Film vs. Digital
Both film and digital capture have significant advantages and disadvantages. It is critical to examine all the factors before you even consider switching to digital. For example, it's important to understand how digital cameras record images. It's also important to know, up front, how much data must be captured by a digital camera for the intended output requirements. One potential advantage of conventional film is that it contains a great deal of data. Even a 35mm image on fine grain film can produce a gigantic file when scanned with a high-resolution, high-quality scanner. With digital cameras, there is a tradeoff between file size and the methods used to record the data. However, as we will see, the amount of data a digital camera can produce may not necessarily indicate how large a file can be output, nor the quality of the file.
Another issue to consider is whether the camera is one that can be used as a single, self-contained unit, much like a conventional camera, or if the camera has to be tethered to a computer. Most instant capture cameras have been designed to work in the field with removable storage devices for storing the image files. Scanning cameras are usually intended for use in the studio and must be connected to a host computer. Many photographers have used these "studio" cameras successfully on location when mated with some kind of laptop computer.
Area array CCD cameras have two disadvantages: First, the file sizes they produce are limited and the cost is based on the maximum resolution they can provide. It's very expensive to produce a high resolution, two dimensional CCD. Cameras that record 2,048x2,048 pixels and produce 12MB files can cost $20,000 or more. Another disadvantage is that two-dimensional CCDs are monochromatic. In order to capture a color image instantly, the CCD must have a matrix of colored filters on or in front of it. Software is required to interpolate the data and create a color file.
The resulting digital files usually contain color artifacts can be eliminated with the software. The alternative, which only a few camera manufacturers have undertaken, is to use three separate CCDs and a beam splitter to expose each CCD through red, green, and blue filters. This method results in true, non-interpolated color, but increases the camera cost to far more than a single CCD capture camera of similar resolution.
There are a number of cameras that use single CCDs with rotating red, green, and blue filters. Each filter moves in front of the lens of CCD during exposure to create "true" non-interpolated color files. However, this technique also forbids any movement of the subject, since the rotation and individual exposures can take several seconds to complete. Therefore, this would not be considered an instantaneous capture camera. While this kind of system is significantly faster than a scanning camera, and the color is indeed true, these cameras still suffer the resolution vs. cost consequences of instant capture CCD cameras.
Scanning cameras use a single row of tri-linear CCDs and record the image by moving the row of sensors over the film place, thereby capturing "true" color. These cameras are very similar to flatbed scanners, except they are usually housed in camera backs that attach to conventional film cameras, either medium format or 4x5-inch view cameras. While it may take several minutes to create a file, these cameras are capable of producing extremely high-resolution images, much like conventional scanners.
What may not be obvious is that the light source can present problems with many scanning cameras, because it must be a flicker-free continuous light source. For example, halogen lighting can have variances in the line voltage and the resulting flicker will be recorded during the scanning process. High-frequency fluorescent or high-modulation (HMI) lighting systems, both of which are very expensive, are two solutions to this problem. At least one scanning camera manufacturer has come up with another answer: Phase One's scanning camera backs feature a "flicker suppression" technology to minimize the effects of light variance.
Which One is Right?
Why Film, Why Digital?
Digital capture in a studio situation provides the photographer a great deal more control over the photo imaging process. The availability of a large calibrated display with a digital densitometer offers much more critical information during setup than running a Polaroid test. More important, why shoot film when it will most likely end up being scanned and separated for final output anyway? The digital camera eliminates multiple steps in the prepress printing process. This issue alone is one of the most compelling reasons to shoot digitally.
One must also consider the cost savings by eliminating film, instant prints, and processing. While I'm the first to admit that professional digital cameras and supporting hardware and software can be costly, many photo studios have paid for this investment quickly in film costs alone.
There is one other major issue to consider in the film vs. digital debate: the quality of the digital capture. Each type of film has its own biases. Kodachrome has a unique look that differs vastly from Fujichrome, and many photographers use a film based on its particular characteristics. However, isn't it also true that these unique characteristics can cause problems with color fidelity? If each color film has its own color gamut (the number of colors it can successfully and accurately render), then wouldn't it be easier to achieve accurate color reproductions by avoiding film altogether?
What is the difference in resolution between digital capture and scanned film? Film images can be enlarged only to a certain degree before their quality begins to degrade. Whether scanning at very high resolution or cranking up the enlarger in the darkroom, the film becomes more "grainy" as it is enlarged. But digitally captured files have no grain. One advantage that digital camera manufacturers have claimed repeatedly is that while their files may not seem large, you can use software to double or triple their size with no noticeable degradation in image quality. Could this indeed be true and, if so, by what factor?
In order to test film vs. digital, I decided to use two completely different kinds of digital camera backs designed for completely different kinds of shooting situations. Both backs fit on conventional cameras, which meant I was able to load a film back on the same camera and shoot the subject both ways. I took the conventional film back to my shop and scanned each piece of film on a drum scanner to the same file size as the digital camera back produced. The next step was to output all files and compare them for quality and accuracy. I then resampled each file and attempted to determine how well each file underwent this resizing.
