March 1998 - Also see the updated article "The Fall of Film Production"

We are crossing a dividing line in the technical business of television and post production. The line is a sharp separation between what we've known as television and what will become of it. Producers ask about the future value of production material as they realize the images they create are mortal in the face of technological change. The question of how to prepare production material for future generations is an excellent one which strikes at the very heart of what the Digital Television future is all about.

Phrases like "future-proofing" come into being, indicating an uncertain outcome of today's efforts and the desire to use the produced material in a future application. Produced material is correctly considered a valuable asset that must be protected. Super16mm and 35mm film formats as well as High Definition Television images come to mind as being future-proof. The produced material needs to be used in today's broadcast and distribution channels as well as the Digital Television standard in the near future and any future application that may come along. The producer must decide on the potential future value of the material before proceeding on a production path.

The answer to future-proofing material is not a simple one. We will discuss the different broadcast video standards and explain some of the issues involved with each. We will address some current assumptions about film and video quality and will cover how they relate to the coming Digital Television standards. We will also discuss how the production standards chosen affect the immediate delivery and long term shelf life of the product.

The DTV (Digital Television) and ATV (Advanced Television) terms are commonly used interchangeably, but they are distinctly separate items. We'll decipher how video and film formats relate to future-proofing but first lets talk about the DTV standards.

Digital Television:
Digital Television (DTV) was originally scheduled to replace all existing terrestrial analog NTSC television transmissions in the U.S. by the end of 2006 but this date has been pushed back to February 17th, 2009. This doesn't necessarily affect home video formats, direct satellite transmission or cable television but the range of services and potential improvement in image quality will probably drive those industries as well. Several simultaneous Standard Definition Television (SDTV) image streams or a single High Definition Television (HDTV) image will make up the television programming broadcasts. SDTV is considered roughly the same quality level as today's television broadcasts and HDTV relates to a number of higher definition video standards. In any case, a television image in SDTV or HDTV will be transmitted in 16:9 aspect ratio. Both of these broad television formats are considered to be "ATV".

Advanced Television Standards:
In fact, there are 18 different television standards that may be broadcast under the name "Advanced Television". This may seem like a lot of different standards but the ability to tailor a digital signal to a task specific function could lead to many more "standards". The ATSC has constrained the list of possibilities to only 18.

You may be able to count more or less of them depending on how deep you get on permutations (and we will resist describing all of them) but it seems there will be only a few standards in general use. The standards are named with the number of scan lines and one of two scanning types; interlaced or progressive.

A "480i" standard means the television screen contains 480 usable scan lines with interlaced scanning, roughly the equivalent of our current NTSC broadcast standard. Horizontally, there are 704 active picture elements (pixels) on each line for 16:9 images. The "1080i" standard has 1080 displayable lines and 1920 pixels across the screen. The "i" with these numbers stands for "interlace" which describes a television frame that is broken into two "fields", transmitted sequentially and reassembled as a complete frame at the home receiver. This is the principle of current NTSC television and will be continued into the DTV world.

The antonym of interlace is "progressive" where the entire frame is transmitted as one element. Using progressive scanning dramatically increases the apparent resolution of an image but has other penalties in bandwidth requirements and receiver manufacturing costs. There are heated arguments over which scanning format to choose for broadcast. Each network and service provider faced with this decision believe they have the right answer. As conventional wisdom changes like the wind, other scanning formats will rise and fall in popularity as technology progresses. Fortunately, the receiver manufacturers belonging to the Consumer Electronics Manufacturer's Association (CEMA) will build DTV receivers that will decode and display all 18 broadcast standards.

Digital Television Services:
The DTV transmission is a digital broadcast service that is not necessarily an exclusive television programming channel as we know it. A single DTV channel may include a variety of data services sharing the channel space. The broadcaster's selection of a pixel count and scan type affects the picture quality reaching the home and the amount of broadcast real estate needed to get it there. They have the ability to sell data services over the same channel shared by television images. The issue of picture quality boils down to the digital data rates reserved for the television image.

