by Matt Hurwitz
photos by Terry Lilly

		From left: Panavision's John Galt, EFILM President Joe Matza, UPMs Cleve Landsberg and Gail Fortmuller, and Jeffery Pierce, Society of Motion Picture and Television Engineers. Click image for larger view.

The discussion was moderated by AD/UPM/TC New Technology Committee Chair Susan Zwerman, who introduced a panel of five DV experts who each discussed a different aspect of DV production. In addition, an informative set of handouts put together by the New Technology Committee was given out to attendees. [Click here to view Digital Odyssey handouts]

Cinematographer and former Vice President of the American Society of Cinematographers and of the International Cinematographers Guild Local 600, Steven Poster, talked about differences in lighting between digital video and film processes. "A colleague once said, 'If you can light, it doesn't matter what you shoot with; conversely, if you can't light, it doesn't matter what you shoot with."

Poster, who has just finished shooting the upcoming Stuart Little 2, described the two elements of the recording medium that affect the cinematographer's ability to control the image: dynamic range and resolution. Dynamic range is the range of tones of light which can be recorded (either on film or video) and still maintain discernable detail before complete white ("highlight") and before absolute black ("shadow"). In addition, he explained, "the number of shades of gray between those extremes defines a quality that affects the way an audience understands" the image, adding that 35mm motion picture film is still the benchmark, as it has the greatest dynamic range.

Poster noted that HD digital imaging has a great ability to record shadow detail, but does so in a somewhat unnatural way. "Film has a natural roll-off of detail on the lower end," as we see it in nature, whereas HD has a more linear "dive into the shadows."

With regard to resolution, the cinematographer demonstrated the cameraman's desire to utilize a format with a resolution close to that of film, showing several example clips images from recent productions produced on film, as benchmarks for which to strive. A clip from Stuart Little 2 showed the computer-generated mouse's antics on a real-life set. The animation was produced at 2K (about 2000 pixels of) resolution, which is common for visual effects. The background photography, shot on film, was scanned, frame by frame, into a computer, and the mouse star was animated into the scene at that resolution. However, demonstrations showed that scanning film at 2K produced a background image that didn't match the detail found in the visual effects (whiskers, fur, etc.). So it was decided to scan the film at 4K resolution, a resolution which approximates that of film itself, and then "down-rez" (reduce the resolution via a computer algorithm) to 2K for addition of the visual effects. Then, when transferring the finished image back to film, the product was "recorded out" (using a laser recorder) out onto the film stock at 4K resolution, again, about that of film. That combination, of 4K real-life backgrounds and 2K–produced visual effects, created startlingly lifelike and entertaining imagery.

"There are many really good and valid reasons to shoot in digital," Poster told the audience, "but I agree with John Bailey," cinematographer for The Anniversary Party, a well-known film shot using digital video, "that making digital imitate film is not one of those reasons. Digital has its own unique look and feel. The most successful uses of digital imaging have been the projects that use these qualities and have not tried to hide them." He also cautioned against over-simplistic approaches to the use of digital video. "Remember: artistic ability, writing, direction, cinematography and acting are skills; technology is a tool. Tools aren't skills; skills use tools."

Panavision's John Galt, the company's Senior Vice President of Advanced Digital Imaging, explained some of the basics of digital video, helping to elucidate the many technical terms often heard when discussing DV. Galt explained the differences between the different categories of digital video formats, from high definition to digital video to mini-digital video. Besides the different categories of digital cameras systems, it was noted that proprietary system names such as "DVC Pro," "Digital BetaCam" and "HDCam" are often confused with the formats (such as those mentioned above) utilized by these system.

A number of parameters can affect in which category a digital video format belongs, among them data rate and resolution. Data rate describes the number of megabytes of information the camera captures and records per second; i.e. relating to the amount of data the camera produces when yielding a frame of image. This can vary with format, as some formats utilize "compression," a mathematical reduction in the amount of information represented in the image. The higher quality formats, such as "D1" and HD have higher data rates than those systems which produce lower quality images, such as mini-DV systems.

Resolution is defined by several things: the number of tiny picture elements, or "pixels," an image is broken down into to form a frame of camera image, as well as the degree of compression and lens quality. DV formats typically utilize NTSC resolution, the standard resolution used in video in the United States, which is 720 pixels across by 480 pixels high. In other words, the image is broken into rows of picture elements numbering 720 across by 480 rows high. HD formats use much higher resolution (1920 x 1080), typically around the 2K resolution which is closer to that of motion picture film.

Galt noted that, "none of the above defines the actual image quality obtainable with these recording formats," adding that "even within the same format, there is great variation with digital image quality."

