DB here, still in Hong Kong:

You are a film director. How do you convey to an audience that a character mysteriously sees hidden personalities in other characters?

Filmmakers are problem-solvers, at least sometimes, and I’ve argued elsewhere that we can often explain their creative choices as efforts to pose and solve problems. I ran across an intriguing example during my stay here in Hong Kong. I interviewed Tina Baz, the editor for Johnnie To’s film Mad Detective (2007), which I blogged about from Vancouver in October.

Tina has a fascinating history. Lebanese and French-educated, she attended art school in Beirut and film school in Paris. Armed with a maitrise from the Sorbonne, she began editing. Over a dozen years, she has built up an impressive career. She was editor on The Mourning Forest (2007), A Perfect Day (2005), and several other high-profile projects.

Mad Detective was Tina’s first film for Johnnie To, as well as her first experience with the Hong Kong crime genre. She was accustomed, she says, to the more psychologically-driven narratives of French cinema, and she had to work in a new way during the seven weeks it took to edit the film. This is an unusually long editing time for Hong Kong filmmaking, and it reflects the complicated tasks this film set her.

If you haven’t seen Mad Detective, you will probably want to stop reading. The scenes I’m discussing come fairly early in the film, but they do contain some surprises you might not want to know about in advance.

Headquarters

The rough cut of the film, adhering closely to Wai Ka-fai’s script, displayed a complicated flashback structure. The plot began fairly far into the action and then jumped back to establish the exposition. Central to that exposition is the film’s premise: Detective Bun, played by Lau Ching-wan, solves cases by extraordinary telepathic/ intuitive/ visionary insights. But the original cut blurred that concept, mixing it with an early revelation of the identity of the villain, a corrupt cop named Ko Chi-wai, and background on Bun’s young partner Ho. How to cut through the clutter?

The first step to a solution was to straighten out the chronology and to focus tightly on Bun from the start. The audience needed to understand the film’s eccentric psychological premise—specifically, that Bun’s gift allows him to detect multiple personalities buried within another person.

In talking with me, Tina went over three scenes that establish Bun’s peculiar abilities. (1) The sequences furnish striking examples of how filmmakers find precise technical solutions to storytelling problems. This trio of scenes also follow Hollywood’s Rule of Three. According to this, a key piece of information should be presented three times, preferably in different ways. (2) (Once for the smart people, once for the average people, and once for slow Joe in the back row.) Tina smiled. “This is not something that French directors would say.” The Rule of Three can be seen as another aid to solving the broad problem of clarity—making sure that the audience understands the story.

Discussing all three scenes adequately would take a lot of space and require dozens of stills, so I’ll concentrate on the first two. These should suffice to show how the filmmakers solved a tricky narrational problem.

The first scene takes place in police headquarters and it’s the simplest. It starts with Bun absorbing details of Ho’s investigation. We hear, offscreen, a whining woman complaining that a maniac can’t solve a case. We think at first that it’s the voice of another cop, but after a second complaint from offscreen, Bun looks up. Cut to the whining woman, who stands among the cops.

Bun rises and walks to a pudgy cop who grins ingratiatingly. Now cut from the cop to the whining woman, in a similar composition against the same background.

An over-the-shoulder shot presents Bun staring as we hear the fat cop asking, “Do you recognize me?”

After the fat cop flatters him, Bun head-butts him. Bending over the fallen cop, Bun says, “Shut up. I see and hear you, bitch!” Cut from the bloody-nosed cop to Bun to the bloody-nosed woman.

After Ho dismisses the team, Bun slowly approaches and confides: “I can see the inner personalities of a person” (below). This line of dialogue crystallizes what we’ve seen–the bitchy woman inside the ingratiating man–and prepares for the next few scenes.

On the street

The second vision scene is more complicated, and Tina had several discussions with To and Wai about whether to include it. The new problem, based on the key line above, is to establish that Bun can see not one but several hidden personalities in Ko, the crooked cop. How can this premise be shown?

