end result to HD doesn't create what most archivists would agree is the basis for film preservation - but it does provide distributors with a format suitable for the majority of current release platforms (DCP, Blu-ray/DVD, etc).
In any case, what's encouraging is the high amount of restoration work currently going on internationally. Rights-holders of film libraries have started to see the value in restoring their titles to make them accessible for HD broadcast and online platforms. DVD/Blu-ray sales have helped create new audiences for archive titles, and the improved presentation quality has made viewers more aware and appreciative of the practice of film restoration on the whole. Back when I started, Criterion was the only company around that devoted time, attention and a decent budget to giving a film the best presentation possible on video, but now in addition to well-established labels like Eureka/Masters of Cinema, the BFI, Milestone and Kino/Lorber, you've got all these new labels like Olive in the US, Second Run in the UK and Edition Filmmuseum in Germany resurrecting lost cinema or rarely-seen classics. It's the work of companies like these that is really helping to bring new interest in film restoration. I mean, it's great when Casablanca or Singin' In The Rain gets treated to another state-of-the-art restoration, but there now seems to be an actively growing interest in discovering new films and film collections from all corners of the globe. A project like Milestone using Kickstarter to fund a new restoration of Shirley Clarke's Portrait of Jason is a great example of the public taking an active role in film restoration. Likewise for Distribpix's recent restoration of The Opening of Misty Beethoven - that film now looks better than anyone probably ever expected it to!
2) The tools for film restoration have evolved to the point that the lay viewer is under the impression that it can perform miracles, which to some extent seems objectively true. So let's look at it from the other end of the telescope, as it were, and talk about the things that the tools CAN'T do.
Well, as I said there’s a great amount of things that digital tools can do to repair the issues that have always blighted older film titles that we couldn’t do just a few years ago. That said, it’s important to bear in mind that in the most fundamental sense, no amount of restoration, digital or otherwise, can significantly improve the quality of the image of the film element that it’s sourced from. The issues I’ve mentioned - dirt, scratches, stability, flicker, missing or damaged frames, etc – these can all be improved significantly through the careful application of digital tools. But the basic details within that original image, meaning the film grain, the level of detail, etc – these can only be improved within the parameters of what the original element would allow.
It’s like up-rezing an image from SD to HD. An increase in detail doesn’t happen simply because you’ve added more pixels. Of course one can give the illusion of increased detail or sharpness through digital enhancement, but then it’s no longer the film you’re working with, it’s something else. I work to create the best representation of an old film possible, but that means keeping the results within the historic bounds of what that film would be able to achieve at the time – be it its film grain, the saturation of its colors, the level of its contrast, etc. A film from 1930 shouldn’t look like a film from 1950, and it most definitely shouldn’t resemble a film from 2013. Whatever tools you’re using should be done in service of what the film looked like at the time of release, not some ill-advised notion of what it could look like now. That, for me, is probably the most important thing.
Unrealistic expectations are a common source of frustration for film restorers, as often the best existing film elements simply won’t allow for a "miraculous" result. The project is only as good as the material it's sourced from, and not all films have been subjected to the same treatment over the years. Often with older titles the original negative has been lost, or is far too damaged to work from. In this case you might be relying on the next best thing, like a second-generation interpositive or fine grain element. But there may be problems with these materials as well. Often restorations have to make use of more than one element, each subjected to vastly different histories, and then it can prove a real challenge to make the overall results consistent. To put it simply, every film is different, and every restoration has its own set of challenges.
One interesting development of late has been the discussion of the differences between what we’re now able to see digitally in comparison to what was visible on a film print. The Wizard of Oz scenario is well-known – in Warner Brothers' new Technicolor restoration you could suddenly see wires holding up the Scarecrow, whereas in all previous releases of the film, they were invisible. Clearly back in 1938 Victor Fleming and his Director of Photography understood enough about the photochemical process that they could expect the wires to be invisible by the time theatrical prints were created in the printing chain, but in working digitally from the first generation elements, the wires are now there, clear as day. Likewise, I've heard it recently remarked that with Vistavision films such as Vertigo we’re now seeing a level of detail on Blu-ray that was never intended for the cinema, as theatrical prints would have been made from separate matrices reduced for printing. So how does one approach situations like these? It’s an interesting dilemma.
