American Cinematographer - July 2010 - 72

Filmmakers’ Forum
I
Capturing an Episodic Drama With a DSLR By Gale Tattersall Some filmmakers tend to think of their equipment as the starting point, but when executive producer/director Greg Yaitanes and I began to plan the finale for the sixth season of House, M.D., we started with the script. According to that blueprint, a lot of the action would take place in a collapsed building, which meant that in order to film it, the crew would be spending about five days on their hands and knees and bellies on sets that were about 2½' tall. Our production designer, Jeremy Cassells, had designed the sets so that everything could move away and float up, but we realized very early on that the time to do this shot by shot would kill us unless we worked with small cameras. I had shot some commercials for Canon Japan on the Universal Studios backlot, and the clients asked us to shoot the spots with the 5D Mark II DSLR. I thought they were absolutely nuts. How could you photograph a serious commercial with a tiny stills camera? But I did some tests, and the results were actually quite beautiful. When I went back on House, I showed Greg what I’d done with the 5D, and we went through some fairly extensive tests before making the decision to use it on House. At first, we decided to use it on just a few scenes in the episode “Lockdown,” which involved a newborn baby. (The rules for filming infants are incredibly strict, and the idea of putting a 60-pound camera with a 50-pound geared head over the baby seemed ridiculous.) Using the 5D for those scenes worked well, and Greg suggested we shoot an entire episode with it. We’d shot all six seasons of House on 3perf Super 35mm, although we’d captured some shots here and there with digital cameras. We’d used the Panasonic HVX-200 for portability in our operating-room set, and we had also recently used the Weisscam HS-2 to capture a particular shot at 2,000 fps. Committing to shooting an entire episode digitally for the first time would be difficult, but every piece of equipment has its advantages and disadvantages — it’s just a matter of going with the things that are wonderful and avoiding the areas of weakness. One of the things I discovered during testing was that the 5D
72 July 2010

“I’ll be surprised if many television shows don’t add a 5D, 7D (or the newer 1D MK IV) to their toolkits.”

brings something unique to the table: a largeformat sensor that lets you play with a very shallow depth-of-field. I would argue that if you shot on 70mm film in the 16x9 aspect ratio, you still couldn’t achieve less depth-offield than with a wide-open T1.2 or T1.4 lens on a 5D. The script for the season finale was a very emotional one; it’s all about what’s going on in House’s head, so we really wanted to separate him from reality. On our hospital sets, a lot of distracting architectural elements vie for attention in the frame, and with the 5D, we could throw backgrounds completely out of focus, putting House (played by Hugh Laurie) in his own world. It’s an incredible look, but it created a nightmarish problem with focus-pulling. Part of the problem is the shallow depth-of-field — a Canon EF 50mm T1.0 lens has almost none. We did a lot of close-ups, and sometimes focusing on one eye would make the other go soft, and the end of the actor’s nose would be completely gone. Also, Canon lenses were made for still photography, so the barrel rotation between minimum focus and infinity is incredibly small — just a 2mm movement of the focus ring might make a focus difference of 10" in the action zone. If I hadn’t had such great focus pullers, I would have been in serious trouble. Don Carlson, my A-camera focus puller, opted to completely separate himself from video village and the camera and just eyeball everything off a 7" Marshall monitor, which has an edge peaking mode that displays the focus differential of a given shot as a colored highlight. We used BarTech remote-focus units, which were well-suited to the Canon lenses; the remote-focus units we use with our full-sized cinema lenses proved too powerful. (We rigged everything using Redrock Micro support equipment.) It tends to complicate things when the cameras are moving, the actors are moving, and you have to calculate a 1" plane of focus from 10' away, so we tried to minimize camera moves. Having seen the rehearsals, Don knew which way the actors were likely to move. In our tests, we also encountered the banding and aliasing caused by the 5D’s highly compressed 8-bit H.264 recording system. Banding breaks up tonal separations into distinctive gradations, but we found that if we added a tiny bit of grain in post — something

American Cinematographer

American Cinematographer - July 2010

Table of Contents for the Digital Edition of American Cinematographer - July 2010