The major scene coming up shows the two partners confronting Ko at a restaurant. There Bun will envision some of Ko’s inner people manifesting themselves. But there’s also dramatic conflict to be conveyed in the scene, which culminates in a beating in the men’s room. Again, too much information! Tina, To, and their colleagues decided to include an intermediate scene, one that had little dramatic consequence but got audiences more comfortable with the premise.

Ho and Bun trail the suspect along the street, first in a car and then on foot. The results have the kind of precision I’ve pointed to in another Milkyway film, PTU.

In the car, Tina’s cutting establishes the differences between the two cops’ points of view. A two-shot of Ho and Bun yields to a straightforward POV shot of Ko striding along the sidewalk.

But a cut to Bun watching is followed by a shot of seven people in Ko’s place. In a nice instant of uncertainty, a few heads emerge from behind a parked van and are only partially visible. At first glance we might take them for passersby walking ahead of Ko.

Before we can fully identify this band of spirits, and as the car draws abreast, a cut to Ho watching leads to a shot of Ko, whistling.

Another cut to Bun, and now we see, as he apparently does, a full view of the platoon of Ko-personalities striding along whistling.

This mildly wacko variant is consistent with the narration’s mixed attitude toward Bun’s visions: are they supernatural insights or hallucinations?

Bun leaps from the car and pursues Ko’s band of personalities, and a new variation shows him and them in the same frame. But this is re-framed by Ho’s POV, which reasserts that to anyone but Bun, Ko is all by himself.

Abruptly, the visions halt and turn, staring at Bun. By now we are trained to understand what’s happening without a corresponding objective shot of Ko. Ko, we infer, has turned around to check on who’s following him.

Likewise, we can grasp the image of Bun passing through the group of figures as simply Bun walking past Ko.

At this point, we have no need to see the objective action. To and Tina have stressed the absence of an objective view by showing Ho, in the background, looking in another direction and missing this byplay (above). After Bun has passed, the troupe of avatars walks in their new direction, and then we get the objective shot of them “as” Ko.

This effect is anchored when Ho turns and notices Ko going into the restaurant.

As a meticulous touch, Ko’s departure from the frame reveals Bun still walking away in the distance. The geography is simple and impeccable.

The idea of seeing “inner personalities” has been made concrete while undergoing a lot of variation. Sometimes shots of the spirits are sandwiched between shots of Bun looking; sometimes not. Sometimes we have Ho to verify what’s really happening, but sometimes not. And sometimes Bun can be in the same frame as the avatars. Otherwise, Tina points out, the scenes would become boring.

Once the street scene has laid out how Bun sees Ko Chi-wai, the next scene, unfolding in the restaurant and a men’s toilet, can play with the premise still more. Tina explains that the creative team constantly asked, “How far can we go with Bun’s point of view, his vision?” Quite far, it turns out. The premise gets varied further across the whole movie, in imaginative ways. Which is to say that a solution to a problem can pose new problems, which the filmmakers go on to grapple with.

In our discussion, Tina noted that the street scene goes beyond simply giving us information. There’s a disquieting emotional effect when the spirits turn and confront Bun, their expressions ominously blank. Finding a solution to a creative problem often yields an unexpected payoff like this.

Bonus features

As I indicated, I can’t pursue the felicities of the restaurant scene here. Instead, here are some quick final thoughts.

*Mad Detective shows once more the power of basic techniques of performance (e.g., the act of looking), framing, and editing. To and Wai have no need for fancy CGI effects to evoke the “inner personalities” of Ko or the other cop. Our minds make the necessary connections. Lev Kuleshov would have admired the engineering economy of this sequence.

*In passing Tina and I talked about actors’ eyes, a special interest of mine. (3) Tina had noticed that actors try not to blink. She agreed that blinks pose problems for editors: “Blinking is very hard on cutting.” She remarked that Elia Suleiman would strain to keep from blinking until tears flowed from his eyes.

*Mad Detective was signed by both Johnnie To and Wai Ka-fai, but their joint projects are usually filmed by To, with Wai supplying the script and consulting as production proceeds. Tina confirms To’s habit of shooting with the whole scene’s layout in his head, using very few takes per shot and not a lot of coverage. The film is “very pre-cut,” she says. You can see this, I think, in the way To relies on matches on action and ends one shot with something that prepares you for the next one.

