Heahter Weaver
It’s not video

The "notes" section of this Powerpoint document contains the preparatory guide to a presentation given at Digital Heritage: Symposium on Video Art in Germany from 1963 to the Present held in Dusseldorf, Germany in July, 2005. The presentation described the processes and methods used in restoring the 1975 video art piece, The Eternal Frame, citing specific technical issues encountered.

The Bay Area Video Coalition (BAVC) is a non-profit multimedia arts access and training center located in San Francisco, California, U.S.A. BAVC offers a full range of audio and video post production services including obsolete analog tape preservation and restoration. Learn more about BAVC at www.bavc.org.

In 2003 BAVC produced the educational DVD, Playback: Preserving Analog Videotape to demystify the process and issues surrounding video preservation. When the DVD project was first envisioned artists Chip Lord and Doug Hall of the 1970's video collective, Ant Farm, agreed to participate in a case study as they worked with BAVC to preserve the 28 source tapes, primarily half inch open reel, that were used in the making of The Eternal Frame and to reconstruct the piece from the newly remastered source tapes.

Since the original open reel master of The Eternal Frame had already been through the preservation process several times for archiving and distribution prior to 2003, in instances where the original source material could not be found, footage was taken from a previous transfer of the half inch open reel edited master.

At this time in history preserving a work of video art is an act of migration. Program content is transferred from an obsolete format onto a modern format. Restoration is the process of compensating for loss and damage which may have occurred to audio and video after the work was originally created.

Neither of these processes require accessing the original source tapes that were used in the making of an original edited piece of video art. The original source tapes were accessed to perform a reconstruction because the source tapes were available and previous copies of the edited master that I reviewed were far from identical in terms of areas of degradation, contrast , and overall signal quality. It was my hope that, having been handled less, the source tapes would likely be in better condition than the open reel edited master.

Several transfers of the edited master made prior to 2003 were carefully reviewed and compared along with a transfer made in 2003of the open reel edited master to Digital Betacam. The 2003 transfer of the open reel edited master contained far more dropout than the previous transfers, indicating that the original open reel master had continued to degrade over time. An earlier transfer of the open reel master to 1inch type C displayed a noisier signal with less resolution. Other transfers displayed varying degrees of difference in video levels, hue, degradation, amount of headswitch present and vertical blanking width.

Overall, the best transfer of the open reel master was a transfer made in 1999 to Digital Betacam. When source material was not available, most replacement shots were taken from this transfer of the edited master.

Although it was assumed that each copy of The Eternal Frame that I reviewed was made from the original halfinch open reel master, it was not clear that this was the case. Each copy I reviewed was far from identical. The visual intentions of the artists became vague. It appeared that it would be beneficial to have a first generation master on a modern format, color corrected to the artist's specifications that could be reproduced correctly using modern professional standards.

Each half inch open reel source tape was cleaned using a Recortec tape cleaning machine. The Recortec cleaning machine was designed for BAVC in 1994 and was based on cleaning machines used for data tape in the 1970's. The tape to be cleaned is loaded onto the machine and travels through pellon cloth, over a series of slotted grids and through a vaccuum chamber. The tape is cleaned as it travels to the take-up reel and once again as it travels back to its original spool. The 3/4 inch sources were cleaned using an RTI VT-3100 cleaning machine. RTI manufacturers tape cleaning machines for various formats.

After the tapes were cleaned the video and audio signal was transferred from the obsolete tape, through a time base corrector and onto betacam SP tape. The decision to transfer the source material to betacam SP was a financial decision. It would have been far more costly to transfer the 28 source tapes to digital betacam.

I felt that this was a reasonable choice, as I knew that during the reconstruction the source tapes would be played back using a Sony DVW-A500 digital betacam tape player that immediately transcodes the component analog signal on the betacam sp tape into a component serial digital signal before that signal is routed through any processing equipment or cable.

The time base corrector is used to stabilize the image by replacing the synchronizing signal from the original tape with a stable and strong synchronizing signal for recording onto the new tape.

