ARTFORUM | Conservation at a Crossroads


Bruce Conner, CROSSROADS (1976, 35mm, b&w/sound, 37min.); © Conner Family Trust, San Francisco

Bruce Conner, CROSSROADS (1976, 35mm, b&w/sound, 37min.); © Conner Family Trust, San Francisco


Ross Lipman on the Restoration of a Film by Bruce Conner

IN 1996, I was working at the Pacific Film Archive in Berkeley, California. Although I was the cataloguer, Edith Kramer, the PFA’s director at the time, knew I had a background in film printing and processing, and she let me hang around the screening room when the archive was preserving Crossroads (1976), Bruce Conner’s profound reworking of the 1946 Bikini Atoll atomic-bomb test footage.1 The film, a thirty-seven-minute montage choreographing twenty-three black-and-white shots of the underwater nuclear explosion to the accompaniment of a transcendent dual score by pioneering synthesist Patrick Gleeson and composer Terry Riley, is today considered Conner’s masterpiece and one of the most provocative and compelling works to address the atomic era. Having only recently joined the PFA, I was honored to be a fly on the wall as my colleagues worked.

At various stages along the way, Conner would come by the theater with Michael Friend, then of the Academy of Motion Picture Arts and Sciences (AMPAS), who was overseeing the project.2 Flash forward to 2012, when the artist’s estate approached the UCLA Film & Television Archive (where I now work as a film preservationist) to collaborate on a new edition of Crossroads: Having witnessed Conner and his team’s extraordinary dedication and the brilliant results of their efforts, I embraced the opportunity to revisit the film with both excitement and curiosity. What had changed in the archival field since the excellent—one might think definitive—author-approved 1996 preservation?3 The short answer is everything. The long answer requires discussion.

If the fundamental goal of moving-image conservation and preservation is saving works for posterity, could further intervention really be needed so soon? Clearly, technological change can create a need for action. However, in so doing, it also quietly demands a reconsideration of the fundamental terms of art’s permanence. (Similar issues arise in the conservation of painting and, to a far greater degree, in that of postwar works made in unconventional media.) One can certainly aim for posterity, but such aims must realistically be tempered with the understanding that even our best efforts will fall short, in practice as well as in theory. Under current conditions of relentless technological transformation, moving-image conservation, preservation, and restoration4 may no longer be seen as guarantors of timelessness but, increasingly, as dynamic functions of the present historical moment.

Let’s, then, describe that present. The most fundamental sea change wrought by the digital revolution is the loss of the singular work. Art conservation has traditionally been based on the maintenance of a unique physical object. In a parallel concept, moving-image works—which are mechanically reproduced in multiple copies—have demanded maintenance of their own medium integrity. Both notions are now, if not obsolete, in danger of becoming so. Rosalind Krauss’s pioneering declaration of a new postmedium condition presciently anticipated our contemporary digital culture, in that the digitization process translates physical properties into disembodied, abstract numeric values.5 Moving-image works are, in practice, subsequently reconstituted in a variety of physical forms—but what is the meaning of medium integrity in such a context?

What was formerly thought of as a singular work may now be seen as a constantly changing entity as it is repeatedly presented over time—here projected as a photochemical film print; there digitally reconstituted for all manner and size of screen. This new dispensation challenges prior notions of authenticity, even those applicable to mass-reproduced works. The authentic singular work is today largely a thing of the past, and moving-image restorationists, if they are to succeed in their mission, must envision multitiered strategies that address conventional notions of archival procedure while simultaneously considering the limitations of current exhibition contexts, even as they seek to improve them.6 Our work on Crossroads offered a unique opportunity to engage these issues in both theory and practice.


The works of Bruce Conner are particularly salient to this discussion, in that his oeuvre encompasses not just films but assemblage, collage, photography, painting, drawings, and conceptual pranks, and accordingly spanned a range of black-box and white-cube environments (though neither term enjoyed widespread usage when many of the pieces were created).7 Conner, as filmmaker and artist, adeptly bridged the worlds of international cinematheques and art galleries. Accordingly, his work must bead dressed from both film-archive and museum perspectives. His deep investment in formal and presentational qualities only increases the relevance of Crossroads to the above concerns.

