From the Summer 1994 issue of the Operating Cameraman

This is a revised version of a paper presented by the author and Edmund DiGiulio at the 135th SMPTE Technical Conference in Los Angeles, in November 1993. It reexamines the process of converting motion picture film images to television. New specifications are proposed for the area of the film frame to be transferred. With a 16x9 display (either widescreen NTSC or HDTV), incorporating the proposal would allow essentially the same image to be seen on the monitor as that seen in the theater. For present 4x3 displays, more of the film area would be visible, and the picture would match the theatrical image as nearly as the aspect ration difference allows.

We are reaching a turning point. With the arrival of widescreen TV systems (Improved Definition TV, or IDTV), and with HDTV looming on the horizon, we have the opportunity to correct some of the non ideal practices that have become common with present film-to-video transfers. We can start over from scratch. But there's not much time to waste. Images are already being composed, and film transferred, in non-standard ways for eventual widescreen TV use.

The primary area of concern is practices to be followed with 16x9 television systems. However, it may be possible to establish some common guidelines for all transfers. The proposals presented below are intended to achieve the ultimate goal that a video transfer of a feature film recreates the original theatrical experience as closely as possible. At the same time, as much of the available film area as possible should be used in film shot for TV. For these two things to happen, film and television must become more closely integrated than they are at present.

The main problem with film and television compatibility now is the difference in aspect ratios (the ratio of picture width to height). With 16x9 television systems this difference is much less, and the problem is greatly reduced. Of course the problem would be eliminated if the aspect ratios were identical.

Aside from the aspect ratio difference, there are two things that must occur in order to accomplish the goal of having the video release closely match the theatrical showing: more of the film frame must be available in the transmitted video signal, and the receiver overscan amount must be reduced from that typically seen in contemporary receivers. Both of these changes are necessary to optimize film on TV, but it should be emphasized that they are also independent. Transfers could incorporate an increased scanned area at any time, and there would be an improvement in the result available to everyone, whether their TV set was readjusted or not.

The Film Image and Present Transfers

When 35mm motion pictures were first introduced about 100 years ago, every system used a unique format, preventing the possibility of exchanging media between systems. Within a short time, the film stock and sprocket hole dimensions became standardized, as well as the 1 in. by 3/4 in. frame. This standard was accepted internationally and lasted into the 1920's.

When the soundtrack was added to the film alongside the picture, it shook things up again for a short time, but eventually a new aperture was chosen by representatives of the Motion Picture Academy, the American Society of Cinematographers (ASC), the SMPE, and the American Projection Society. This frame was accepted as the international standard for sound films, and has become known as the Academy frame. The original Academy aspect ratio was 4:3 (1.333:1), but has become 1.37:1 due to revisions of the standard over the years.

Early in the development of television, engineers wisely realized that a great amount of television programming would come from film materials. This led to the selection of the 4x3 aspect ratio for the picture tube. Unfortunately, the standard aspect ratio for films changed shortly thereafter, largely as a result of television's popularity. By the early 1950's, broadcasts were going most of the day and into the evening, sales of receivers were picking up, and theater attendance was dropping. Naturally, people in the film industry became concerned about their future. But this changed with the introduction of widescreen movies. Actually, widescreen formats had been experimented with almost since the beginning of motion pictures, but they had not been done very well until the arrival of Cinerama in 1952.

Audiences loved the larger, wider screen. The threat of competition from television merely added more momentum to the theaters' push for widescreen formats. The next three years saw the introduction of three additional widescreen systems: CinemaScope in 1953, VistaVision in 1954 and the 70mm Todd-AO format in 1955. These new widescreen processes drew audiences back into theaters and revitalized the movie business.

Even though the official 'standard' for films was still the Academy format, with its approximate 4x3 aspect ratio, it became common practice for theaters to project the frame larger than usual and to mask the top and bottom of the picture to obtain a wider image. Thus, the cropped 1.85 standard was born. Other countries adopted ratios of 1.85 or 1.66:1, although a 1.75:1 frame was reportedly used by Warner Brothers and Disney for a period of time.