This thinking is certainly on the minds of many broadcasters as they work out the financial models in their DTV future. It is possible to "bit starve" the television image in favor of data payload on the DTV channel thus trading image quality to make room for other paying services. It is also possible to increase the quality of the television image beyond the intended "Table 3" constraints. By using some proposed data tricks, one network has spoken of broadcasting sporting events at 90 frames per second at HDTV resolution. Time and funding will tell if that noble effort will succeed.

The earliest over the air DTV broadcasts will simply be standard definition television connected to a DTV encoder carrying existing programming. These broadcasts will be the "480i" variety. Broadcasters will gradually begin integrating a library of programming intended for future DTV transmission. First, 16:9 aspect programming with standard resolution is the easiest thing to accomplish. In the future, higher resolution images will become more commonplace as the older programs and production equipment are retired.

The production standard used is not necessarily the same as the broadcast standard. Of primary concern to producers is the quality of the original material and it's future value. Broadcasters will be converting images from their native production format to fit into their broadcast chain. Regardless of the original image quality (pixel count), the common denominator in all produced material will be the image aspect ratio.

Aspect Ratio:
The current NTSC broadcasts are in 4:3 aspect ratio. This means that no matter what the screen size is, the image will measure 4 units wide and 3 units tall. The primary feature of the ATV formats is a 16:9 picture aspect ratio, which comes out to be about 20% wider than a 4:3 image of equal height. Think of a 4:3 aspect ratio as 12:9 when comparing it to 16:9. Independent of the aspect ratio is the number of scan lines available on the screen and the number of pixels available across the width of the screen. The higher the line and pixel count, the better the potential resolution of the image.

One of the available realities in the ATV world is the need to incorporate images from current tape libraries. The largest change for ATV and biggest hurdle to using existing material is the issue of image aspect ratio. Current video libraries are all 4:3 aspect ratio and must be converted to fit in a 16:9 world, whether it is HDTV or SDTV. Essentially, all available 4:3 aspect program material has become obsolete. The producer must decide to either blow up the picture so the original image sides fill the screen, or allow black side panels on the 16:9 screen thus keeping the original aspect ratio of the source image.

The penalty for blowing up the picture is that the top and/or bottom of the screen will be removed creating a framing problem. Things normally in the frame may get cut off, or a medium shot of a person's face becomes a close-up, each changing the meaning of the image. In addition to the framing problems, a blowup from a video original degrades the image quality with visible artifacts. The producer must make compromises when reframing each scene during the blowup process.

A producer with film elements available, especially widescreen film, will have the advantage of re-transferring the image elements and reassembling an ATV compatible product, possibly reusing the entire audio track. Film shot in 4:3 ratio will present the same difficulty while deciding where to reframe the image, but degradations caused by refaming are quite minimal when done at the telecine transfer step compared to a similar action using video as the source. Standard definition video material finished in 16:9 format may be applied directly as an SDTV product.

Major manufacturers of professional video camera equipment such as Sony, Panasonic, Ikegami, Philips and others offer standard resolution NTSC cameras capable of switching between the current 4:3 aspect and the 16:9 widescreen ATV aspect. These cameras will allow producers to create video images in the correct aspect ratio for ATV product, making it easier to reversion video originated material for future broadcast. The DTV standard does not define the image resolution required for broadcast of an ATV image allowing both standard and high resolution images. The producer should consider the alternatives presented with the various film and video formats when thinking of immediate, short term program delivery and future-proofing program material.

Scanning Systems:
The number of scanning lines available on the video picture become the limiting factor for vertical resolution. More scan lines in the television system generally translate to higher vertical resolution. The issue of interlaced scan versus progressive scan also comes into play when judging picture quality. A progressive scan picture with only 720 scan lines ("720p") has nearly the same apparent vertical resolution as 1080 lines with interlaced scanning ("1080i"). The interlaced scan method is a form of compression that degrades the picture slightly.