*To follows Hong Kong tradition of adding all the sound in postproduction. This makes for economy and efficiency in shooting, but it also gives HK films the fluency of silent cinema. The filmmaker gains a freedom of camera placement and is encouraged to think about how to tell the story visually. Tina and Shan Ding, To’s all-around assistant, pointed out that sync sound might make the director depend more on dialogue and static conversation scenes. Still, I think that sometimes they wish for a scene shot with direct sound.

*As with many Milkyway films, several scenes were filmed at the company headquarters. An earlier blog here discussed the ambitious rooftop set for Exiled. For Mad Detective, the fistfight in the restaurant toilet was filmed in the main men’s room of the Milkyway building, with a fake urinal added. Shan Ding provides documentation in the snapshot below, using a double for Lam Suet.

To come: Another Milkyway interview, this one with ace sound designer Martin Chappell. And a final batch of impressions from the Hong Kong International Film Festival, which ends on Sunday.

(1) If you know Mad Detective, you’re aware that Bun’s ability to discern hidden personalities is displayed in an earlier scene. But given the way in which that story action is presented, we can’t say that that his powers are established there. This scene is one of many clever narrational stratagems in a film that benefits from being rewatched.

(2) See Bordwell, Janet Staiger, and Kristin Thompson, The Classical Hollywood Cinema: Film Style and Mode of Production to 1960 (New York: Columbia University Press, 1985), 31.

(3) Laugh if you must, but I wrote an essay about this. “Who Blinked First?” is available in Poetics of Cinema.

DB here:

In Number, Please? (1920) Harold Lloyd plays a lovesick boy who’s been jilted by his girl. Moping at an amusement park, he sees her arrive with a new beau.

He shifts to another spot to watch them. When she notices him, she scorns him, and he reacts.

She and the escort stroll past, then she turns and cuddles up to him, making sure Harold notices.

Her flirting precipitates the rest of the action in this very funny short.

In this scene from Number, Please? Harold and the couple aren’t shown in the same frame. The action is built entirely out of singles of Harold and two-shots of the couple, with an especially emphasized close-up of the girl’s snooty reaction. The sequence is rapidly paced and carefully matched in angle; note the shift in eyelines as the girl and her beau walk a little way and then she looks back at Harold.

This approach to building a scene was consolidated in American studio cinema in the late 1910s, as we noted recently, and it soon spread around the world. One of the places it caught on most fervently was Soviet Russia.

 

Kuleshov glories in the gaps

The great director and theorist Lev Kuleshov always claimed that he and his associates learned the power of editing from American cinema. Russian films were slowly paced, consisting of long tableaus occasionally broken by an inserted closer view of an actor. (For examples, see my Bauer blog entry from the summer.) Kuleshov noted that the Hollywood films brought into Russia grabbed audiences’ interest more intently, and Kuleshov attributed this effect to the fact that the Americans exploited editing more fully, creating the scene out of many shots.

Kuleshov’s example was the formulaic scene of a man sitting at his desk and deciding to commit suicide. The Russians, Kuleshov claimed, would handle this all in one distant framing, with the result that the key actions were just part of the overall view. By contrast, Americans would shoot the scene in a series of close-ups: the man’s face, his hand taking a pistol out of a desk drawer, his finger tightening on the trigger, and so on. This gave the scene a powerful concreteness, and was cheaper to film besides (no need to have a full set).

But most American filmmakers didn’t create the scene wholly out of close-ups. Typically there would be an establishing shot before the action was broken down into detail shots. The process has come to be called analytical editing. (We discuss it in Chapter 6 of Film Art.) As the label implies, the cutting analyzes an orienting view into its important details.

Less commonly, as in our Number, Please? example, American directors could create a scene entirely out of separate areas of space, without ever showing a master framing. This technique, usually called constructive editing, remains common today as well, especially in action scenes.