For feature films, the ASC Manual suggests protecting the full Academy frame for eventual video release. This practice may be followed occasionally, but this writer knows of no widescreen features photographed with such protection. Most cinematographers do protect some of the frame above and below the 1.85:1 area, typically 1.75 or 1:66:1 equivalent frames. The fact that the full Academy height is not typically usable makes transfers of features to 4x3 video that much more difficult. Of course, film shot specifically for television may still be composed for the Academy aperture.

Despite the uncertainty about the specifications for the HDTV broadcast format that will be ultimately approved by the FCC, one thing is certain. The new television system (or systems) will use an aspect ratio of 16x9, or 1.78:1. Even though other countries may not agree on the frame rate, or the number of scan lines that make up the picture, television engineers worldwide have all agreed on the 1.78:1 aspect ratio. Being closer to 1.85 simplifies the film transfer process somewhat, but there is still a difference to be dealt with.

It is well known that some cinematographers are unhappy with the 1.78:1 aspect ratio. It is not within the scope of this paper to argue for or against this number. Suffice it to say that the number was chosen as being the most fair to formats ranging from 1.33 to 2.4:1. In other words, equal cropping or matting (in area) is required for main theatrical standards of 1.66 and 1.85:1. It is the opinion of the author that despite the objections that have been raised, the 1.78 aspect ratio for the TV tube is not likely to change at this late date in HDTV development. Of course, it is quite likely that receivers will have sufficient intelligence that the geometry of the raster-scanned area can adapt to various aspect ratios. But obviously, the shape of the tube must be fixed, and it looks like 16x9 will be the shape.

Those who know the history of conflict between the television and film industries might wonder, won't this new widescreen TV hurt theater attendance again, and simply force movies to some other new incompatible format? It's pretty clear that the answer to this question is no. People will always go to the movies, for the high-quality picture and sound, for the social interaction, or just to get out of the house. The film industry has gotten over viewing television as an enemy, and the video transfer is considered a normal part of the post-production process. Many films lose money in the theatrical release but earn a profit from video sales. And let us not forget that a large portion of television program material is shot on film. This includes essentially all dramatic shows. You could almost say that there are not separate film and television industries any longer. They have almost completely merged into a single industry. At the present time, it is reasonable to expect that everyone in this industry should be willing to consider changes that would make film and television more compatible.

This is a good place to make a plea for a global theatrical projection standard. For the last 40 years we have used 1.85:1 for features in the U.S., while much of the rest of the world has used 1.66:1. Since the aspect ratio for advanced TV systems is firmly established at 1.78:1, features and television could be made more compatible by adopting approximately the same frame for film.

Warner Brothers Technical Operations has said that, starting the 1993 fall season, film shot for television by any of their production companies will use a 1.78:1 frame, so as to be widescreen-TV compatible. The next logical step would be to adopt a similar frame as the standard aspect ratio for all non-anamorphic film, including theatrical projection. A 1.78:1 frame is a fair choice for a global standard, because it favors neither 1.85 nor 1.66 users. It would be a compromise for both sides. If this change could be brought about, for the first time since the 1950's we would once again have true global theatrical compatibility, and at the same time, eliminate any framing differences between theatrical and television display. This is simply too good an idea to ignore. This concept is being promoted in Europe by Rune Ericson of Sweden.

The actual difference between 1.85 and 1.78 ratios is slight, only 4%. On a typical 20-ft wide theater screen, the difference amounts to an increase in height of about 5 in. This difference could probably be accommodated by a minor adjustment of the screen masking and would not require a different screen or projection lens. The impact to production is essentially nil. As mentioned earlier, most, if not all cinematographers already protect a portion of the frame above and below the 1.85:1 area. On film and in the viewfinder, the difference in picture height between 1.85 and 1.78 amounts to about .018 in., hardly more than the thickness of the lines on the groundglass.