The current NTSC analog television scanning system is nearly identical to the 480i ATV standard. With the same number of scan lines delivered to the home as 480p (progressive), the home viewer will perceive a much higher resolution image. If television programming is created in a progressive scan standard and delivered to the home in that manner, many of the artifacts attributed to interlace will disappear.

The expense of manufacturing a large tube-type progressive scan display system is high compared to interlaced displays. It is more likely that the home receiver will have an interlaced display and the progressive scan material will be converted to interlace at the home receiver. Film is well suited to a progressive scan delivery system. Hopefully, the technical and economic hurdles will be overcome so we may actually see it in the home.

Large screen flat panel displays are coming to market that will allow a progressive scan image to be displayed correctly. An image that was created as an interlaced product will carry the artifacts of interlacing to any progressive scan display. You can successfully make an interlaced image from a progressive image but the reverse is not true.

Image Quality Considerations:
Video cameras have gotten very good in the areas of resolution, dynamic range, sensitivity and noise. Film stocks have steadily improved over time as well. We must consider these areas when talking about picture quality in any format.

Image Resolution:
The subject of image resolution, or sharpness, will be the real key to future-proofing. Please forgive me as I tech-out for a moment here. The measurement of horizontal resolution in an image is the maximum number of black and white vertical bars that can be visually resolved within the horizontal dimension equal to the picture height. In other words, no matter what the picture size or aspect ratio is, you carve out a square on the screen (where width equals height) and count how many black and white vertical bars you can cram into that area and still see them. This is true for film or video and is expressed as "TVL/PH", or "TV Lines per Picture Height". The vertical bars are considered vertical "lines" which are not to be confused with the fixed number of active scan lines available on the television screen.

The resolution measurement for a camera involves shooting a test chart with a series of patches containing measured vertical black and white bars of different packing densities. To measure resolution of a video camera, a video waveform monitor will directly display the ability to resolve each vertical line in the patches. For film, a microdensitometer, essentially a microscope with a light meter, is used to examine the image of the black and white bar patches and determine how well the film can separate them. With each test patch that has bars closer together, the cameras have a harder time resolving the individual bars and tend to progressively blur them together until they turn a flat gray at the extreme upper limit of resolving power.

Measuring how much the black and white bars blend together is expressed as a percentage of what they were originally, namely 100% black and 100% white. A 100% response indicates that nothing was lost in the camera. It's possible to have a measurement of over 100% after gamma and aperture correction, but we'll discard that discussion for now. An 80% response on a higher resolution patch is considered very good, showing only mild degradation. Once you get a high enough packing density of black and white bars and the residual falls into the 20% range, you can start to write off the existence of any significant resolution elements.

A test like this will show that Super16mm film can resolve fewer vertical lines than some current standard resolution video cameras. A present day NTSC video camera can resolve upwards of 750 vertical lines whereas Super16mm film has lost half of its resolution powers at around 500 lines. These numbers represent what is available in the camera and does not take into account what happens to the signal when processed further in a video system.

Once either of these images are converted to a digital video recording at 4:3 (standard television) aspect ratio, the resolution is limited to 567 TVL/PH on a D2 machine and 535 TVL/PH on a D1 or Digital Betacam machine. The limits occur due to the available pixel count per line of the digital television system in use.

If a 4:3 video image is stretched horizontally about 20% to a 16:9 aspect ratio, whether film or video originated, the horizontal resolution of a D1 or Digital Betacam image is reduced to 402 TVL/PH. There will be fewer pixels available inside your square resolution test area because they've been pulled horizontally to make the screen wider. Even so, the video camera, which started with more resolution, has a measurable sharpness advantage over Super16mm film. Based on this, a high quality standard definition video camera will have a measurable resolution advantage over Super16mm film in the DTV world.