The present project was conceived as a new working model for moving-image museum mastering, to exist as a limited edition of six (each including archival 35-mm copies and digital media, as further described below), with production overseen by Michelle Silva on behalf of the Michael Kohn Gallery in Los Angeles (which represents the artist’s film work) and the Conner Family Trust, and myself on behalf of the UCLA Film & Television Archive. Our goal was to attain the highest standards in archival film preservation and simultaneously to ensure the maintenance of Crossroads’s unique aesthetic character in translation to the new media forms needed for contemporary exhibition, through careful control of the digital remastering process.

If these two parallel aims seem contradictory to some, the contradiction is not as great as one might think. Throughout every step, we held two well-established principles firmly in mind: first, to never do anything irreversible; second, to document our intentions and methodology so as to not falsely represent the nature of our work.9 The realization of our efforts can be seen as an attempt at versioning, with distinct iterations created in both photochemical and digital media. These iterations targeted both cinematheque and gallery spaces in a variety of ways.

A short recap of Crossroads’s unique production history might be useful in understanding the thinking behind our dual strategy. In 1974, Conner contacted the US National Archives with a request to access the government’s nuclear-test footage. As Conner described the project for the Pentagon’s perusal, “there would not be any commentary or dialogue involved. Music and sound effects would accompany the images in a proper and natural manner. The spectacle of the underwater detonation would be the substance of the film production.”10 To Conner’s surprise, he was granted approval, and in characteristically subversive manner, he completed the film to coincide with the American bicentennial.

Both halves of Crossroads’s dual sound track were recorded at Gleeson’s studio, Different Fur Music, in San Francisco. Although Riley’s score was mixed in stereo,11 Dolby Stereo for film was not yet readily available.12 Consequently, the film was released in 35-mm standard mono sound, as well as in 16 mm, which is inherently mono.

The 1996 joint preservation by PFA and AMPAS utilized 35-mm film with mono sound and served as an excellent preservation of the original 1976 release. However, in 2003, Conner himself went back to the original stereo recordings of Riley’s score and released a new version—in standard-definition video. At this time, he also reframed many of the pictorial compositions to eliminate or reduce visible scratches in the source footage, and he made a slight editorial change in the pacing of the end credits while leaving their text unchanged.

When we approached Crossroads anew, we thus confronted a work that had itself metamorphosed under the artist’s guidance over the years, from conception to initial realization to subsequent revision, at key points responding to the practical limitations and production climate of the moment. We, too, would continue that process, even as we sought to preserve the film’s essence in the translation to digital media.


Our practical work began with a careful replication of the authorized 1996 photochemical preservation. The original effort had been conducted at the Cinetech laboratory in Los Angeles, which now, like many photochemical labs, was in the process of winding down its operations as a consequence of the paradigmatic changes that were so precipitously transforming the industry. We created a single new “check” print, which emulated the 1996 print, and it was completed and viewed on Cinetech’s very last day of business, October 10, 2012. This print would serve as the starting point for our subsequent labors.

The primary preservation elements of the earlier project were 35-mm fine-grain master positives, still in perfect condition in the vaults of the PFA and of UCLA, respectively.13 Given the thoroughness of the prior endeavor, no further work was needed to ensure archival preservation of the 1976 version with full medium integrity retained. But what of Conner’s revisions and, particularly, the demands of current exhibition practices? While museum conservators are well aware of the importance of medium integrity, nevertheless (we were not surprised to learn) many of the purchasing institutions had no viable way to properly present 35-mm film.14

The museums acknowledged the ethical need to maintain 35-mm prints, but simultaneously required high-quality digital editions to effectively exhibit the work. This practical imperative intersected with our studies of Crossroads’s production history, giving rise to a strategy that supplemented and ultimately complemented the 1996 archival work. The following is an overview of our efforts, which should provide a snapshot of both the promise and technological limits of the present historical moment.