Adopting 1.78:1 as a new projection standard has been officially proposed to the SMPTE Projection committee. Of course, it is not the function of the SMPTE to create new standards, but rather to recognize practices which have become common in the industry. A change of this type might find acceptance as an EG (engineering guideline) or RP (recommended practice), and eventually become a standard in the future. It is hoped that the proposal will be seriously considered by producers and exhibitors as one that represents an intelligent move toward global standardization and video compatibility, especially when the impact to production and exhibition is so trivial.

Present Transfer Standards

Due to the mechanical tolerances of cameras, printers and projectors, it is not possible to project the entire photographed film frame. Doing so would result in one or more edges of the frame frequently becoming visible on the screen. With any projected format there is always a safety margin. According to ANSI/SMPTE Standard 195, the overhead, or amount of image that exists on 35mm film that is not projected is about 4.5% of the height and width.

The amount of the film frame that gets transferred to video is officially defined in ANSI/SMPTE Standard 96. The area to be covered by the video pickup device is to be 96% of the projectable area, or about 92% of the total height and width of the image. So in a transfer executed to the present specification, at least 8% of the height and width is lost, even when viewed on a video display showing every bit of the video picture. Of course, video displays don't show the entire picture. The typical loss at the receiver is about 10% in height and width, leaving about 86% to be seen at home.

The telecine alignment pattern, including the "safe action" and "safe title" areas, is defined in SMPTE RP 27.3. The SMPTE sells alignment films made to this design. They were originally intended to help align a projector and video camera on a telecine chain, and are still used today for calibrating modern transfer equipment. It is important to realize that there is no inherent connection between the dimensions of film scanned for television and the receiver safe action and safe title areas. The safe areas are defined as a fixed percentage of the transmitted TV picture. In other words, with a video picture originating on film, if we increase the area of film being scanned, the safe areas on film increase a corresponding amount. Conversely, if the scanned area of film is decreased, the safe areas on film also decrease in size.

Some people may wonder why it is desirable to change the scanned area at this time. They may reason that the system has worked for all these years, so why change it? There are two motivations for this effort. The first is the need for a standard for transfers to 16x9 television. To understand why a change is appropriate for 4x3 television as well, more historical background is necessary.

When Standard 96 was created, movies were transferred to video at the local TV station, by means of a tube-based camera and film projector on which regular 16 and 35mm distribution prints and kinescope films were run, unsupervised. Of course this was done live, as there was no other method of storing the picture. The scanned area defined by the standard was purposefully made conservative, taking into account the tolerance buildups associated with the process, and the desire to avoid having any unused areas of the picture tube on the home receiver. The standard intentionally cut off a portion of the image on all four sides to ensure ample leeway for film weave and framing errors. It also may have been felt that, due to the small size of receivers of that era, the long and medium shots that are common in films would be too small for the home viewer to see what was going on in the scene.

There have been substantial improvements in technology since the time the standards were created. Besides an increase in the size of the average receiver, there have been tremendous advances in the design of receivers and broadcast equipment, and there have been improvements to film and the film transfer process as well. Perhaps the most significant change is that the transfer is no longer done at the local TV station, from a regular print. The video transfer is now executed at one of a few post-production laboratories that specializes in that type of work. The modern transfer equipment used is much more stable than the old telecine chain, and the transfer specialist is usually working with a pristine film element such as an interpositive, internegative, or even the original camera negative in some cases.

The net result of these improvements is that the laboratories are ignoring the ANSI standard. They have found that they can increase the scanned area of the film frame without causing any problem whatsoever. This pleases the customer, because less of the image is cropped off. One might be tempted to accuse these labs of violating the standard, until one realizes that the standard was never intended to apply to this modern transfer scenario.