Noise:
Kodak has converted the measurement of film granularity to the equivalent of video noise. They calculated that Kodak EXR5254 film in a Super35mm format, a size used for 16:9 production, has a 50db signal to noise ratio. Signal to noise in television is a measurement of how much the picture content overpowers background noise. A number of 50db means that the noise or grain pattern is .01% of the picture content. Every increment of 10db is a multiplication factor of 10, so a 60db ratio is one-tenth the noise of 50db and 40db is ten times the noise of 50db. A higher number is better. The Sony HDC-500 HDTV video camera measures at a 54db signal to noise ratio, slightly better than the Super35mm film stock. Comparing that to Super16mm with only 42db and 16mm at 40db, the Super16mm and 16mm film doesn't compare favorably. By these tests, Super16mm film has more than ten times the noise of a present day HDTV camera.

Dynamic Range:
Film is acknowledged to have a minimum dynamic range of about 8 or 9 stops. That is the lighting difference between the brightest and darkest object in a scene without overexposing the image and without losing detail to noise or film grain. Jeff Cree, Sony's guru on video cameras, demonstrated how a Sony DVW-700 video camera can make a remarkable picture on a table-top scene with 11 stops difference between lightest and darkest objects. A properly exposed video camera, without any clipped elements in the picture, can reasonably be expected to approach the exposure quality of a film originated image.

Sensitivity:
Film is no doubt the most flexible format for working in varying lighting conditions. The exposure index of a video camera cannot be adjusted like a film camera and extremely sensitive film stocks can make excellent exposures with candle light. There is no such thing as "fast" video tape. However, some video cameras have signal to noise ratios in the 60 to 65db area which allow for additional video "gain" to be added without dragging up the noise in the blacks. With these cameras, reasonably good pictures can be made in extremely low light situations that will rival most standard film stocks.

Standard Definition Television:
The current component digital standard, considered the top of the heap today, is ironically the lowest acceptable image quality in the realm of ATV. An official CBS Engineering document written by Henry Mahler concluded that the lowest quality image available in our current television standard is a component digital recording at 16:9 aspect ratio. It was rated lower than even composite digital (D2) images in his report. The 16:9 SDTV images we can make today will match the quality of SDTV transmissions on a DTV channel and can be included in a product intended for HDTV distribution if necessary.

High Definition Television:
The term "High Definition Television" is considered anything that is better than what we get today. Any scan line count greater than 480 is generally considered "High Definition". Even 480 lines transmitted as progressive scan is considered a "High Definition" image. The top of the heap would be the 1080 line HDTV standard which several broadcasters have elected to support.

The 1080 HDTV standard will point out some of the inherent shortcomings of Super16mm film. Joe Flaherty, Senior Vice President of CBS, gave a speech in 1997 where he spoke of his "concern about the long term asset value of Super16mm material as HDTV product because of Super16mm's low performance". He also showed several objective tests that compared the various film and video formats with compelling results. For example, resolving an image that demands 600 TVL/PH showed that an HDTV video camera can attain an 80% response, 35mm film has a 73% response, Super16mm has a 36% response and regular 16mm film only has a 23% response. Looking at the visual comparison of an HDTV camera and 35mm film transfer to HDTV shows little difference between them. Looking at Super16mm is a stark contrast to the 35mm film and HDTV video camera. Mr. Flaherty concluded that Super16mm film is not acceptable if the final destination is intended to be an HDTV standard, and therefore could not be considered a future-proof imaging format.

To be fair, the tests performed by CBS were met by the film community with howls of disapproval. Accusations were made about creating results born of vested interest against Super16mm film. It has essentially brought on a minor war between several interested parties. We've seen some very good looking Super16mm film and can hardly complain about the quality or apologize for the lack of resolution. However, the material shown by Mr. Flaherty was presented in a scientific, factual manner without an overt bias to any format. In fact, care was taken not to treat any format more favorably than another. For instance, a telecine colorist would normally crank in almost twice the noise reduction and image enhancement into a Super16mm film transfer than a 35mm film. This correction was apparently not done in these tests. Handling the Super16mm in the same way as the 35mm simply pointed out some differences between them.