To complement our photochemical film print, we began our digital work by scanning and archiving the original 35-mm negative to 4K-resolution files.15 Then Silva and I conducted a close review of the differing content in the 1976 photochemical and 2003 standard-definition video versions. We ultimately determined that, though we would fully preserve both versions, our primary exhibition edition would be one that in essence had never existed before: a high-resolution reinterpretation of the final, 2003 low-resolution Conner cut. This would be remastered for appropriate presentation in a variety of contemporary exhibition conditions, including both 35-mm film and digital media. Motion-picture archivists generally refer to this type of “new” version as a reconstruction. In this case, the reconstruction represents the primary public component of what is in reality a multitiered strategy that at once recognizes an archival imperative to respect the artist’s original vision and self-consciously exhibits a subjectivity that is becoming increasingly acknowledged in restoration work.


Early digital restorations were frequently conducted by laboratory technicians who were accustomed to working on new Hollywood films, with the result that many exhibited a degree of polish and, to many critics, sterility that was not typical of historical titles. As our collective understanding of the process has progressed, archives and laboratories alike have become increasingly aware of photochemical image characteristics that should ideally be retained in translation to digital media. Primary among these is film grain, which fundamentally differs from the regularized structure of pixels. Whereas early efforts frequently removed grain as undesirable, newer productions have made efforts to retain it.

As this concept has gained credence, it has become more and more common to remove grain for a variety of reasons, and to later generate digital grain to replace it, following various stages of image processing. However, not all grain is alike. Even when“replacement grain” is specifically modeled on the grain removed from the original source, transformations occur. For Crossroads, it was determined that retention of grain character was a high priority, taking precedence over digital cleanup when conflict between the two arose.17 This decision would set the table for the remainder of our work.


Just as human subjectivity plays a greater role in the restoration process than many care to admit, so too does “digital subjectivity.” When a computer assesses an image to distinguish intended content from incidental damage, a degree of “interpretation” is necessarily built into the computation process. This interpretive act is far from infallible. As one example, dirt-removal software works more successfully on a degrained image, precisely because computers easily mistake grain for dirt. Continuous vertical “tramline” scratches present even greater challenges for software, which can’t easily distinguish them from vertical lines that may happen to appear in the photographic composition. In such cases, digital hand-painting is usually the preferred method, in that it restores human discretion to the process. Unfortunately, hand-painting is extremely expensive.

The question then becomes, In what areas, and to what degree, can a project compromise? For Crossroads, we used automated cleanup just to the point where grain patterns would become noticeably affected were the process to continue; from there on, we engaged in extensive hand-painting. Certain production “problems” were intentionally retained as part of the work.18 The result is a new version that is substantially cleaner than the original copies but still far from pristine. While future tools may enable a more thorough cleanup, the current work embodies an engaged dance between the film’s content and the limits of contemporary technology—as did the original.


A common selling point of digital cinema is that its image is more stable than that of motion-picture film, which, when projected, often exhibits a subtle jitter that the human brain steadies in the act of perception. To my mind, stabilization is overrated. When the jitter is not pronounced, the brain’s internal regulating process compensates for the movement quite well; an overstabilized image, on the other hand, often loses a certain dynamism. In the case of Crossroads, selected shots exhibited a jitter that could be traced back to environmental conditions at the time of filming the atomic explosion and hence demanded retention. Others displayed a jitter that resulted from camera problems, creating effects that we know displeased Conner and could be improved upon. In some cases, unstable images were the result of camera problems that could not be digitally corrected without causing collateral defects, such as softening or blurring of the image. In such cases the stabilization was dialed back, and the images retained both sharpness and the unwanted movement.


Conner’s meticulous production notes expressly state a desire to retain exposure variations within selected shots. Where indicated, these exposure variations were retained. Elsewhere, digital color- and density-correction tools enabled a degree of pulse elimination not easily obtained by means of photochemical film grading.19 In some cases, this process resulted in a flattening of the image that drained it of character, so again the process was dialed back to more faithfully emulate the original—flaws included.