Tapes holding information to be preserved are reviewed in multiple areas to determine where the video levels should be set for the transfer. Reviewing the tapes in several different areas is necessary because many tapes that are preserved do not contain a bar and tone test signal. When the test signal is present, it sometimes has no relationship with program content.

The black level is set using the darkest area or scene on the tape while the white or video level is set using the brightest area on the tape. This stage of the transfer process is critical. If a video signal is too bright all levels over a certain brightness level will be clipped to white. If portions of the signal are too dark, the levels under a certain level will be clipped to black on the new copy of the work. Once this detail is lost it cannot be recovered. This also holds true when transferring video into a nonlinear editing system or other hard disk based system.

The exposure of the open reel source tapes used in the making of The Eternal Frame varied widely. Upon reconstruction, I noticed that some areas of a particular source tape would be too dark while other areas would be too bright, resulting in a loss of detail in those areas. These source tapes had to be retransferred, optimizing video level for each section being used. When The Eternal Frame was initially edited, the editor was able to make video level adjustments and would have done so for each shot. This was evident during reconstruction as there would have been no other way to preserve the detail that existed on the edited master.

When material is in color, hue and saturation levels can be determined using skin tones but in cases where bars on not present or are suspect, it is generally safest to leave hue and saturation at their default settings.

At BAVC, our policy is to review the signal from the obsolete tape both directly and through a time base corrector. This is done because sometimes a time base corrector can introduce unwanted artifacts to a video signal. We have several different makes and models of time base correctors. The time base corrector that produces the most stable image is used for each transfer. In reviewing the source tapes transferred for the reconstruction, I encountered several specific problems with the transfers caused by the time base corrector.

In one case the time base corrector was not set properly and was not replacing the old, unstable synchronizing pulse from the original source tape. This caused the image on the tape to be lower in the frame that it should have been , produced dark bands across the image and the image occasionally drifted off of the screen. When running the same footage through a different type of time base corrector, the unstable sync produced vertical jitter or jumps in the image. Adjustments were made to the time base corrector and the tape was able to be played properly

Headswitch is a phenomena of helical scan tape formats. It occurs just prior to the vertical or field synchronizing information within a video signal. While headswitch can interfere with the synchronizing information of the video signal, often its only manifestation is the appearance of unsightly lines at the bottom edge of the video image. This aberration is usually not long enough in duration to interfere with lines of video that make up the active picture area. It often can not be seen without the use of a professional monitor set to the overscan position and therefore is usually not apparent when a work of video art is viewed in a television broadcast or in a gallery.

In the transfer process, other issues can arise from the way in which the obsolete tape was originally recorded.

When the tape to video head contact or video head speed is inconsistent during a color recording, velocity error is often the result. In this case, Velocity error manifested itself as horizontal red and green stripes covering an image. It is most commonly found in 2" recordings but any helical scan format is subject to this type of aberration.

While velocity error can sometimes be diminished by playing a tape back in a different machine the velocity error we found was recorded into the 3/4 inch source tape and could not be adjusted. Skew error appears as bending in the top part of the video frame and is a result of improper tape tension as it is being pulled over the record or playback heads on the video drum.

After the tapes were successfully transferred to betacam SP, each tape was logged and the segments used in The Eternal Frame were noted. An Avid 1000 system was used to create a frame accurate edit decision list for the reconstruction.

First an edited master of The Eternal Frame was digitized into the avid. The necessary portions of footage from the newly remastered sources was then digitized. In the avid the source footage was aligned over the edited master for each cut, guaranteeing an exact match. An electronic file of all edits was generated from the avid which could be imported into the linear online suite for reconstruction.

The edit decision list was loaded into the linear digital component online editing suite to build a color corrected master reconstructed from the original source tapes.

The linear suite was chosen because, when editing from tape to tape in a linear suite, there is no compression, footage can be color corrected in real time, and multiple sources can easily be run simultaneously. Having the ability to run multiple sources together at one time allowed for thorough comparisons of the newly transferred original source footage and corresponding shots on several different pre-existing transfers of the edited master to ensure that the best overall source was chosen for each cut.