Phase One and 4x5 Film
The PowerPhase back was tethered to a Macintosh 9600 dual 200MHz machine with 756MB of RAM and a Radius PressView 17SR display. Starline's aim in setting up this studio was to shoot very large, original artwork for output to posters on their six-color presses. Shooting colorful artwork created with numerous pigments and paints is a difficult task. For my first test, I copied a very vibrant and colorful painting by artist Tavlos on Fujichrome Provia 4x5-inch film and the PowerPhase back. The cost for 10 sheets of Provia film was $26.88.
Setting up the artwork for the shoot was simple, since Starline has a vertical copy stand rig for artwork. I decided to start shooting with the digital back, as I would be able to preview all my work on a calibrated 17-inch Radius PressView display. If I had decided to shoot with film, I would have likely used a number of Polaroid tests to examine the exposure, lighting, and so on. Since I was able to capture the prescan onto the PressView, I could take advantage of Phase One software's pre-shooting and setup features such as the ability to:
After making the prescan with the camera, a process that takes about 20 seconds, I was able to crop the full image of the painting and produce a 121MB file in less than 14 minutes, using an f/11 lens opening. The file was saved to disk in RGB, but the camera can do on-the-fly CMYK conversion using a custom ColorSync profile. Since RGB files have a wider color gamut and Starline intends to use the file for various output needs, I decided to forgo the CMYK conversion at this time. Note that the only difference in using the PowerPhase back vs. Fujichrome 4x5 film is that a special filter on the lens was required to eliminate IR light to the scanning back. The filter factor was half an f/stop, so when it came time to shoot conventional film, I simply closed the lens down to f/11.5 and shot the painting at 1/4 second.
I exposed four sheets of film: two at normal exposure, one at a half stop over my normal exposure, and one at a half stop under normal exposure. There's no need to bracket exposures when using the digital camera back, because the software provides far more robust data about the file, including a histogram and densitometer reading. A back that is capable of providing 6,000x8,400-pixel resolution produces a staggering 50,400,000 pixels, each with a numeric value that can be read with the digital densitometer. Think of having a 50,400,000-point spot meter, and you can see how capturing a file digitally makes a conventional light meter obsolete.
The ability to target a specific highlight and shadow range in the prescan, as well as pull a tone curve prior to capturing 14 bits of data, should give you some idea of the control available with digital capture. Color balancing a digital camera is as simple as clicking an eyedropper over a gray card placed on the set for the prescan. Balancing conventional film means shooting film tests with various filter packs. Often, a variety of gel filters need to be placed over the lends, and the photographer must use the same E-6 lab and film emulsion batch to ensure consistent color. Should the film manufacturer or E-6 lab vary even slightly, the resulting film will not be consistent. These small variations can easily be corrected when the film is scanned, if the scanner operator has a reference point on which to set a neutral value in the software, that is similar to the scanning camera.
I had three sheets of 4x5 film processed, keeping the last exposed sheet in the holder, in case I wanted to push or pull the film a fraction of a stop during processing. Fortunately, the film I exposed at the normal exposure was right on the money. I think knowing the exact exposure from the Phase One back and then closing down the lens a half stop for the IR filter I removed was the key to this. I had to wait 24 hours for my film to be processed at a cost of $2 per sheet. The 4x5 film was scanned to exactly 121MB on a ScanMate 5000 drum scanner to create an RGB file. I attempted to pull out as much detail as possible in the prescan and color balance the highlights and shadows to neutral. The time to create the scan, including oil mounting the film on the drum scanner and spotting the resulting file, was 20 minutes. Not having to spot out dust in the digitally captured file is a great advantage in my book.
When I output the two prints to the Kodak dye-sub printer at 100 percent resolution, I was further amazed by the quality differences between the two prints. Anyone looking at the print from the film image would say it looked great, but it didn't compare to the quality of the print made from the Phase One digital file. Again, the grain in the film was apparent, with no sign of grain in the digital capture. It almost looked like the difference between seeing an 8x10 print made from a 35mm va. an 8x10 print made from large-format film.
Another interesting factor I noticed between the two files was color. With the PowerPhase camera, I gray-balanced the prescan (hence, the high-resolution scan). The elephant in the image is not completely neutral gray when read with the densitometer, but it's very close. When I scanned the actual 4x5 film, I set the scanning software's gray balance to the same area I set the gray balance with the digital camera. Yet the elephant in the prescan still looked somewhat blue. I noticed that other colors in the transparency scan were not falling into the same hue and saturation as that of the digital camera file. It appeared to me that the film had a unique color response that didn't compare to what I expected nor did it match the original accurately.
The file from the digital camera was more authentic. I had to use many of the tools available in the drum scanning software to achieve a prescan preview that was close to what the Phase One back produced. I had to adjust the shadow and highlight curve to record the detail I saw in the purely digital capture. But film has a characteristic shoulder and toe, known as an H&D curve, that required this kind of curve adjustment; the Phase One back didn't have this limitation. The digital back appeared to capture a significantly wider tonal range than the E-6 film.
My conclusion is that the
image captured from the Phase One back was of far superior quality.
While you may not see this difference when printed with a halftone dot
in the magazine, you can easily see the difference on screen and on
a high-quality dye-sublimation print. I also think that ability to accurately
gray-balance at the time of the shoot, plus the full control over the
tonal curve with 14 bits of data, and the increased tonal range of the
Phase One proved to me that film lost this contest by a long shot!