When a work is translated from one medium to another, its image is frequently cropped, often broadly and sometimes subtly. When remastering Crossroads onto standard-definition video in 2003, Conner (in a salient example of subtle cropping) intentionally magnified selected shots to eliminate visible scratches. The 2003 edition thus had different framings than the initial 1976 film release. When it came time to conduct our principal work, we found we could indeed eliminate selected tramline scratches via this method, reducing the need for digital cleanup. We further found that the tightly controlled frame lines of digital cinema allowed us to be more conservative in our cropping than Conner had been in 2003. We thus created a version with framings that matched neither the 1976 nor the 2003 versions but sat somewhere between the two.20 Non-practice-based notions of ethical purity may make such repositionings sound heretical, but they are common practice and are here acknowledged, however slight they may be.


The minor adjustment to framing described above is but part of a larger phenomenon. As I’ve argued before, it is fallacious to think that “perfect emulation” of a static original is the consensus aim of restorationists—or, in fact, even possible.21 The perceptual experience of photochemical film is different from that of digital images when viewed on monitors, and yet again in projection. In our work on Crossroads, we thus utilized available color-correction tools to moderate density and contrast, optimizing the image for presentation in contemporary digital environments while simultaneously retaining its “film” character in terms of grain structure, flicker, and image stability.

While this might worry theorists unversed in technology, such variances are again commonplace, if often misunderstood. Moreover, it is ontologically impossible to replicate one medium in another. In medium translations, interpretation is intrinsically part of the work process, whether one wishes to acknowledge it or not. To presume otherwise and relinquish human intervention in translation is to presume that the technology enabling translation is itself neutral; yet believing so is itself subjective faith. Some would argue that medium translations should not be undertaken, period. While I deeply value medium integrity, I would temper my own vocal support of it (as I have previously22) with the suggestion that although it is essential for some works, it isn’t for others. And while I accept the fact that certain works must be lost to the winds of history, not all works in obsolescent media must necessarily be forfeited. Another type of cultural loss ensues when works that might successfully survive translation are withheld from it, in adherence to a rigid Platonic ideal. Ironically, some works can be if not “restored” then reimagined and reembodied, precisely in their transformation. The challenge in this enterprise is skillful execution.


Upon completion of the digital version of Crossroads, we conducted extensive testing to ensure the highest possible quality in digital-to-film recording and subsequent 35-mm film prints.23 As we had already interpreted the tonal character of the image, it did not make sense to base this work on side-by-side comparisons with older prints, which would by definition yield a mismatch. We conducted such studies only upon completion of the work, for research purposes exclusively. Similarly, we retained our working premise that the new 35-mm prints constitute a different medium than the digital version. Every decision we made was in the service of achieving the most satisfying 35-mm black-and-white film rendition of the digital work possible.24 Upon project completion, we had thus created a total of three different versions of Crossroads: 35-mm analog/photochemical, digital, and 35-mm digital film-out; each with distinct image characteristics acknowledged.


Just as the picture of Crossroads was subjectively rendered, so too was the sound. As mentioned, Conner had not initially released the film in stereo because of practical industrial limitations at the time of completion. As archivists, we were ethically committed to the preservation of the work as released. However, once we had secured that goal,25 we evaluated the source material (as we had done with the picture) to determine the best strategy for our primary exhibition versions. Again we modeled our work on Conner’s final 2003 remastering, and again subtly reenvisioned it for contemporary technology. In so doing, we similarly transgressed the received wisdom and generally accepted parameters of film preservation. By default and intent alike, we adapted the acoustical dimensioning to current exhibition conditions.

Riley’s score, as I have noted, was initially mixed in two-channel stereo. After ensuring archival preservation of the unretouched 1976 mono optical track, we digitally restored the sound. Audio Mechanics studio founder John Polito, in close dialogue with Silva and myself, conformed the original quarter-inch-mix master to the final edit. He then carefully removed system-induced noise generated by rerecording while retaining sound anomalies traceable to the original recordings. The cleaned-up files were archived in both mono and stereo.

