We ran the test at two facilities: Sony's Hi-Definition Center and Digital Image, formerly 4MC. In many ways, their processes are very similar. Sony's method is to up-convert the DigiBeta tape to their hi-def standard, then use an electron beam recorder to make a 35mm, sequential, black-and-white, three-color YCM, reversal, separation positive on 5360. From that they print a color negative on (EXR 50D) 5245, and then do answer prints on 2383. Digital Image also uses an electron beam recorder to make a sequential, black-and-white separation positive, doing so in 16mm with 7360. They then make an optical blow-up to 35mm color negative on 5245 and answer print on 2383.
Both tests produced nearly identical, and good, results. Subjectively, I would compare them to how Super 16 looks when blown up to 35mm. In both tests, the exposure latitude - based on an evaluation of the print in projection and on a light box - was a good seven stops between the blackest black and the whitest white. Color saturation was very good, and the blacks were quiet and solid. Overall, the look was better than I had hoped - it really didn't look like video. In addition, the difference between the Sony and Digital Image processes was minimal when projected. At first glance, it might seem strange since the Sony process starts with an up-conversion to hi-def and uses 35mm all the way. That alone would lead one to think the Sony process is perhaps sharper and less grainy. But the results on screen - in terms of grain, sharpness, and saturation - were essentially the same. Digital Betacam may be the limiting factor. That is, if we had started on hi-def, we might have seen a difference. But in this case we didn't. Both tests, by the way, were preceded by color bars, which was the only information used to set up for the transfer. That, and the fact that both procedures utilize the same film emulsions processed to the same standards, is probably why both tests also look almost identical in terms of color balance, brightness and contrast.
Armed with this information and the confidence gained from the tests, we forged ahead. I shot this exactly as I would a film. Our camera crew consisted of operator Michael Santy, first assistant Greg Cosh and I. Our only lens was a Fujinon 14:1 (8.5mm - 119mm) zoom lens behind a Chrosziel 4 x 4 matte box and studio style follow focus system. Camera support was a Sachtler 20 Plus fluid head, the usual assortment of sticks, hi-hat, and other mounting equipment, plus a Super PeeWee dolly.
Much can be said about the pros and cons of digital production. In many ways, I view the whole process as a re-invention of the wheel. It begs one to question, "If it isn't broke, why fix it?" On the other hand, the game is changing fast and the quality is increasing rapidly. But I've heard a lot of talk lately about the notion that video requires less light than film. That simply isn't true. Just look at the numbers published by the manufacturers. The Digital Betacam we used is rated by Sony at an E.I. equivalent of 280. The High Definition digital cameras are slower than that. Eastman Kodak's Vision 320 (5277) is rated by them at E.I. 320. From my recent, direct experience (see: "Nebraska… Shooting a No Budget Feature" ICG Magazine, October '97) I know that a feature shot in Super 16 on Vision 320 will meet or exceed the technical image quality of the Digital Betacam to 35mm I've seen so far.Warm Blooded Killers wasn't done on digital video because it's better, or requires less light, or less people or less whatever, than film. It doesn't. The project was done on digital video because tape raw stock cost less money than film up-front, and the quality is acceptable.
It's like the old notion I've heard from more than one actor, that they just don't look good on video. Not true! These ideas share the same, misguided assumption: That the "look" of a piece of photography, whether it be stills or motion pictures, is more about the camera-the technology-than it is about the photographer, the lighting and the lensing. Such a notion is completely wrong! If you want a film or a video to look like news or a documentary, then you can shoot it handheld, with a small format camera with no lighting and a couple of people on the crew. If you want it to look likeGone With The Wind then shoot it on a large format camera with whatever lighting and personnel you need to get the look and control you want. That is really what film and video lighting is all about. If you doubt what I'm saying just take a look at any quality television studio where an anchor person sits in front of a state-of-the-art, studio camera. If video requires so little light then why are these anchor people surrounded by so much light? Because that's what it takes to make them look pretty. That won't change as long as we care about how things and people look.
In the second setup, a young man sitting at a table is illuminated by a practical, Tiffany-style lamp at 2800° Kelvin. The kitchen appearing in the background is illuminated by COOL WHITE fluorescent light at 4100° Kelvin. Again, data points are indicated on the picture. These results convinced me that I could get a "film noir" look.
