David Bordwell's website on cinema   click for CV
    %62or%64%77e%6cl%40%77%69%73c%2e%65%64%75

Home

Blog

Books

Film Art: An Introduction

Christopher Nolan: A Labyrinth of Linkages pdf online

Pandora’s Digital Box: Films, Files, and the Future of Movies pdf online

Planet Hong Kong, second edition pdf online

The Way Hollywood Tells It pdf online

Poetics of Cinema pdf online

Chapter 3 | Three Dimensions of Film Narrative pdf online

Figures Traced In Light

Ozu and the Poetics of Cinema pdf online

Exporting Entertainment: America in the World Film Market 1907–1934 pdf online

Video

CinemaScope: The Modern Miracle You See Without Glasses

How Motion Pictures Became the Movies

Constructive editing in Pickpocket: A video essay

Essays

A Celestial Cinémathèque? or, Film Archives and Me: A Semi-Personal History added September 2014

Shklovsky and His “Monument to a Scientific Error”

Murder Culture: Adventures in 1940s Suspense

The Viewer’s Share: Models of Mind in Explaining Film

Common Sense + Film Theory = Common-Sense Film Theory?

Mad Detective: Doubling Down

The Classical Hollywood Cinema Twenty-Five Years Along

Nordisk and the Tableau Aesthetic

William Cameron Menzies: One Forceful, Impressive Idea

Another Shaw Production: Anamorphic Adventures in Hong Kong

Paolo Gioli’s Vertical Cinema

(Re)Discovering Charles Dekeukeleire

Doing Film History

The Hook: Scene Transitions in Classical Cinema

Anatomy of the Action Picture

Hearing Voices

Preface, Croatian edition, On the History of Film Style

Slavoj Žižek: Say Anything

Film and the Historical Return

Studying Cinema

Articles

Book Reports

Observations on film art

Cutting remarks: On THE GOOD GERMAN, classical style, and the Police Tactical Unit

Wednesday | November 15, 2006   open printable version open printable version

For about ten days Kristin and I are off the grid, unable to check email, let alone blog.

In the meantime we’ll backlog some blogs (backblogs) to be posted at intervals. When we get back, we’ll probably divulge where we went.

This is a followup to my entry on The Good German of some days back. I want to talk a bit about editing technique in classical and contemporary film.

Recall that Dave Kehr’s fine reportage on Soderbergh’s Good German appeared in the New York Times. The film is set in postwar Berlin during the late 1940s, and Soderbergh is trying to film it with the production procedures of the time. He’s using boom microphones and incandescent lights, and he’s asking actors to deliver the lines in the forthright manner characteristic of classic studio acting. He’s using prime lenses of fixed focal length (as opposed to today’s zoom lenses with minutely adjustable focal length). Those prime lenses will often be wide-angle ones (e.g., 32mm), the sort that Soderbergh discovered that Michael Curtiz employed at Warners.

Soderbergh will also avoid multiple-camera shooting, which is common for most scenes nowadays; he will shoot single-camera. This has fascinating implications for staging and editing choices. Dave Kehr explains:

If there is a single word that sums up the difference between filmmaking at the middle of the 20th century and the filmmaking of today, it is “coverage.” Derived from television, it refers to the increasingly common practice of using multiple cameras for a scene (just as television would cover a football game) and having the actors run through a complete sequence in a few different registers. The lighting tends to be bright and diffused, without shadows, which makes it easier for the different cameras to capture matching images.

The advantage for directors is that they no longer need to make hard and fast decisions about where the camera will go for a particular scene or how the performances will be pitched. The idea is to pump as much coverage as possible into the editing room, where the final decisions about what goes where will be made.

The danger for a director is that with so much material available, the original vision may be drowned or never really defined; and the sheer amount of exposed film makes it possible for executives to step in (after the director has completed his union-mandated first cut) and rearrange the material to follow the latest market-research reports.