While praising analytical editing, Kuleshov was particularly taken with constructive editing, because that shows that cinema can call on the spectator’s tacit understanding to assemble the separate shots. Kuleshov realized that we will build a sense of the scene’s space and action out of separate shots without need for the comprehensive view supplied by an establishing shot.

What the Americans developed, the Soviets thought seriously about. Around 1921, Kuleshov and his students mounted some experiments, several of which he discusses in his books and essays. He probably didn’t need to experiment; the American films were full of examples. Indeed, the Number Please? passage is more intricate than any experiment Kuleshov seems to have mounted. But he had a bit of the engineer about him, and he sought to break the technique into its simplest components.

For one thing, constructive editing offered production efficiencies. You could film two actors separately, at different times, and then cut them together. Further, Kuleshov saw that editing could abolish real-world constraints. It created events that existed only on the screen, with an assist from the viewer’s mind.

This is perhaps best seen in his experiment involving an “artificial person.” Evidently it wasn’t a case of constructive editing, because it seems to have begun with an establishing shot. The first shot shows a girl sitting at her vanity table putting on makeup and slippers. A series of close-ups of lips, hands, legs, and the like were derived from different women, and they were edited together to create the impression of a single woman. (Something of this effect survives in the idea of a body double in contemporary films and TV commercials.) But the point is that the woman on screen, made out of different parts, doesn’t exist in the real world.

The same possibility could apply to geography, if we delete the establishing shot. As Kuleshov describes his experiment, we initially get a shot of a woman walking along a Moscow street. She stops and waves, looking offscreen. Cut to a man on a street that is in actuality two miles away. He smiles at her and they meet in yet a third location, shaking hands. Then together they look offscreen; cut to the Capitol in Washington DC. Kuleshov saw the potential for imaginary geography as both a useful production procedure and a demonstration that editing could create a purely cinematic space, one not beholden to reality.

Kuleshov’s most famous experiment, the one he identified with the “Kuleshov effect” proper, involves a stock shot of the actor Ivan Mosjoukin, taken from an existing film. In his writing he’s rather vague and laconic about the results.

I alternated the same shot of Mosjoukin with various other shots (a plate of soup, a girl, a child’s coffin), and these shots acquired a different meaning. The discovery stunned me—so convinced was I of the enormous power of montage. (1)

Kuleshov’s pupil the great director V. I. Pudovkin offers a different description of the shots: a plate of soup, a dead woman in a coffin, a little girl playing with a teddy bear. He also goes farther in reporting how the audience responded. People read emotions into the neutral expression on Mosjoukin’s face.

The audience raved about the actor’s refined acting. They pointed out his weighted pensiveness over the forgotten soup. They were touched by the profound sorrow in his eyes as he looked upon the dead woman, and admired the light, happy smile with which he feasted his eyes upon the girl at play. But we knew that in all three cases the face was exactly the same. (2)

Now it isn’t only geography or a human body that has been created by editing; it’s an emotion.

These experiments have been poorly documented, and only a couple have survived. One consists of fragments of the imaginary geography exercise. Here is Alexandra Khokhlova waving, but we don’t have the answering shot of the male actor responding. The two meet at the foot of a statue.

After the two shake hands, they look up and out of the frame, but unfortunately we lack the shot of the Capitol.

Kristin and other scholars have written more about these surviving fragments, and their essays are published along with Kuleshov’s proposal for funding the experiments and his wife Khokhlova’s memoir of filming them. (3)

It’s worth taking these prototypes of constructive editing apart a little bit. Clearly, there are several cues that prompt us to see the shots as continuous.

One cue is a common background, or at least a consistent one. Kuleshov thought that sometimes a neutral black background worked best, especially for close shots, as you can see with the handshake shot. Another cue is roughly consistent lighting from shot to shot. Yet another is the presumption of temporal continuity; no moments seem to be omitted in the cut from shot A to shot B. It never occurs to us to consider that Kuleshov’s man is looking at something hours or days before the soup is laid on the table in the second shot.