Diary of a Digital to 35mm Feature
by Chuck Barbee
First Published in ICG Magazine - June 2000
Summer 1999 was turning out to be a very slow period for me. At the end of June, I was working the SOC booth at ShowBiz Expo when a producer approached me, placed his card in my hand and proceeded speak about a film with the inviting title Warm Blooded Killers. The man in question, Dan Wulkan, had such great energy and enthusiasm for this digital-to-35mm feature that I found it difficult to not be drawn in, even though its projected $25,000 budget seemed impossibly low. In spite of that figure, I agreed to meet the following week with Dan, writer/director Stephen Langford and co-director Nicholas Siapkaris to discuss the possibilities.
I was eager for an opportunity to shoot a digital feature actually to be printed to 35mm. I've had numerous opportunities to shoot low/no-budget digital features, but have turned down most of them because they didn't stand a good chance of getting to that final step. My motivation for doing such a project is seeing a projected 35mm print. A 35mm film transferal is a necessary phase for any widely distributed digital feature - for the time being, at least - and I want to be comfortable with that process. Meeting with Stephen and Nick convinced me that this project stood a very good chance of making it to film.
Warm Blooded Killers is an off-beat, dark comedy which offered me the chance to shoot video in a very dark, film noir style: interiors alternate between a dark, strip-club atmosphere or are daytime shots brightened only with natural light coming in through windows. In my mind's eye, I saw many interior scenes played against bright windows, with little or no fill on the camera side of the subject. Traditionally, such shooting is a total no-no for video. I've long held the view a lot more can be done with video if it's thought of as reversal film. Warm Blooded Killers was a good vehicle with which to put that notion to a test.
I believe that the ideal digital-to-film camera will be a 24 frame-per-second, progressive scan, high-definition camcorder the same size and configuration as the HDW 700. But that isn't available yet, and it would have been out of our budget range anyway as was the HDW 700. The best camera within our budget limitations was a widescreen (16:9), 30 frame-per-second, interlace scan, Sony DVW D-700 Digital Betacam (DigiBeta) rented to us by Plus 8 Video.
I pushed for a Plus 8 camera package because I knew they would bend over backwards to make sure that the D-700 was set-up optimally for the tape-to-film process. I could also count on them to be there for me during the testing and shooting phase, just as a good camera rental house should.
Since I didn't want to blindly take on this shoot, I also shot a test and printed it to 35mm. I wanted to test the exposure latitude of the entire camera-recorder-scanner-film system to see just how far it would go. I also wanted to test a range of panning speeds to evaluate the best rates for avoiding the "stuttering" that tends to occur in the 30-to-24-frame translation. This stuttering looks similar to what occurs when a film camera pans too rapidly. Sony deals with this problem by utilizing a blurring algorithm, which tends to smooth the stuttering. But it becomes apparent again when rapid motion stops and the image's moving parts suddenly sharpen quite noticeably. Digital Image, where the finished feature was ultimately transferred, does not use the blurring technique, preferring instead to maintain a uniformly sharp image.
In addition to testing exposure latitude and panning speeds, I wanted to compare and evaluate a couple of zoom lenses and test the effects of increasing gain in the camera (the equivalent of force-developing film). Since we could only afford a 2.5-minute test, I had to plan carefully. I shot lens comparisons, gain tests, gray scale and Macbeth charts and two actual static "scenes" with stand-ins, each 20 seconds long. One was a day interior, the other a night interior with mixed tungsten and fluorescent light. I also shot the tests "clean" - with no diffusion or softening of the image. The camera itself had already been set-up with very low edge gain or detail. Both Sony and Digital Image recommend this for their processes, and it's also happens to be the way that Plus 8 prefers to setup its cameras.
As I mentioned before, experience has taught me that shooting video is a lot like shooting reversal film. With negative film you expose for the shadows and let highlights take care of themselves. With reversal film and video, you have to expose for, or protect, the highlights. Once the highlight detail is gone, it's gone forever. On the other hand, video seems to hold detail pretty far down into the blacks. With this idea in mind, I set up a couple of test shots which would give me a pretty good, quantitative evaluation of the system's latitude and sensitivity.