During the studio era it was more typical for directors to arrive on the set, block out their shots and light them with the use of stand-ins; the actors were then summoned from their dressing rooms and, after a brief rehearsal, they would film the lines needed in the individual shot. The crew would then break down the camera and move it to the next setup, as determined by the director.

“That kind of staging is a lost art,” Mr. Soderbergh said, “which is too bad. The reason they no longer work that way is because it means making choices, real choices, and sticking to them. It means shooting things in a way that basically only cut together in one order. That’s not what people do now. They want all the options they can get in the editing room.”

While the editing process now routinely takes months, “I had a pretty polished cut of ‘The Good German’ two days after we wrapped,” Mr. Soderbergh said. “It was shot to go together very, very specifically.”

Let’s start with the concept of coverage.

Shooting to Protect

Consider a traditional conversation scene.

In the classical Hollywood style, most such scenes were shot with a single camera. (In the silent years, big films would be shot with a second camera to provide a negative for foreign markets. This means that in some films, the overseas versions were slightly different, shot for shot, from their US counterparts.)

Your conversation scene will be edited. So how do you film it? Working with only one camera, directors developed procedures for “covering” a scene.

Here’s the default. The entire scene would be played out in a single, distant take, usually called the master shot (sometimes “master scene”–for a long time, filmmakers confusingly called shots “scenes”). The master shot provided an uninterrupted record of the whole scene, and thus served as a reference point for overall set geography, where actors were standing, and so on.

Once the master shot was finished, the filmmakers went on to break the scene down into closer positions. That meant repeating portions of the scene already played in the master. They shot the medium two-shots of the actors talking, the shots of each actor taken from over the shoulder of the other actor, the singles of each player–which might be medium-shots, medium close-ups, or close-ups.

(Sometimes the filmmakers also shot inserts, the detail shots of objects or printed matter that would be inserted into the action, but usually they were filmed another time, or by an entirely different crew under the supervision of an assistant or second-unit director. Often the hands we see in an insert close-up aren’t those of the actor.)

For each different camera position, the actor would have to repeat the relevant bit of dialogue, gesture, or whatever that had already been played out in the master. Must have gotten boring, no?

The question arises: How much coverage do you shoot? Do you have to film from all those angles, and do you have to repeat everything to the max?

One answer says yes. Overshoot. This is the shoot and protect option. Film lots more footage than you can possibly ever use. Film is cheap, and you want lots of options in the editing room. Since in the studio system, directors rarely supervised the editing of their films, the editor worked with the producer, and the producer wanted to readjust the film in the cutting phase–inserting a close-up of the star, tightening the pace, and so on.

Shoot and protect is an insurance policy, and producers still insist on it. In The Way Hollywood Tells It, I quote Swedish cinematographer Sven Nykvist’s remark that American directors shot a lot more coverage than European ones. “I believe that comes from the fact that producers usually have the final cut and they want to have all the material they can get.” In the studio era, producers might want changes and demand reshoots of portions of scenes so the sequence would cut together better.

During the studio era, the most famous director who relied on overshooting is George Stevens. In an interview with Kristin and me, William Hornbeck talked about how Stevens was maniacal about reshooting the scene from every angle he could think of. Then he would run the footage over and over, trying to decide what takes to use. One result of this is a fairly choppy style. The barroom fight in Shane is almost wholly created in the editing room, and some of the cuts are downright clumsy.

Or take the barbecue scene of Giant, where the cuts struggle to extract a throughline of action. Elizabeth Taylor stands on the edge of the crowd, and abruptly Stevens dissolves to an extreme long shot of the festivities.

giant-1.jpg giant-2.jpg

So is the scene ending, as such a reestablishing shot might suggest?
No, because we dissolve again to her, now walking to the tree in the foreground, captured in a high angle from a crane.