The technology of the recording medium was described, as well as the variations found in different systems. Digital video images are captured, instead of on frames of film, on a "CCD." A CCD, which stands for Charge-Coupled Device, is a special chip that converts light into electronic signals, which can then be stored on tape. CCDs range in size from 1/4 inch for lower quality image cameras, to 2/3 inch and one inch for high definition video. Each CCD can produce up to 2.2 million pixel images (a typical pixel count for digital video), or even 3.3 million pixels, which, as Galt pointed out, doesn't necessarily increase the quality of the image. Some cameras use a prism to split the incoming light image into three light beams (R-G-B), each hitting its own CCD (in other words, one 2.2 million pixel CCD for each of the three primary colors, making a total of 6.6 million pixels per image). Others use a single CCD chip that captures the image with a mosaic pattern of color 2.2 million picture elements.

		From left: Panavision's John Galt, EFILM President Joe Matza, UPMs Cleve Landsberg and Gail Fortmuller, and Jeffery Pierce, Society of Motion Picture and Television Engineers. Click image for larger view.

Another topic that is often the source of great confusion, interlaced vs. progressive, was explained. Television in this country, operated at the NTSC standard, is recorded and transmitted at a frame rate of 30 frames per second (fps). It was found, however, that bright images at this frame rate tended to flicker. So, RCA television pioneer Vladimir Zworkin in 1934 introduced interlace to reduce the problem. Interlace breaks each of those 30 frames into two fields. Each field carries every other line of the vertical resolution of the scanned image — one, say, the even rows, and the other carrying the odd rows, meshed — or "interlaced" — to form a whole frame's worth of image. So, therefore, there are 60 fields — 30 frames made up of two fields each — scanned each second in the interlace system.

When transferring interlace images — particularly those recorded in the 30 fps NTSC format, a very complex mathematical computer algorithm is required to convert that frame rate to the 24 fps used with motion picture film. To some degree, this involves "compression," dropping of some of the image field components. This process can produce some "smearing" of some images, particularly images with moving objects. It is for this reason that filmmakers using DV will often shoot their project using a PAL–format camera system. PAL, the standard for television in most of Europe, though also an interlace format, uses a frame rate of 25 fps, which more closely approximates the 24 fps film frame rate.

A progressive video format does not have the two interlaced fields. It captures the image in a single pass for each frame. Progressive format DV cameras (usually identified by a "P" in their model names), as described above, capture the image with a CCD chip, instead of an electron beam scanner. The light from each frame's worth of image hits all of the pixels on the chip at the same time, producing a single frame of image at whatever resolution is specified. This approximates, then, the very manner in which a film camera captures images onto its own film medium.

So for video, one preferred format is one that combines both this progressive-type frame capture, and does so at a frame rate identical to that of motion picture film cameras. Thus the "24P" HD camera, a popular Sony camera system outfitted with Panavision lenses, which captures images in the progressive format at 24 frames per second. With its nearly film-level resolution (1920 x 1080) and film-friendly frame standard, this 24P format produces the closest to film that DV can provide. It transfers to film at the 24 fps frame rate, producing little in the way of the "smearing" described above for action movement, and with little "artifacting" (loss of information due to compression. This can appear on screen as an image with jagged edges), again, mostly due to the compatible frame rate.

Galt doesn't support the idea that digital video is a simple medium to jump into, nor that one can trust outright the claims of any manufacturer. "If you're shooting in film," he says, "you're dealing with a whole community and infrastructure that has been doing this for 100 years, so you're in the same hands most of the time. Once you get into digital imaging, there's a lot of 'snake oil,' and there's a lot of people selling that snake oil, and you have to be awfully, awfully careful. Test the system you're going to use, and make sure that you get people who have your best interest in mind to work with you."

EFILM President Joe Matza next discussed the process of transferring video to film. Completed DV or HD projects are copied onto a digital disk recorder, and then, using proprietary software (which handles, among other things, frame rate and resolution transformations up to the target 2K resolution for film output), are transferred to a computer. The image is then written, frame-by-frame, using a laser film recorder, onto fine-grained intermediate film stock. This laser writing can produce either a positive or a negative, depending on if one is seeking an inter-positive or an inter-negative.

Film-to-film processing is also accomplished, using a similar operation, for such things as film restoration and preservation, or, as described above, the insertion of visual effects. Film is scanned, one frame at a time, at either 2K or 4K, and the images are then available in the computer environment for any needed adjustments. They are then recorded out, using the laser film recorder once again, to film.

Matza also noted the growth of use of this very same process for "digital mastering." This process involves scanning in or transferring the source/camera material to the computer environment, and utilizing the lab's facilities for process shots, titles, color-timing, etc. The advantage, he says, is in avoiding generational loss, for example, in printing a cross-fade, since the fades are performed digitally, not optically.

The digital process can also produce high-quality inter-negatives, which, in turn, produce top-quality prints. "From your digital master," he explains, "you're going to be able to make multiple negatives — original digital negatives from that single digital master, and bypass the IP/IN process. The print in the theater will be a first-generation print off the original."