It has been suggested that an even more objective test would have been to show projected film against the telecine transfer to prove or disprove the telecine's ability to handle Super16mm film. In any case, it is generally acknowledged in the film production community that 35mm film has a distinct advantage over Super16mm in all aspects except cost.

The following drawing is an indication of the difference between the area of a 35mm film frame and a Super16mm film frame:

The flexibilities of working in a 35mm film format will also allow adjustments to the images in the form of blowups and framing corrections in future product without suffering degradations as severe as those in Super16mm.

HDTV video cameras that exist now boast 1,000 TVL/PH of horizontal resolution, exceeding the available resolution of 35mm film. The potential exists for an HDTV video production to exceed the quality of an original film negative. The disadvantage of using a video format to acquire original images is a degraded flexibility for future reversioning. Once an image is limited by a video standard, the image resolution and aspect ratio is a permanent part of the image wherever it goes.

There are valid fears of future technical advances making the new HDTV standards obsolete. For instance, using an interlaced HDTV video standard for production will not allow smooth integration of the images into a possible future progressive scan product. A 35mm film original, on the other hand, can be converted to any television standard in the present or future without fear of making the images obsolete.

Creating an HDTV video product using the highest pixel count possible would be the best choice for future reversioning of video originated material. The highest quality HDTV video standard approaches the upper limits of what the human eye can detect and future compromises during reversioning will minimize the impact on image resolution. However, the pixels of a digital video image are in fixed rows and columns which translates directly from scan lines and horizontal pixel count. Technically, there is a danger of introducing artifacts into a video image called "aliasing" when altering the original placement of pixels during any conversion process. Since film has no regular pixel structure, there can be no aliasing artifacts when adjusting the position of a film image.

There are several alternative paths to making good ATV pictures, each with their rewards and troubles.

Upconversion to HDTV:
Technically, standard resolution television images can be converted to HDTV images with the use of an upconvertor. This device is a television standards converter that will interpolate, or "line double" standard resolution images to effectively be HDTV. If elements of current video tape libraries are to be included in HDTV product, upconversion is the only answer. Decisions about aspect ratio and framing will be encountered during upconversion of 4:3 programs. Programming finished as 16:9 SDTV video may be upconverted without regard to aspect ratio decisions.

There will be a strong budgetary temptation to use upconversion as a means to create HDTV masters using standard component digital editing equipment. A Digital Betacam master can be upconverted for delivery as an HDTV program. Even though high quality upconversions subjectively look appealing on an HDTV monitor, the upconvertor cannot manufacture resolution that does not exist in the original material. The television picture may be HDTV in an electrical sense, but not in image quality.

There will also be a strong temptation for some service bureaus to offer SDTV upconversion as a means to create HDTV programming without educating the client that it isn't true "high definition". It allows the service bureau to extend the useful life of their installed equipment base and possibly delay purchasing significant HDTV equipment. They can charge the client less than what full resolution HDTV would cost and demonstrate the quality of the upconverted images on monitors not likely to show the differences. The client who is not prepared to understand the issues is subject to getting hoodwinked into accepting it as true HDTV. This will not help the client in efforts to future-proof the product.

The issues of upconversion relate to image quality. A standard definition image will turn into a standard definition image with more scan lines. Increasing the scan line count will reduce some of the problems associated with our current television system. The image, however, is still short on the high frequency detail that makes a higher resolution image. Also, a standard image with 350,000 pixels upconverted to a two million pixel image will challenge the DTV encoder unnecessarily and degrade the image further at the home DTV receiver. Since the DTV standards allow for broadcast of what is essentially our current television resolution, the image will look better if it is transmitted as SDTV and not upconverted to an artificially high pixel count.