giant-3.jpg giant-4.jpg

Most directors would have moved Elizabeth Taylor away from the crowd in a single tracking shot, letting the camera follow her approach to the tree. Stevens apparently hadn’t planned or filmed such a shot, so he had to use two dissolves, although there doesn’t seem to be a time lapse. Once Liz gets to the tree, the high angle obscures her face and gesture, so a cut tries to clarify:

giant-5.jpg giant-6.jpg
Stevens cuts to a head-on shot of her starting to swing around the tree. But she doesn’t get a chance to complete this gesture because the tree blocks her face and arms. So Stevens cuts away to James Dean, already walking leftward to the car–another interrupted action.

giant-7.jpg giant-8.jpg

Dean settles down to watch the barbecue from a distance, but instead of getting anything resembling a point-of-view shot, we go back to the uninformative high angle, in which a cowpoke offers Taylor a drink.

giant-9.jpg giant-10.jpg

We can’t really see the first man’s gesture, or her reaction, so the dialogue has to carry the whole scene. And this action is interrupted again by another man proposing a toast to the couple. The high angle adds nothing to the scene. Didn’t Stevens film this bit of action from a more straightforward setup? If not, maybe he didn’t protect himself. Whatever happened, this passage seems to be very awkward filmmaking.

Granted, the cuts between Taylor and Dean do parallel them as outsiders and prefigure their alliance later in the film. But the same points could be made with smoother cuts and greater integrity of gesture. The scene’s fragmentation prefigures today’s interruptive editing.

Someone might argue that the jumpy cutting conveys something expressive about the scene (maybe Taylor’s anxiety about making a mistake in front of her new husband’s friends). But this is an all-purpose alibi, which can be invoked to defend any case of poor craftsmanship. My shot’s out of focus because…um…the hero’s uncertain of his identity. Moreover, if we defend the cutting as projecting Taylor’s anxiety, we’re tacitly granting that the scene has failed to convey that through performance, staging, and other techniques.

Lest it seem I’m being too mean to Stevens, I’ll just point out that occasionally he took some chances, even using single-shot high-angle staging reminiscent of Mizoguchi. Here’s an example from A Place in the Sun.

place-1.jpg place2.jpg place-3.jpg

Shooting Cut to Cut

“Shooting for protection” sounds a bit defensive and wussy, like you’re afraid to make a choice. Suppose you try another tactic. Suppose you shoot only what you think you will finally cut together.

You might still shoot the master shot for insurance, but maybe not. On the whole, you’d shoot only the angles you’d decided on in advance, or perhaps you would add a few pieces for protection. Basically there would be only one way to cut the scene together. This was called filming “cut to cut” or “cutting in the camera.”

Hitchcock is the most famous exponent of this tactic, thanks largely to his claims that he preplanned everything that appeared on screen. After working out the script, shot lists, and storyboards, he knew exactly what he wanted, he said, and so the actual filming was a little tedious. I think he once remarked that he wished he could phone a film in. In truth, Sir Alfred exaggerated the precision of his planning. Bill Krohn’s excellent book Hitchcock at Work debunks “the myth of the storyboard” (which were often made after filming, to show reporters) and Krohn offers several examples of set improvisation, last-minute changes, tightening during editing, and so on (pp. 9-16). Nevertheless, on the whole Hitch probably shot significantly less coverage than most directors.

So too, evidently, did John Ford. Ford didn’t pre-plan his films to the minute degree Hitchcock claimed, but he achieved enough power in the system to avoid the shoot-and-protect method. According to editor Elmo Williams, whom I once asked about this, Ford had the reputation of giving his cutters almost no range of setups to choose from. If you give ‘em a close-up, he supposedly said, they’ll use it. Ford apparently kept track of all the shots he wanted in his head, which may explain some of the mismatches that we find in his films. But it did mean that when the shooting was finished, he could hand the footage over to an editor and go out on his boat.

Another director who filmed coverage sparingly was W. S. Van Dyke, known as Wun-Shot Woody. You can read more about him here.