I like to deal with video camera sensitivity in film terms. Film sensitivity is expressed in EI or DIN numbers. To change film's sensitivity, you "push" or "pull" the developing process. With video, sensitivity is expressed in engineering terms - such as 2000 lux @ f-8 with 89.9% reflectance, which is the Sony DVW D-700's rated sensitivity. Unlike film, this type of rating is telling you what is needed to get white, not what is needed for mid-gray. Like film, you can change video sensitivity by increasing or decreasing the gain or amplification. Unlike film, video gain is expressed in decibels or dB, not f-stops: 6 dB equals one f-stop. The DVW D-700 has several pre-set gain positions. They run from -3dB to +18dB in 3 dB (half-stop) increments.
Since the typical video-style expression of sensitivity is so different, many cinematographers shooting video tend to work backwards, using a light meter to determine an equivalent EI number for judging video exposure. For the DVW D-600 & 700 cameras, Sony kindly publishes the exposure index number equivalents for all of the preset gain settings. Sony's numbers are: -3dB= 280EI, 0dB=400EI, +3dB=560EI, +6dB=800EI, +9dB=1120EI, +12dB=1600EI, +15dB=2250EI and +18dB=3200EI.
I did my own correlation using a combination of reflected and incident readings on gray cards and actual scenes. I evaluated exposure subjectively on a monitor, objectively on a waveform monitor and in the camera's sharp black-and-white viewfinder. Professional video camera viewfinders can indicate when the picture reaches certain preset exposure levels. You see little diagonal lines - Zebra bars - in the areas of the picture that have reached the preset level. I had set this camera's Zebra bars to indicate 100 IRE, essentially full exposure. That way, I can easily protect the highlights and judge the rest of the exposure by eye, meter and experience. Actually, black-and-white viewfinders are great for judging exposure too, because what you see is strictly luminance values, but that can be a hindrance when composing because it can mask color-related problems. Even though I used a waveform monitor to evaluate and correlate EI numbers and lens stops, I wouldn't be using one on location - I wanted to keep the "engineering" side as low profile as possible. The camera crew would be working in complete "film style." The DVW 700 facilitates this by use of a setup card which is pre-programmed and can be read as often as necessary to ensure the camera's set-up parameters are consistent.
In my tests, I came up with slightly different EI numbers than Sony, but that is probably due to the fact that our camera was set up in a configuration different from Sony's. Our camera was optimized for the film transfer process. I used a quieter, less noisy/grainy setup that was also slightly slower. As with film, you juggle speed and grain. I finally determined the camera/setup speed to be equivalent to approximately EI 200 at -3dB gain, so at +6 it would be 400 and so on.
The data below shows the correlation I found with this setup between IRE and f-stops. Using an EI of 200, I lit a gray card to about 237 foot-candles. That setting gave me mid- gray on the waveform monitor right where it should have been at about 56 IRE - the "crossover" point. Running the lens iris up and down through its full range, I then observed the following:
f2.5 = 100 IRE (max level for good detail and color saturation)
f2.8 = 97 IRE (two stops over mid-gray)
f4.0 = 77 IRE (one stop over mid-gray)
f5.6 = 56 IRE (mid-gray)
f8 = 40 IRE (one stop below mid-gray)
f11 = 27 IRE (2 stops below mid-gray)
f16 = 17 IRE (3 stops below mid-gray)
f22 = 10 IRE (4 stops below mid-gray)
At four stops below mid-gray, separation between mid-gray and black remained discernable, though barely.
The first test "scene" occurred in a living room with a woman sitting in a chair, reading in morning light streaming through a big bay window. The window does not feature any color or density. The only artificial illumination is fill light from a 400-watt Joker HMI bounced off the white plaster wall, just to the right of the lens. The bounced HMI's color temperature ran 4000° Kelvin at the subject. I set the camera's color balance to a preset 3200° Kelvin (similar to tungsten balanced film), used its internal daylight correction filter (similar to an 85), and exposed for the exterior as seen through the window. The camera's internal auto-iris readings and a 40-degree "average" reflectance meter reading put the picture's window area at about f8. An incident reading at the woman's face hit about f4.06. My spot meter readings are as indicated at the numbered points in the picture.
WARM BLOODED KILLERS