Chip Lord and Doug Hall were present during the reconstruction to guide the color correction.

The half inch open reel sources were originally black and white sources. In some cases, the time base corrector had introduced a slight color cast to the image.

All color was subtracted when appropriate in the color correction process and video levels were adjusted to create a master with rich blacks and ample contrast while staying within the technical specifications of the medium.

In most cases the newly remastered original source material was clearly superior to the corresponding footage on any of the edited masters for each individual cut. Occassionally, only a portion of the original source footage used in a cut had degraded to the point where that portion of the cut was superior within the edited master. In these instances the source footage was aligned with the edited master and a cut was added to conceal the degradation. The cut becomes a match frame cut; a seamless cut to and from identical footage.

As we moved through the piece artists Chip Lord and Doug Hall began to notice that I had far more control of the image than they had when The Eternal Frame was first edited in 1975. Rather than just setting overall black, white and contrast levels for each cut, Chip and Doug began to wonder what else was possible in order to enhance the appearance of their video.

Pip Lauenson, a conservator from the Tate Museum in London, England, who was also present during the reconstruction, noted that in some instances the suggested improvements may result in the creation of a new piece of art rather than a reconstruction.

The color correction stage of reconstruction was the first phase of what is considered restoration. Opinions differ widely on what should and should not be done when restoring or merely preserving a work of video art.

This page of the presentation contains live video excerpted from the educational DVD, Playback: Preserving Analog Videotape
show video

Chip’s idea was to take the single long cut where there were camera aperture shifts and turn that single cut into two cuts, color correcting the underexposed segment and the overexposed segment so that they more closely matched and then to add a cross dissolve between the cuts to counteract the way the scene was originally recorded by the camera. This is a color correction technique that I use frequently when online editing new work.

To perform the modification that Chip was suggesting would have required dubbing the source tape with the aperture shift onto another tape, synchronizing them perfectly and having an edit controller perform a dissolve from one source to the other at the exact time to counteract the aperture shift. Considering that the edit controller would have had to be capable of controlling 3 different machines at one time and since half inch open reel tapes do not have timecode, this would have been very difficult if not impossible to accomplish. Pip recognized this immediately and suggested that if we undermined the original technical limitations of the time we would be creating a new work. To quote Pip "You could remake the work and make a different piece with your footage." Once the boundaries were set on what we felt should and should not be altered in the restoration process with the guidance of Pip, the color correction stage of the restoration process went smoothly.

In summary, the boundaries of the color correction and reconstruction sessions were defined as follows:

We will not undermine the technological limitations of the time the work was created. We will not zoom in or reposition the image in the frame to obscure the effects of scew error or headswitch. We will ajdust the video levels for each shot, optimizing white and black levels, saturation and hue. We will counteract any artifacts of our reconstruction process.

As Rob Riley pointed out in the segment from the educational DVD, Playback: Preserving Analog Videotape, sometimes "the degeneration of the image is considered to be part of the "pulse" or feeling of the work.” I spoke to Chip and Doug at length on this issue prior to the commencement of our restoration sessions. While Chip and Doug certainly did not want The Eternal Frame to look like it had been shot and edited recently, they were in no way attached to any degradation that appeared due to tape deterioration. In fact, some of the original editing decisions were heavily influenced by their desire to avoid areas where degradation, such as dropout, was already present on their source tapes in 1975. From this discussion it was decided that we would repair any defects in the footage that were a result of deterioration or degradation caused over time or in the transfer process from open reel and 3/4 inch to betacam sp

The most prevalent artifact of deterioration found was dropout. Dropout occurs when the surface of the tape is scratched, resulting in a loss of magnetic particles. Dropouts look like recorded in white or black horizontal lines of varying lengths on the screen. They are generally random in nature.

A curious form of dropout also appeared in the newly transferred footage. Unlike the usual random white or black recorded in lines, this dropout creates a white-black-white pixel pattern that travels from the bottom right edge of the frame to the top left edge of the frame over a period of several seconds and then begins at the bottom again. This type of dropout can be caused by a small yet long horizontal scratch along the surface of the entire tape.