Broadcasters who are making the move to HDTV realize that upconversion will be necessary for all existing material, but they stress that upconversion is unacceptable when the opportunity for native HDTV production is available. They also stress that upconverted material must not be intercut with native HDTV material because of the dramatic resolution differences. All new production for several networks will mostly come from 35mm film transferred to HDTV formats.

Broadcast television will see HDTV originated commercials, a likely early contributor to HDTV material, intercut with upconverted SDTV program material. The visible differences between these image types may accelerate the desire to replace standard resolution material as quickly as possible.

Downconversion from HDTV:
High quality original images will allow for conversion to any lesser standard. The opposite is not true for upconverted images since the highest image quality available will be limited by the originating image standard. In order to future-proof new production, television producers should consider the shift to 35mm film. Film can be transferred to the coming HDTV standards without compromise.

Broadcasters will be simulcasting material in both HDTV and current NTSC channels for a number of years. CBS and NBC will be deriving the NTSC simulcasts from downconverted HDTV source material when possible and will avoid upconversion.

The use of 35mm film has historically outlasted video originated material and will also allow future television standards to be accommodated. The only reason shows like "I Love Lucy" are still around is because they were originated on film. The first few years of "Johnny Carson", originated on video, don't exist anymore. Some film producers I've talked to in Hollywood are advocating originating on 35mm and cutting the film negative for program finishing. That way the finished product exists as a complete entity that can be pulled out of the can years from now and run exactly like it was cut.

Image Compression on Transmission:
Compression is going to be upon us in the DTV world. The compression scheme for broadcast is called MPEG2 which can take the data required to create a video image and pack it more efficiently before it is broadcast. Our current NTSC television is an analog compression scheme where color is added to a monochrome picture by using otherwise wasted parts of the television transmitter power curve. Every compression scheme has its artifacts. MPEG2 and NTSC are no exception. The DTV broadcasts reaching the home will contain artifacts not present in the original material. We are exchanging one set of artifacts (NTSC) for another (MPEG).

The MPEG2 compression scheme has the ability to adapt to picture content. A video image is broadcast as a series of still frames, one after the other. MPEG2 takes advantage of the fact that much of a video frame is usually identical to the previous frame as well as the following frame. Instead of transmitting an entire video frame every time, the MPEG2 transmission scheme only needs to transmit a complete image every 8 to 15 frames. The rest of the frames are created by transmitting only what is different between the frames. With a relatively still scene, where the only thing moving may be someone's mouth, very little data needs to be transmitted to keep that scene in motion. As the image becomes more complex, the MPEG2 data rate will rise to accommodate the additional data needed to complete the frames.

The MPEG2 ATV encoder will be able to detect the presence of film originated material. Film, which runs at 24 frames per second in the U.S., must be transferred to video using a method that divides the 24 frames into the 30 available television frames. Every other film frame is held for 1.5 television frames, or three fields. Since the extra fields are redundant data, the MPEG2 encoder removes them and saves the transmission bandwidth. The home television receiver is told of the omission and will repeat the redundant fields during the display process.

The home television receiver is going to be a bag of tricks by itself. The set manufacturers will be trying to figure out how to make the sets cheaper so people will buy them. Along with that comes all kinds of schemes on how give the public a range of seemingly identical television receiver offerings with different price points that in reality perform wildly different. Be on the lookout for DTV receivers that can receive all DTV transmissions, either SDTV or HDTV, but convert everything to display on a less expensive standard resolution screen. Even though the transmitter is sending HDTV, the receiver is showing something less than HDTV.

Stress on MPEG2:
As a picture gets more complex with large amounts of fast motion and changes to the image, the MPEG2 compressor may be overrun with data that it cannot transmit fast enough. The MPEG2 encoder may decide to discard the high resolution elements of the image allowing the frames to be completed at some lower resolution. Fortunately, the human eye cannot resolve detail in fast motion anyway, so there is less need to transmit it. If done properly, the MPEG2 encoder will be able to significantly mask the absence of detail without calling too much attention to the failure mode it is in.