3) What are your favorite/favored tools, or the ones you find yourself applying more often? What have you had a hand in developing?
I can’t claim to have had a personal hand in the development of any specific software tool, but I’m fortunate that I work with such an amazing team of technicians at Deluxe Labs in London, who are always working to find new ways of meeting the numerous challenges this kind of work presents. We’ve worked together on so many projects over the years for the BFI, Eureka and most recently Arrow (Zombie Flesh Eaters) we’ve developed a great rapport together, which is crucial when you're working on archive titles.
With picture restoration there's a variety of tools at hand that benefit some films more than others. I'm always cautious about any so-called "automated" processes, as it's so easy for these tools to have an unwanted effect on film grain and detail even if applied carefully. So while it takes far longer to do, nothing works better for me than old-fashioned frame-by-frame cleanup, performed manually. It can be incredibly complicated, time-consuming work, but there's just no substitute for treating each individual frame with the utmost care. Full disclosure, though - I don't do any of the hand-on stuff myself anymore. As a restoration supervisor, I oversee all the steps of a project from the initial research and selection of film materials to the reviewing of scans to the full grading of the film to all stages of picture and sound restoration. Whenever possible we'll work with the director or director of photography (if alive and available) and we always work to deliver the highest quality representation of a film's original release within the parameters the budget and time frame allow for.
4) The two projects that were the springboards for this correspondence were pictures that seem to have almost as wide an aesthetic gulf as a temporal one: Carl Theodor Dreyer's 1928 masterwork The Passion Of Joan Of Arc, which you supervised the Eureka!/Masters Of Cinema Blu-ray version of, and the aforementioned Zombie Flesh Eaters, the notorious 1979 Lucio Fulci horror cult fave. In terms of materials, each one must have presented unique challenges. But there's also a question of properly serving a given film's aesthetic. I understand I'm constructing a bit of a big tent question here but I'm interested in your thoughts on all of these considerations.
One could hardly choose two more different films than those two, could they? Though I still think they'd make a great double bill! Hey, they both feature heavy doses of pain and suffering, both base the reasons for their violence in religion of one kind or another, and both really know the power of a well-executed close-up. Although it's true that only one of the films features a punch-up between a zombie and a shark...
Anyway, the fundamental criteria for me is always to present the film as close to how it originally appeared in the cinema. That means restoring the film to the highest quality possible but not employing any means to "update" or "improve" the image or the sound in some misguided effort to refashion it to fit in better with modern-day expectations. I want the results of something I restore to appear as a film artifact, not a digital one, which is why I've been using the term "preserving a film's photochemical integrity" lately.
The Passion of Joan of Arc project grew out of Eureka's interest in releasing the film on Blu-ray and the Danish Film Institute's desire to see Carl Dreyer's original Danish version made available to the public. As far as I'm aware, Gaumont are still planning to restore it themselves, but the version they eventually produce will almost surely be the French version we've all been familiar with for years. So as this stood as the only likely chance to have Dreyer's debut version seen to properly, we wanted to make sure we presented Joan in the best and most accurate representation possible.