So there are degrees of coverage, from the maximal coverage of shoot-and-protect to the minimalism of cutting in the camera. Of course you can avoid coverage altogether by shooting your scene in a single take, from a single position (as Hou Hsiao-hsien often does) or with a moving camera. Producers hate this. Christine Vachon calls it “a ‘macho’ style that leaves no way of changing pacing or helping unsteady performances” (The Way Hollywood Tells It, 153). The director who doesn’t provide extensive coverage is a producer’s nightmare.

Camera A + Camera B + Camera C…+Camera N = ?

What about shooting the scene with more than one camera? Before the 1970s, multiple cameras were used chiefly for action sequences, and especially for things that couldn’t be repeated, like an explosion or cars hurtling off a cliff. Multiple cameras were also used for a brief period when sound first came in; watch nearly any film from 1929-1931 on Turner Classic Movies, and you’ll see scenes filmed much as TV sitcoms and soaps are filmed today, with several cameras. (For more on this, see The Classical Hollywood Cinema, Chapter 23.)

On the whole, though, multiple cameras were rarely used until some time in the early 1980s, when shooting schedules were accelerated. Faced with the need to turn out more footage quickly, directors began using “B” cameras to pick up alternate angles. Now a big picture may have “C” camera plus a Steadicam roaming around to grab details. One of my favorite quotes (in The Way, p. 154) is from John Mathieson, DP for Gladiator, who after explaining that by using seven cameras for some scenes, “I was thinking, ‘Someone has got to be getting something good.’” The old Hollywood directors, thinking through the scene shot by shot, knew when they were getting something good; now everybody just hopes.

In effect, as Dave Kehr’s article implies, multiple-camera shooting is today’s equivalent of classic coverage. Soderbergh is right to note that this practice postpones decisions until the editing stage. But it has other knock-on effects on the way the movie looks. I’ll consider just two, which I expand on in The Way.

1) Having reams of footage tends to speed up the cutting rate. Rather than sticking with a few angles, directors offered many choices tend to like a little bit of angle 1, the middle part of angle 2, another stretch of angle 3, and the last bit of angle 4. Four shots arise in the editing process, when maybe a single well-chosen one would do the job.

There are many sources of the fast cutting that rules contemporary cinema, but probably the practice of shooting with A, B, C, and N cameras is one.

2) Multiple camera shooting limits the freedom of camera position that single-camera shooting allows. Obviously, you can’t put Camera B in a position where it will be seen by camera A. In fact, you tend to keep all the cameras at a fair distance from the scene, using long lenses to enlarge the actors.

The implication is that the cameras really don’t work their way into the space as much as they do in single-camera shooting. You can’t put the camera between characters as easily, so Ozu’s characteristic frontal framings aren’t as feasible. Or, going back to Soderbergh’s beloved 1940s, take the variety of angles we find in the conversations between Spade and Gutman in The Maltese Falcon. These can’t easily be replicated with multi-cam staging.

Medium two-shots are also problematic with multiple cameras. (Hence perhaps all the singles and close-ups we get in contemporary films.) And all cinematographers grant that lighting an overall scene so that it looks right from several camera angles requires many compromises.

So I look forward to seeing how, in an age when single-camera coverage is rare, Soderbergh will handle these matters. Will The Good German‘s cutting rate be slower than is common nowadays? (The 1940s and early 1950s saw the widest disparity in average shot lengths in Hollywood’s history, from very short shots to ASLs over thirty seconds.) Will the camera penetrate the space in the manner of the ubiquitous setups of the studio years?

Finally, I’d just like to note that Soderbergh didn’t have to return to the 1940s for inspiration. Classic techniques of coverage, even cutting in the camera, are amply on display in other cinemas.

Mackendrick’s Axiom

Today’s choppy editing style is interruptive rather than integrative. A typical director today doesn’t let a shot ripen to full implication. The cut interrupts the shot rather than coming at the moment that will carry forward what’s been developing in the image.