One of the things that can stress an MPEG2 encoded television image is noise. Active noise, or film grain, can be construed as motion to the MPEG2 compressor. Noise or film grain is also a high resolution image element that adds to the complexity of the image. If the noise becomes excessive, the picture quality may be compromised if the required data rate overruns the DTV channel's ability to transmit it. The presence of noise decreases the headroom the MPEG2 encoder has before entering a failure mode. This is yet another reason to avoid using Super16mm film in favor of HDTV video or 35mm film.

Another pitfall of film is gate weave. Using the steadiest possible film transport in a telecine will reduce the amount of interframe motion that can tax an MPEG2 compression scheme. Using 35mm film instead of Super16mm makes it easier to create steady film transfers. Of course, HDTV video originated material has no gate weave.

Compression in Post Production:
There has been a lot written about compression in post production. Compression has always been with us. The question becomes "how much compression can we stand?" The NTSC television standard is an analog compression scheme that compresses the color about 6:1 before adding it to the transmitted picture. The component digital 4:2:2 standard is also a compressed image where half of the color samples are missing. That's 2:1 compression in the color samples. Digital Betacam compresses a little more than 2:1 to make digital component recordings on Betacam tape. All of these compression schemes exist for one reason; to economically perform a recording or transport function that otherwise wouldn't be possible.

The HDTV video signal contains almost six times the data of a standard resolution image. To record that kind of data economically on technology available today requires the use of compression. For example, a full bandwidth HDTV digital tape recorder (Toshiba/Philips D6 format) costs $400,000 today. A video recorder that can record an almost identical picture with 4:1 compression (Panasonic D5 format) costs $95,000. Most people will accept the compression as long as they can't see the picture degradation and the D5 format does a very good job. The Sony HDCam format uses 6:1 compression. The HDCam shoulderable camera and studio recorders will be priced even lower than the D5 format.

Out of the 1920 pixels available in HDTV, the HDCam format will only record 1440 of them. Fortunately, there is very little detail information available in any standard scene beyond 1440 horizontal pixels. The resolution differences between the HDCam format and a full 1920 pixel recording are nearly invisible. The pictures are nothing to apologize for and the format will find its way into HDTV post production despite the theoretical quality reduction. For future-proofing, care must be taken to select a video recording format that provides the best cost/performance ratio.

Compression damages the ability to do multiple generation work, but it can have its place in areas where you only expect to go two or three generations. Transferring film original to a compressed video format is not a bad choice as long as the compression has no first generation losses. Cascading more than one compression scheme during post production may generate additional image artifacts and should be monitored to minimize them. As a point of comparison, the home delivery of HDTV images will incorporate 50:1 compression ratios and is not likely to be damaged by minor artifacts accumulated in post production. However, once compression artifacts enter a finished product they cannot be removed.

Film, especially 35mm formats and above, is currently considered to be the ultimate uncompressed, unadulterated image carrier available. Actually, film itself has compression characteristics. Film does an excellent job of compressing lighting ratios found in reality to the grains of the chemical storage media. Shooting an image of the sun, for instance, does not yield a film image as bright as the sun. Film will scale the relative exposure of the scene to what it can reproduce.

The 24 frame exposure rate of film conserves film stock while making an acceptable compromise in motion artifacts, sometimes known as "judder". The frame rate compresses the real time "reality" of life into brief time slices. Increasing the frame rate to 30 frames per second will improve the judder, noise and the apparent resolution of the film by putting more photosensitive grains in the path of the image. The ultimate film speed that will perfectly match the projected DTV standards would be 60 frames per second. That isn't likely to occur in normal production because of cost.