The next step was to format the two-channel stereo version for exhibition, a process much less straightforward than one might think. Although 35-mm prints with Dolby Digital sound tracks can accommodate a straight rendition of two-channel sound within its native 5.1 channel configuration, this practice is prohibitively expensive for virtually all but studio releases because of the proprietary fee for its usage. The more viable option for independent artworks, Dolby Stereo, entails spreading two stereo channels to four: left, center, right, and surround.26 For our 35-mm stereo prints we utilized this process by necessity, and the resulting track consequently features a slightly different acoustical dimensioning than either prior version, if only as a result of the system’s default encoding process.

For our digital version, we needed to anticipate a multiplicity of playback environments, encompassing traditional cinematheque, museum, and gallery spaces. Crucially, we had to consider the many variables at play in presenting audiovisual works in the white cube or in semicontrolled black-cube environments within museums. For this key stage of the work, we again did not adhere to strict literalist notions of reproduction. Silva’s direct collaboration with the artist proved vital at this juncture. In joining her intimate knowledge of Conner’s working method with the material evidence of his subsequent revision of Crossroads, we witnessed not a desire to freeze the picture and sound quality in the exhibition technology of one era, but rather a clear intent to move and adapt. We thus decided, in dialogue with both composers, to unobtrusively dimension the film’s two scores within the 5.1 stereo format.

I should stress that we did nothing resembling the effects-based sound spreading sometimes undertaken for commercial films, whereby non-surround audio is expanded to 5.1 to create an illusion that noises are originating in specific places around the theater. Polito, to the contrary, employed different techniques for each of the two scores to distribute the overall sound environmentally.27 While doing the above work, we also ensured that our 5.1 version was 100 percent mono-compatible, meaning that, if played back in a mono system, the output would emulate the mono original and not a corrupted version of it. Not all methods of spreading allow this.


By this point, I hope to have made clear that our reconstruction actively interpreted Crossroads in its rendition of the film. The last aspect of our work that should be mentioned in this regard involved the film’s editorial content. Conner’s 2003 video remastering featured only one notable change from the 1976 original: He readjusted the timing of the film’s end credits. The original release—taking up an idea of Riley’s—featured an extended section of black screen while the music continued after the final image’s fade-out and before the end credits at last commenced in silence. In 2003, Conner truncated the black ever so slightly, so that the credits began shortly before the music’s conclusion and ended in silence. While I know at least one purist who feels that the full black duration makes a crucial difference, both Silva and I agreed that Conner’s 2003 choice, which minimally reduces that expanse of black, does nothing to diminish the metaphysical meditation suggested by Riley’s desired delay, while alleviating the arguably unnecessary durational challenge posed by the credits in isolation. Although this value judgment was one of several reasons we chose the 2003 edition as the model for our primary viewing version, we nevertheless fully preserved both versions in the archives of UCLA.

The issue of duration itself should also be raised in the context of larger problems of gallery exhibition. Crossroads is relatively well suited to museum presentation in that its reception is not completely dependent on sequential viewing, thanks to its repeated depiction of the bomb’s explosion from multiple angles. Ambulatory viewers experiencing only a portion of the work can at least glimpse and sense aspects of the larger ideas behind it. That said, crucial things are missed in such a context. One of Conner’s major structuring devices is his placement of Gleeson’s score—which uses re-creations of representational and quotidian sounds, including explosions, countdowns, and birdsong—prior to Riley’s transcendental, nonrepresentational music. In gallery exhibitions, not only will passing viewers largely miss this dynamic, but continuous looping without interruption makes the piece cyclical—an interesting strategy in its own right, but not Conner’s intent. Looping also creates the probability that some viewers will experience the Gleeson/Riley sequencing in reverse—and have to mentally reconstruct Conner’s design.

Likewise, the film’s last shot, which lingers over the sea for an extended time after the bomb’s blast, cannot be fully experienced via mere glimpses. In this respect, Crossroads is hardly unique: Key aspects of most moving-image works cannot be fully conveyed in the context of itinerant, nonsequential gallery viewing. This troublesome condition is only one of many that point to systemic challenges posed by the traditional nature of museums themselves, which are ill adapted for audiovisual works, even though such works arguably constitute the core expressive form of contemporary life.