Even though transfers are of much better quality now, a rather unfortunate effect of this situation is that different labs are using a different scanning aperture. A transfer executed at one facility will include a different amount of the frame than a transfer of the same film at another facility. So there is no standard currently in effect. The present approved standard does not reflect the current wants and needs of the industry. A principal objective of this article is to provide a recommendation for transfers that will resolve, or at least improve this situation.

A New Telecine Standard

Now let us examine what can be changed in the transfer process to help accomplish the goal of making the video image more closely match the theatrical image. First of all, we know there is substantial image loss in the typical receiver. The scanning raster is purposely adjusted to cover an area that is larger than the visible portion of the screen, resulting in a noticeable amount of the image being cropped around the edges. This is referred to as "overscan." In a survey of receiver overscan that was recently completed, the least amount of vertical and horizontal loss found in any set was 6% of the total image. Knowing that, by recording a reduced portion of the film image, we are adding insult to injury. With the care and supervision that takes place throughout a modern film transfer, there is no longer any reason not to transmit the entire projected film image. In fact, we can include essentially the entire camera image, and rely on the receiver overscan to hide the excess picture content at the edges. In this way, a receiver adjusted for a small amount of overscan would show the projected film image almost exactly, whereas a set having the typical excess amount would show less of the picture, but still more than what is seen now.

It seems apparent that several standards for transfers of 35mm film to video are required, depending on the nature of the film original and the particular destination video format. The Super 35 format for film is being used more and more, and requires a standard for transfer to 4x3 and 16x9 video systems. On the other hand, we have built up a 60-year history of film material that in many cases does not have any image outside the Academy area. Therefore we should still continue to support that format for transfers, or something close to it.

In the early days of the Academy format, the camera aperture was reduced in size, which left the film unexposed in the soundtrack area, and provided a thick frame line between frames. That reduced aperture is seldom (if ever) used now. Typically, cameras expose the entire (silent) frame. Today, the only difference between a camera set up for Academy versus one set up for Super 35 is that the lens is shifted laterally 0.050 in. for optical centering of the image.

Although film is almost never projected in Academy format today, it is the projectable Academy area that would ideally be visible on a 4x3 television display (assuming an Academy-format original). So if one were to transmit the entire film camera image area, a receiver adjusted for exactly 4.5% overscan would end up displaying the projectable area of film exactly. Efforts are currently underway to persuade receiver manufacturers to aim for a target overscan of 4%. If we assume that every set will have at least a 2% minimum overscan amount, that allows an approximate 2.5% safety margin to accommodate film weave and telecine alignment error.

There is a slight aspect ratio difference between Academy-format film at 1.37:1, and 4x3 TV, at 1.33:1. People are often heard to make the comment that the two are exactly the same. This is a misconception. The old, silent aperture (which is the same as Super 35 today) was indeed 1.33:1. The Academy aperture that was originally selected restored this ratio, but the present aperture is slightly wider. This means that if one were to exactly match the width of the Academy image to the width of the TV picture, the height would not exactly match. Conversely, if one matches the height of the film image to the height of the TV picture, the width will not exactly match. So the choice must be made whether to match the height or the width (or perhaps neither, as is the case at present). Of course, in a transfer to 16x9 HDTV, one would match the width, because there would be unused image height on the film anyway, and there is no need to retain NTSC compatibility.

On the other hand, 16x9 widescreen-NTSC, or IDTV sets, share the same video signal as 4x3 NTSC sets. The IDTV set simply displays a selected number of lines out of the middle of the normal NTSC picture. While there could be separate home video versions for 4x3 NTSC and 16x9 IDTV sets, this is undesirable and should not be encouraged. For television broadcasts, there is no choice. They must share the same signal.