DGA unit production manager (UPM) Gail Fortmuller talked about the changes her production team went through when transitioning from film to high definition video for CBS's Diagnosis Murder, for which Gail has been UPM since 1999.

After filming Diagnosis for seven years, CBS lowered the production company's budget, so it was decided to explore using high definition video as an alternative. After one short week to adapt to the change, the team began using HD.

Difficulties arose mainly, says Fortmuller, from trying to "make something look like something it wasn't. We wanted a film look, and we were shooting on high def," she explains. "The interior sets looked like plywood with wallpaper on it, which is exactly what it is." The depth of field on the HD camera didn't allow the background images to soften. "Even with all the reality shows out there," adds Gail, "I don't think people are ready to watch a prime-time drama that looks like an afternoon soap opera."

The camera itself also created problems, mainly due to its size. "We only have 11:1 zoom lenses to work with, and that made the length of the camera four feet" — a tight squeeze, as well as a heavy load, for hand-held shooting. Indeed, using the camera as a steadycam was difficult due to the weight and balance ("I think the operator's chiropractor liked it just fine."), and working inside a car was "a nightmare. When we did over-the-shoulders in a car, it seemed to the crew that the actors were eating the matte box, because the camera was so close." The camera seriously limited the directors' creativity, due to its awkwardness.

Some of the advantages the team found included being able to have dailies first thing the next morning — or even the same day, if needed. Inserts using existing stock footage (from film) were edited in seamlessly.

As for costs, Fortmuller had done a dollar comparison, which she shared with the audience. While there were obvious savings in the thousands of dollars for video tape vs. film stock and developing costs, there were other costs, such as for video transfer to film and increased online editing fees. Titling cost four times as much for HD than for film.

The production was able to save about $4,000 per episode. However, Fortmuller says, "As a member of the Directors Guild of America, that savings does not make much sense when the directors feel so stifled working with this format."

Gail Fortmuller was followed by another DGA UPM, Cleve Landsberg, who discussed sound recording when working in the high definition environment. Landsberg, who is a member of the AD/UPM/TC New Technology Committee, described the two methods applied to sound recording when using digital video — single system and double system. Single system involves using the same medium — in this case, the video camera, to record both picture and sound. Double system involves, as in the case of film, a camera to record the image, and a separate sound recording system to record the audio, typically on a sound cart.

One of the advantages of single-system recording, he pointed out, is that the camera yields a picture and sound in sync, and, thus, can be copied immediately for dailies. But single system requires an unwieldy cable harness, since signals and time code still have to be monitored and controlled at the sound cart. And with multi-camera photography, such an operation can become quite cumbersome, especially with hand-held camera work. With additional cables to handle, particularly during shot setup, additional crew members may be required.

Double-system recording is recommended by most for 24P HD work, Cleve noted. Double system, with a tape recorder right at the sound cart, enables the sound mixer to have the greatest control over quality of the recording. In addition, he can record using a multi-channel device, such as a TASCAM DA-88 eight-track digital recorder, allowing better isolation of microphone inputs, thus creating more options during post-production processing. A backup recording, meanwhile, can actually be made on the HD camera's soundtrack, via a mix sent back through a cable harness. This mix, again, makes for a quick turnaround on dailies, and also helps the editor by providing a rough soundtrack to cut against.

Landsberg stressed the importance of communication and, most important of all, planning, when considering sound recording methods. A major concern is being sure the same protocol is used for picture recording as for sound recording. Digital audio recording machines typically have a 29.97 fps frame rate (about 30 fps), while a color 24P HD camera photographs at 23.97 fps (about 24 fps). "Planning ahead cannot be overstated," says Cleve. He impressed upon the audience to plan meetings including all involved (sound mixer, editor, producer and others) to make sure everyone is on the same page in this regard.

Another important concern was also discussed. DV allows for continuous rolling, without breaks between takes. This can result not only in fatigue for the boom operator holding a boom for a long period of time, but for other technicians, as well as the cast, who otherwise might need a moment in between takes to collect themselves.

Jeffery Pierce, former Director of Digital Film at Santa Monica Studios and Vision Art, and a current member of SMPTE, explained the concept of "universal mastering" in the digital domain. Universal mastering basically involves transferring the recorded image (either on film or video) of a project into the computer system stored as digital data to make that material available for all those who have a hand in production, and then available, at end, for production of delivery formats. With the original material available on a server, editors, effects artists, sound artists and others can all access material from remote locations, perform their work, and save their input for the following step to be completed by another artist.

Once completed, a final high-quality digital master is then available for outputting — to a laser film recorder to film, for projection. This digital master can be formatted to any digital video format, such as NTSC, PAL or DVD for release in any market. Streaming media formats, such as Windows Media, can also be derived from the same master.

Special thanks to all of the speakers and the AD/UPM/TC New Technology Committee for their efforts presenting this successful seminar.

Table of Contents     Top of Page