Ironically, much of the painstaking effort that has gone into our work on Crossroads could be undone in practice if viewing conditions in the displaying institution were inadequate. To address this, the Conner Family Trust and Michael Kohn Gallery have negotiated detailed agreements with purchasers that consider both archival and presentational issues, assuring at least a degree of control over both exhibition and future remastering, to maintain the work as needed for posterity within the scope of the limited edition being generated.28

In this way, our work on Crossroads is aimed firmly at the present while remembering the past and anticipating the future. By creating and archiving multiple versions across a variety of platforms, I ultimately wish to suggest that no one version—not ours, nor remastering efforts to come, nor even Conner’s own multiple reworkings of the piece—can be understood as singularly definitive. That notion of authenticity is in transformation as we move further into the postmedium epoch. Crossroads now exists as a work out of time that resurfaces within it. Our multitiered iteration strives to be the best possible rendition in its own historical moment, respecting the prior and enabling the next.

The reconstruction of Bruce Conner’s Crossroads is scheduled to premiere at the Museum of Modern Art in New York on October 28, 2013.

Ross Lipman is Senior Film Preservationist at the UCLA Film & Television Archive in Los Angeles.


1. In an e-mail to the author dated July 25, 2013, Michelle Silva, who currently represents the filmmaker’s estate, noted that Conner preferred to render his titles in all capital letters rather than in italics because he “hated letters that looked like they were falling over.”

2. Friend is currently a technical specialist in asset management, film restoration, and digital mastering at Sony Pictures.

3. I briefly discussed the project in an earlier essay, “Problems of Independent Film Preservation,” Journal of Film Preservation 25, no. 53 (1996): 50–51.

4. It is important to note that these terms have slightly different meanings in the museum and film archiving communities. To generalize definitions that will hopefully work for both of these constituencies: Conservation can be understood as the physical maintenance of a prime artifact, preservation as straightforward duplication in a single medium, and restoration as an interactive process whereby the potentially abstract essence of a work is re-created in a duplicate copy from an imperfect or even fragmentary source or sources by a variety of means.

5. Krauss’s formulation distinguishes between modernist notions of medium specificity and “technical support,” or underlying conventions that help define a medium. The term itself is of great value in describing a digital crisis that encompasses the wholesale dematerialization of artworks across multiple media, and Krauss’s distinction is of particular use in the remastering process.

6. The challenge is so pervasive among contemporary art forms that its consideration has even reached the mainstream media. See Melena Ryzik, “When Artworks Crash: Restorers Face Digital Test,” New York Times, June 9, 2013,

7. My use of the term black box, which traces to the theater, is functional rather than specific, in that the term itself is semifrequently conflated with black cube in the discussion of moving-image works presented in a gallery or museum context. I here use black box to distinguish well-controlled spaces (primarily in cinematheques) from uncontrolled or semi-controlled gallery spaces, while ultimately hoping to foster the greater implementation of more fully controlled environments within museum settings.

8. It should be noted that the six editions were supplemented by copies—termed artist’s proofs by the Conner Family Trust—for the trust and UCLA.

9. In considering methodological ethics, I continually refer back to Eileen Bowser’s classic short essay “Some Principles of Film Restoration,” Griffithiana, no. 38/39 (1990): 170–73.

10. Bruce Conner, letter to Russell Wagner, Directorate for Defense Information, March 28, 1974.

11. Via the use of tape delays at differing speeds, Riley expanded two separate Yamaha organ tracks to sixteen channels, then folded these down to a two-channel stereo mix. For his score, Gleeson carefully generated naturalistic sound effects on a Moog synthesizer.

12. Although Dolby Stereo was introduced in 1975, it was not widely adopted by film laboratories until after the release of A Star Is Born in 1976.

13. The Conner Family Trust transferred its film holdings to UCLA in 2008. In addition to AMPAS’s preservation materials, UCLA additionally houses all of Crossroads’s original edited production elements.