Therefore, since a modern film transfer to NTSC video may end up being displayed on both 4x3 NTSC and widescreen IDTV sets, the most useful thing would be to establish a common scanned width that includes the entire film image width. This would make the area of the frame scanned for TV .864" x .648". Unfortunately, the height of .648" is quite a bit more than the Academy image height. And there are often undesirable objects at the upper and lower limits of the image, such as mic booms at the top, and dolly track at the bottom. Until the time comes when 4x3 receivers are obsolete, a safer practice would be to scan just to the Academy height. This gives a scanned area of .842" x .631", which results in a very slight image loss at the sides, if the video display has more than 2.5% width overscan. A compromise may be feasible, but the latter method has been proposed to the SMPTE working group on telecine practices. With such a transfer, viewed on a 16x9 display adjusted for 2% overscan, one would see essentially the same image seen in the theater (assuming that the proposal of using 1.78:1 for theatrical display was also adopted).

We can now contemplate a telecine alignment chart that embodies the philosophy outlined. The safe action and safe title reference lines would be absent. As already discussed, those things are a fixed percentage of the television picture. If the operator enlarges or reduces the frame slightly, those references on the film become meaningless. Also, they are likely to be revised if TV set manufacturers comply with the request for a reduced overscan. The only important references that should be present are: the actual border of the area to record, which is approximately the same as the Academy image border, and an indication of where the screen will cut off on an ideal display, which is basically the same as the projectable area.

Rather than having separate lines that indicate these two perimeters, a single bold border line is used. The width of this line is significant. The line thickness is 2% of total image width at the sides, and 2% of image height at the top and bottom. The 2% thickness on all sides results in a combined 4% difference, corresponding to ideal receiver overscan. The outer edge indicates the total area of the film frame to be transmitted, and the inner edge indicates the projectable/viewable area.

The single bold reference line simplifies the alignment pattern to the extent that multiple versions of it can be included on one test film. On the "Super 35" film, the full 4x3 aperture, a 16x9 shape of the same width but reduced height, and a smaller 4x3 box comprising an extraction from the center of a 16x9 composition would be represented. Thus one test film would replace three.

Another version of the film would include the same three references, but slightly reduced in size and re-centered for the Academy frame. In the usual case where the full Academy height has not been protected during photography, the preferred 4x3 transfer would use the 16x9 indicator as a letterbox reference. The full image width would be preserved, and a matte would be added at the top and bottom to cover up the mic booms, lights, or other distractions that appear in the 4x3 picture. As a less-desirable alternative, the inner rectangle could be used as a pan-and-scan size reference.

The Warner Brothers plan for using a 16x9 film frame during photography for eventual HDTV use was mentioned. The planned 4x3 TV image is an extraction from the center of the 16x9 film image. The extra image width that will appear on an HDTV set cannot appear on widescreen IDTV sets unless the picture is letterboxed in the NTSC domain, because of the need to share the signal with standard NTSC sets.

Besides being used on alignment films, this same test image, or portions of it, could be helpful to cinematographers if it was applied to the groundglass in the camera viewfinder.

Conclusion

This is a time of change and merging technologies in the audio-visual production community. It is no longer fashionable or desirable to enhance the differences between photo-chemical and photo-conductive capture or display methods. Both technologies have their strengths and weaknesses, and it is time to concentrate less on emphasizing the differences, and concentrate more on reconciling them. Establishing a common screen shape is one step toward maximizing compatibility, and minimizing the impact to compositions in transfers either to or from film. Also, both technologies have an effective "gross image" and a "net usable image", which are entirely equivalent. In film, they are defined as Camera Aperture and Projectable Aperture. In video, it is scanned/transmitted area and displayed area (due to receiver overscan). If the safety margins are adjusted to coincide, rather than adding to one another as they do currently, the loop is complete, bi-directional transfers with not image loss are possible, and the method of delivery to the viewer becomes transparent, as it should be.

Acknowledgments

The author would like to thank Ed DiGiulio for his collaboration and guidance throughout this work, and the other SMPTE members and specialists who assisted: Si Becker, Bob Ringer, Rollie Zavada, Phil Mendelson, Mike Orton, Chris Cookson and Paul Klamer.