Film To Data:
For future-proofing, the best way to preserve film images (other than keep the film in perfect condition) is to record the images as data, not as video. Transferring film to video immediately limits the quality of the images to whatever the television standard allows. If the same film was scanned at high resolution and each frame stored as an image file, the image may be retrieved at a later date and converted to any television standard. A high resolution scan will easily scale to any likely video line count or frame rate. This includes exporting stored images as PAL since the film image is digitally stored frame for frame and not at the mercy of any television frame rate.

One likely preference of ATV broadcasts is to create material with an interlaced scanning technique. This allows material from current video systems, all of which use interlace scanning, to be easily incorporated into an ATV product. Interlaced scanning also can have significant motion artifacts, especially when dealing with film originals transferred via a telecine process. Film is more akin to a progressive scan video system. From a progressive scan original, a conversion can be done to an interlaced product. The opposite is not true. Once the images are scanned with an interlaced scanner, the artifacts are built in to the images. This is another consideration for future-proofing of production images.

Philips is showing the Spirit DataCine that has the ability to scan motion picture film and record the raw digital data onto one of many data archive formats. The scanning is done without regard to current or future television standards and is done in a progressive scan process. The data can be recovered and perfectly adapted to any future television standard since the images have not been touched by any television standard at all. The data from the scanner is good enough to output the images back to film. Degradation of the original digital data recording medium can be monitored and, if necessary, transferred to any future data medium without degrading the images. This theoretically will allow storage of the original data indefinitely, possibly long after the original film has disintegrated.

Several other film to data recorders are in operation designed for creating digital effects on feature films. The Kodak Cineon and Quantel Domino can scan a film negative at enormous resolutions (up to 4,000 pixels by 4,000 lines) into a computer workstation and output the result, including 3D embellishments, back to 35mm film without degradations. These types of data recorders may come into more common use, but they are currently in the "wretched excess" column of standard video post production.

The future-proof image library will be able to incorporate all of the available video, data and film standards. The future value of the image asset will be determined by two things; the quality of the image and the ability to find and retrieve it. Several types of computer based image storage and retrieval systems are in use world-wide. The successful systems will allow standard database architectures and a variety of storage medium options to suit the needs of the library. Accurate data entry, flexible search and retrieval and the highest quality image available will insure the future life of the image asset.

Conclusion:
For an immediate delivery mechanism, HDTV video originated product may be considered better than 35mm film. When the subject of asset futures comes into play, the producer may need to think again about an originating format. Film is the one medium capable of crossing most of the boundaries that exist in program delivery. It can be transferred to all present and future image transport formats including the current NTSC and PAL video frame rates. Particularly, the use of 35mm film closely matches the High Definition television formats coming into being. The quality of Super16mm film is suited for program material with modest performance expectations compared to 35mm film. Transferring 35mm film original to digital data will ensure the longevity and recoverability of the original image assets.

HDTV video production may rival or exceed the image quality of 35mm film and allows for downconversion to any lesser video standard. However, any video image standard will become a limiting factor for future use of the product. The resolution, bit depth and aspect ratio become frozen in the video product and cannot be changed without some compromise. In particular, the subject of interlace versus progressive scan image formats may become a factor in judging the future value of an image asset. Products are being designed and tested that will capture live images at 1080 lines with progressive scanning. The equipment, particularly the recorders, will have to bear enormous data rates to store these images. They will be the direct rival of 35mm film capture when available.

The safest format for the foreseeable future is 35mm film. The next best format is HDTV video origination for all its resolution abilities. Super16mm film allows the same image flexibility as 35mm film without the quality. For a future-proofing function, film has been proven to outlast video tape for durability and certainly outlast video tape formats for popularity. The least desirable format for future-proof images is standard definition video at 16:9 aspect even though it will outperform Super16mm film in an immediate delivery mode. The need for upconversion from lesser formats will certainly reduce the future value of image assets and should be avoided.


Also available is an article on 24 frame HDTV production and distribution.

Steve Wiedemann is Sr. VP, Director of Technology for Henninger Media Services.