The British director Alexander Mackendrick once suggested that at each moment the director must prepare the audience for what is going to happen next. If you’re really good, shot A prepares not only for B but for C and D as well. You can study this process in Lang, Ford, Hawks, and other classical directors, but it’s still visible today, I think, if we know where to look.

Yes, the answer is Asian cinema again. Hong Kong, no less.

Johnnie To Kei-fung began his career in television in the 1970s, made a feature in 1980, and returned to shooting TV series in the early 1980s. He shot single-camera cop shows, martial-arts shows, and other dramatic shows. No shot lists, no storyboards–no time! Given the brisk schedules of TV shooting, To mastered the ability to film one shot while thinking about the next three, the way a chess master conceives a string of moves. He makes the determining decisions about the sequence in the flow of production, then fine-tuning them in editing.

Johnnie To shifted to feature filmmaking in 1986, and he has turned out more than forty films since. Sometimes he finishes three films in a single year. His output includes many extraordinary achievements, such as Lifeline (1997), The Longest Nite (1998), Expect the Unexpected (1998), A Hero Never Dies (1998), The Mission (1999), Throw Down (2004) and several others. His two films about Triad life, Election 1 (2005) and Election 2 (2006) have received wide attention, as has his most recent release, Exiled. These last three have been picked up for US theatrical release.

His style can vary a lot from film to film, from the frenetic action of A Hero Never Dies to the hairtrigger poise of The Mission. But in each case you detect a director who makes his stylistic choices and sticks to them. He can structure a scene in his head and he can shoot the result efficiently. He is, we might say, today’s Woody Van Dyke. Except he’s quite a bit better.

Here’s a sequence that exemplifies Mackendrick’s axiom.

During the climactic gunfight of PTU (2003), a female police officer fires at a gunman. He fires back, and she drops her pistol as she dives back into her car. Then the cop Lo, the hapless slob we’ve been following, seeing the gunman about to fire, dashes out of the shot. This is followed by a shot of the gunman swiveling and firing.

ptu-2.jpg ptu-3.jpg

The next shot is the crucial one. Through a rain-speckled car windshield we see Lo fleeing.
ptu-4.jpg

Why shoot from this fancy angle? It does two things at once. It tells us about the fleeing cop’s progress, continuing the action of the earlier shot of him, but it also puts us in the car, implying that the next phase of the action will take place here, rather than on the street. And so it does:

ptu5.jpg ptu-5.jpg

The wounded gunman staggers against the car, and again, instead of seeing his progress from the outside, we’re in the car. We don’t need to see his face, only the gun: director To films the danger. This shot expands on the previous one, keeping us in the front seat, and then gives us new information. The camera tilts diagonally to the policewoman crouching below the window.

The suspense is heightened when To cuts back to the pistol, bumping along the glass.

ptu-6.jpg

Will the gunman look inside the car? The movement of his arm is continued smoothly in a new angle:

ptu-7.jpg
….as he rushes off to the right in pursuit of Lo. Director To gives us both action and reaction in the same concise shot. The policewoman sobs in relief, and this sub-sequence ends with a pause on her.

The sequence has broader implications as well, since this officer has been suspecting all along that Lo has lost his pistol (which he has). Now she has lost hers. After the smoke has cleared, Lo strolls back to her, and the shot echoes the earlier shots taken through the windshield.

ptu-9.jpg

The irony dawns on her–in the heat of battle, it’s easy to lose your weapon–as she glances down at her discarded gun and retrieves it.

ptu-10.jpg ptu-11.jpg
This sequence couldn’t have attained such precision if it were filmed with several cameras snatching shots, hoping to get something; nor with shoot-and-protect coverage. The passage is conceived cut to cut, in the director’s head. Each shot makes one or more points succinctly and leads us crisply to the next bit of action. When I asked Mr. To about the shot of Lo fleeing, seen through the windshield, he seemed surprised that I’d wonder. “Of course! You have to let people know they’ll see her in the car.”

Comments are closed.

David Bordwell
top of page

comments about the state of this website go to Meg Hamel.