One of the most important cues goes unmentioned by Kuleshov: the very act of looking. Like most commentators, he seems to take it for granted. Yet it’s crucial in prompting us to imagine an overall space in which the actions take place. Knowing the real world as we do, we can infer that if you’re close enough to watch someone, both people are probably in a shared space, such as the arcade strip in the amusement park in Number, Please?

Another cue is facial expression. In his soup/girl/coffin sequence, Kuleshov supposedly picked a shot of Mosjoukin that doesn’t have a clearly identifiable expression. If Mosjoukin was smiling in his interpolated shot, he would presumably seem not grieving but wicked. Normally, though, performers seen in the single shot are expected to express the appropriate emotion more fully, in order to specify what we take the characters’ mental states to be. Our sequence from Number, Please? makes sure we understand the drama by giving the actors unambiguous expressions.

Finally, in the Kuleshov prototypes each shot should be simple. Its action can be stated in a brief sentence. A woman waves. A man responds. A man looks. A plate of soup sits on a table. Even in Number, Please? we can summarize each shot: Harold looks off. His former girlfriend disdains him. That’s to say that the shots are composed to present only one, quickly grasped point of interest.

Filmmakers don’t need to tease apart all these cues; they use them intuitively. Soon after Number, Please?, Hollywood filmmakers were creating amazing passages of eyeline-match editing—the most virtuoso being probably the racetrack sequence in Lubitsch’s Lady Windermere’s Fan (1926). Within a few years of Kuleshov’s experiments, Soviet filmmakers were creating their own masterworks, like Battleship Potemkin (1925) and Kuleshov’s By the Law (1926). Benefiting from a very compressed learning curve, filmmakers took constructive editing to new heights.

 

Constructive editing, dissolved relationships

Sometimes constructive editing is used to save a scene when production goes astray. When Doug Liman reshot the climax of The Bourne Identity, Julia Stiles couldn’t be on set, so singles of her taken from the first version were wedged in among the retakes, to create the impression that she was watching Jason confront his controller. More positively, a carefully calibrated constructive editing is central to a sequence I analyzed a while back in In the City of Sylvia. For over 100 shots, the spatial relations are built up without an overall establishing shot. (4)

Constructive editing can be used systematically throughout a film. A good example is Alan J. Pakula’s thriller Presumed Innocent (1990). Spoilers coming up!

Harrison Ford plays Rusty Sabitch, a prosecutor in the District Attorney’s office who becomes infatuated with a new woman on the staff. He has a brief affair with her, but after she’s broken it off she’s found brutally murdered. He has to investigate the crime without involving himself, but eventually he becomes the prime suspect.

At the start of the film before Rusty learns of the murder, Rusty and his wife Barbara are shown at breakfast, and establishing shots bring them together.

At the office Rusty learns of Carolyn’s death, and after he comes home, the conversation between Rusty and Barbara is treated without an establishing shot. Barbara knows about the past affair, and Rusty is wracked by guilt and shame. The cutting seems to reflect the fact that Carolyn’s death has revived the pain in their marriage.

In a series of flashbacks, Rusty relives his affair with Carolyn. Pakula treats their early encounters by means of constructive editing. The common-background cue is especially helpful in this scene in a kindergarten.

Only after Carolyn and Rusty cooperate and win a child-abuse case does the cutting’s rationale become clear. Pakula has saved the traditional two-shot framing for their moment of passion, as they make furious love in the office.

But soon their affair wanes and Rusty is reduced to watching Carolyn from across the street in point-of-view shots.

After the flashbacks end, Rusty is investigated and eventually charged with Carolyn’s murder. In these scenes Pakula often situates Rusty and Barbara in the same frame, as if the threat to him has healed the breach between them.

Yet in the film’s climax—and here is the big spoiler—they are pulled apart again. The last scene is a sustained monologue by Barbara. As Rusty listens, across twenty-six shots the two are never shown in an establishing shot.

Contrary to the standard Hollywood ending, in which the loving couple embrace in reunion, here they are left divided.

Pakula’s use of constructive editing has effectively traced two arcs: the growth and dissolution of Rusty’s affair, and the reuniting and dissolution of his marriage. In such ways, what might seem a purely local effect, confined to a handful of shots, can create stylistic patterning across a film. The judicious use of constructive editing matches the dramatic development.