Aberrations can also be caused by a magnetic rather than a physical tape scratch.

Oxide material lodged in a guide or in the transport of a play back machine can magnetically change the characteristics of the tape passing by.

Glitches are defined as miscellaneous non-repetitive impulsive noise artifacts. They usually appear to be filled with something other than white or black.

Sometimes glitches are the result of a drop out compensator in a time base corrector attempting to repair a physical dropout on a tape by filling the white or black area with a previous portion of the frame held in memory.

Infrequent analog dropout or other non-repetitive artifacts can be efficiently repaired in an appropriately equipped digital linear suite but given the amount of degradation found in The Eternal Frame we decided the bulk of image repairs were made using an SGI hosted Discreet Logic Smoke system. The Smoke system was chosen because it contains a formidable toolset for making image repairs, allowing the user to combine multiple image enhancement techniques on a series of single frames while maintaining the ability to play back results in real time without the need to render.

Footage was digitized into the Smoke's hard drives uncompressed. It was then deinterlaced so only fields containing artifacts were altered.

Smoke's reveal tool was used to reveal portions of an unaffected field in cases where the subject and camera remained in the same position over a series of frames. The clone tool was used to cover affected areas of a field with similar areas of the same field. In some cases, aberrations were painted out using colors picked from areas surrounding the damaged area.

These methods and more, including the use of desktop tools such as Final Cut Pro and Adobe Photohop in image repair, are outlined in an article entitled Restoring the Eternal Frame published in the July 2003 issue of DV Magazine. It can be accessed online at www.bavc.org/meet/news/articles/restoring.htm

After over 300 hours of image repair, we emerged with a first generation master of The Eternal Frame on digital betacam, each shot being prepared to the artists' specifications. Even using the Smoke system, evidence of degradation, such as dropouts, had to be fixed by hand, field by field. This is a time consuming and labor intensive process.

In response to the film industry's migration towards a digital environment for post-production, software has been developed to automatically detect and repair common defects in film such as dirt, scratches and debris, jitter and weave. Many of these problems looked similar to video aberrations I have encountered.

Curious to discover what would happen if deteriorated video was processed by an automated film repair system, I approached Gary Adams, product manager of Revival, a film restoration software package by da vinci systems, to see if he would be willing to collaborate on a series of tests using native archival video.

Although Revival software was not designed to work with interlaced video and the software was written specifically for film, da vinci agreed to participate. The results we were able to achieve were promising. Revival software was able to catch and repair many dropouts, was able to counteract jitter, and was even able to greatly improve a segment of video containing distortion and multiple glitches.

The Revival system was also able to repair, nearly seamlessly, a half inch open reel in camera edit. An in camera edit on any analog format has a characteristic look and can be found in many early works of video art. I would argue that this is precisely the type of artifact that becomes part of a work and should not be undermined.

Many software based film repair systems, including the da vinci Revival, has a feature where all automated repairs the system makes can be identified, reviewed and retained or deleted by the operator so unwanted repairs can be avoided with human guidance.

Fortunately when working with video, it is not necessary to alter the original work. A separate lossless copy should be made for restoration. It is always possible to reverse any changes made in a restoration session by replacing the segment of video with unwanted changes with the original, unaffected material. As the field of video preservation progresses, and automated image repair systems become more prevalent, it is necessary to remember, as Riley states, that the history of the machine used to create a work of media art is embedded in that work.

Different technologies leave different signatures on media artworks, creating unique challenges for media art conservators. As technology continues to change at a rapid pace, it is important to retain an understanding of outmoded technologies used to create media art and to recognize the signatures that are an integral part of the work that anchor it in the time period and in the way it was produced.

Special Thanks to:
Gary Adams, da vinci
Pip Laurenson, Tate Museum
Ken Zin, Merlin Engineering
Susan XXX, BAVC
Janet Patterson, BAVC