14. The larger implications of these cultural conditions are best saved for a later discussion.

15. More specifically, we generated 4096 x 3112 10-bit log DPX files of each frame on an Arriscan and archived them on LTO 5 tape. All digital picture work was conducted at NT Picture & Sound in Los Angeles under the supervision of engineer Shawn Jones, a brilliant technical innovator.

16. See Pip Laurenson, “Authenticity, Change and Loss in the Conservation of Time-Based Media Installations,” Tate Papers, October 2006, It is also the central premise of my essay “The Gray Zone: A Restorationist’s Travel Guide,” Moving Image 9, no. 2 (2009): 1–29.

17. Additional efforts were made to retain the original’s film-grain character in black background titles and white-screen sections, which might otherwise have been replaced by video blacks and pure whites.

18. While one could make a compelling case that all printed-in scratches and blemishes are part of a given found-footage work, our research on Crossroads clearly indicated that Conner liked some of the “problems” present in the film’s source footage but wished others to be corrected.

19. This pulsing is often called density breathing by film restorationists. Video engineers sometimes use the term shading error. I tend to dislike the latter, as its pejorative connotations are misleading when the characteristic is not problematic.

20. Our initial scans encompassed the entirety of the source negative’s image area, ensuring that all photographed content was retained and archived before we embarked on our final reconstruction.

21. Lipman, “The Gray Zone,” 1–29.

22. Ross Lipman, “Technical Aesthetics in the Preservation of Film Art,” in Big As Life: An American History of 8mm Films (New York and San Francisco: Museum of Modern Art/San Francisco Cinematheque, 1998), 92. This essay perhaps best articulates early formulations of my notions of authenticity in moving-image reproduction, later modified and developed more fully in “The Gray Zone.”

23. We conducted our restoration work in 3K resolution (3072 x 2334) to control costs with maximized efficiency, and then boosted the resolution for 4K film-out. Given that the Bikini Atoll source footage from the National Archives was itself a duplicate, the down-res/up-res process involved a truly minimal compromise. As another point of reference, the projected 4K digital cinema resolution of a 1.37 Academy aperture image is 2960 x 2160—less than our working resolution.

24. Our decisions were ultimately based on a rigorous procedure that is best described as LUT-wedging, wherein Shawn Jones at NT Picture & Sound created eighteen custom-modified LUTs for the transfer. LUT stands for lookup table, and broadly refers to the process whereby a digital image is translated from one platform to another, in this case from log film space for digital monitor display to Fuji 35-mm black-and-white digital separation film 4791. Not surprisingly, in such instances a straight application of math will not generally do the trick, especially when dealing with black-and-white stock. In my experience, some photochemical density grading is frequently needed even when the source file is a fully graded program, and contrast problems are common. Jones’s extensive testing yielded one of the best black-and-white film-outs that I’ve seen to date.

25. The original version was preserved to 35-mm film, DVD-R data discs, and LTO tape.

26. Dolby Stereo uses Lt/Rt (Left total/Right total) matrixing to derive the four tracks. Mono sound is matrixed to front center and two-channel stereo to front left and front right, but a slight sound bleed is diverted to the other speakers in both cases.

27. For Riley’s stereo tracks, he employed careful channel separation to isolate discrete center audio and send decorrelated audio to the surround speakers. For Gleeson’s mono score, he used frequency splitting to a similarly controlled purpose, with the express aim of avoiding the creation of specific effects. It should be noted that if one sent identical signals to different speakers, our hearing process would perceptually create its own front/center emphasis depending on where one was seated. The ear uses delays to help determine directionality, and when identical sounds arrive at both ears at once, we tend to center them. Frequency splitting helps create a sense of surround in this context. For this reason, I jokingly refer to the Gleeson section of the film as being “mono surround,” although the usage doesn’t precisely match the term’s technical definition.

28. Detailed exhibition and duplication agreements between museums and artists are becoming increasingly essential as the technologies that enabled the creation and display of specific works become obsolete.