We had the good fortune to be working from the Danish Film Institute's preservation materials, struck directly from a first generation 35mm print discovered in Oslo. This is as literally good as it gets with this film, a film we're amazingly lucky still exists in any form given the problems it faced from the very beginning. For those unfamiliar with the saga this film was put through, the story bears repeating, so I'll crib a bit from a piece written about the restoration for Moviemail last year:
"The trouble began just six months following the film's Copenhagen premiere in 1928 when the original negative was destroyed in a fire and the two original prints created from this element were subsequently lost. A new second negative, incorporating extra footage not featured in the original version, was subsequently cut together by Dreyer but this material swiftly fell victim to misfortune, and was presumed to have perished in a lab fire as well. Although fires like this were fairly commonplace at a time when highly flammable nitrate film stocks were used, the fact that the film had been destroyed on two separate occasions seemed to imply that The Passion of Joan of Arc was destined not to survive in any form.
Over the years that followed, several incomplete prints and material believed to represent the second lost negative re-surfaced; as a result numerous versions of the film have been edited together and screened for international audiences, but Dreyer’s original version remained lost, presumably forever.
Then in 1981, one of the original prints struck from Dreyer’s first negative was miraculously discovered at a Norwegian psychiatric hospital. The Danish Film Institute (DFI) immediately acquired this material, which came to be known as the "Oslo print" and created a new preservation negative, guaranteeing a secure future for the film and keeping the elements in optimal conditions for years to come. Finally, after over half a century, The Passion of Joan of Arc could finally be seen exactly as Dreyer had originally intended."
Before-and-after restoration images from Joan.
So with all that history facing you, the responsibility you bear to remain faithful to the film and the materials at hand should be always first and foremost on your mind. Given that we had a specific budget and fairly tight schedule to work by, my approach was to do the best job possible but to be wary of not "over-restoring" anything. The entire film required stabilization due to the shrinkage and sprocket wear the print had endured throughout its 80-plus year lifespan. Every join between shots required manually re-setting as there were bumps on literally every cut - a fairly daunting task for a film as creatively and heavily edited as Joan. Density inconsistencies manifested in the form of heavy flickering had to be significantly reduced. There were numerous instances of damaged and missing frames, often requiring new frames to be digitally interpolated, as well as a steady stream of heavy and light scratches throughout. Not all of these issues could be remedied completely, and some problems could only be marginally improved. If you watch the restored version of Joan, the film still exhibits quite a few of these issues, although it's now a fraction of what we started with. The important thing is that none of the work we did compromised the original photochemical look and feel of the film, and the original texture of Joan remains. Fortunately most of the feedback we've received on the Blu-ray has been very positive, so I think on the whole people appreciate this approach, even if the results aren't 100% pristine, which would be a very tall order with any film from this era.
Zombie Flesh Eaters was quite a bit different, in that it's a film that's been widely available for years, most recently in a fairly decent Blu-ray presentation from Blue Underground. But Arrow felt strongly that the film deserved better, so they decided to fund a new restoration of the film on the basis that we would be granted access to the original Techniscope negatives. A little digging revealed that these elements hadn't been made available for transfer since the film's original print run (contrary to a good deal of misinformation that's been circulating on the web), so we had a good opportunity to correct for some home-video crimes of the past and restore the film to its original release presentation.
Techniscope was a fairly popular format in Europe throughout the 1960s and 70s - it was essentially a low-budget version of Cinemascope, using two perforations instead of the standard four. After the initial negative is processed and edited, secondary elements (Interpositive, Internegative, etc) are produced by blowing up the image to fill the 4-perf frame. Having worked on another Techniscope project a couple years ago for the BFI (La Vallée, Barbet Schroeder, 1972) I knew that using the negatives provided the ability to capture a good deal more image area in the film frame than any of the other printing elements in the film chain, as the blow-up process would have forced the cropping of all sides of the image in order to fit within the 4-perf frame. So working from 2K scans of these original negatives, I was able to bring the color palette, the grain replication and all the details present on the negative back to its proper place, as well as reveal a good deal more picture area on all sides of the frame for most of the film. None of the work was exceptionally challenging, although the footage from the famed shark-zombie sequence was blighted by a lab fault that proved impossible to do anything but digitally minimize - fortunately this issue has always been a part of the film, as it was printed into every version anyone's ever seen! [...]"