 

Godard of the gaps

Robert Bresson has made more varied and complex uses of constructive editing across a film, as I tried to show in Narration in the Fiction Film and in an Artforum essay. (5) A more radical approach, somewhat in the purely Kuleshovian spirit, can be seen in Godard’s films since the late 1970s. In presenting a scene, Godard often omits an establishing shot, so constructive editing takes over. But he makes the technique quite abrasive and ambivalent.

In watching films like Number, Please? and Presumed Innocent, we fill in the gaps between shots with ease. Godard, however, makes his images and sounds more fragmentary by equivocating about the Kuleshovian cues. The background elements and lighting don’t match entirely; time seems to be skipped over; glances and facial expressions are ambivalent. Adding to these factors, lines of dialogue slip in from offscreen. Godard does present a dramatic scene taking place, but he also creates a sense that images and sounds have been pried loose from their place in an ongoing action, floating somewhat free and functioning as objects of contemplation for their own sake. The cues that Lloyd insists on and that Kuleshov plays with are ones that Godard suppresses or makes ambiguous.

I’ve mentioned this tendency in a recent entry, but to elaborate a little, consider the science lecture in Hail Mary (Je vous salue Marie, 1985). A professor who has emigrated from an Eastern bloc country is explaining his theory that life on earth began and evolved because it was directed by extraterrestrials. No establishing shot of the classroom shows us where he, Eva, and two male students are located, so we have to construct a rough sense of their positions on the basis of a few cues. As Eva, perched on a windowsill, toys with a Rubik’s cube, we hear the professor’s lecture begin to speculate on whether life could have evolved spontaneously. His remarks about sunlight coincide with a burst of sun on her face.

After the Biblical title, “In those days,” we get a series of shots that allow us to apply our mental schema of a classroom lecture: attentive students looking off, a professor at the blackboard.

          

Lacking an establishing shot, we can’t specify how many people are in the class, nor indeed where Eva and her classmate are sitting—since the professor looks off in several directions during his talk.

At the end of his talk he remarks, still scanning the room, that we can presume that life exists in space. “We come from there.” At that point Godard cuts to the head of another student, seen from behind. The sproingy haircut is a little explosion of yellow, and it’s accompanied by a burst of choral music. And as the shot goes on, we start to notice that the professor is pacing in the background, out of focus.

The student, whom we’ll learn is named Pascal, asks a question (at least the slight head movement suggests that it comes from him), and the professor replies. Pascal scratches his head as the professor continues, still out of focus. If I had to choose one shot that condenses Godard’s strategy of suppression in this sequence, this shot would be my candidate.

At the end of the shot, the professor asks Eva to stand behind Pascal. Cut 180 degrees and somewhat farther back to Pascal, now seen from the front. The professor’s hand brings the Rubik’s cube into the shot and Eva comes up behind Pascal as the professor passes.

     

Later she and the prof will become lovers, but Godard lets them meet first as simply two torsos intersecting behind Pascal. The purpose is a demonstration that a “blind” agent can be steered toward a goal through simple yes/no commands. This models the professor’s theory that an alien intelligence could have guided evolution.

Pascal will twist the facets of the cube under Eva’s hints. Godard makes this a tactile, even erotic exchange, with the close-up of her by his ear and Eva saying, “Yes,” more urgently as Pascal’s hands arrive at a solution, in the close-up surmounting this blog entry.

     
The next two shots of the sequence center on the prof, who has exited frame right from the “three-shot.” Now he’s at the window, recalling the initial shot of Eva; but unlike her he’s little more than a silhouette. As crashing organ music is heard, he seems to be watching the experiment from afar. The second shot, an axial cut-in, coincides with the offscreen voice of a male student: “Were you exiled for these ideas?”

      

“These ideas, and others,” the prof replies. He says he’ll see the class on Monday, evoking the idea that he’s dismissing the offscreen students, and he turns his head slightly, though we can’t be sure they’re on his right. This shot will be held for some time as students quiz him more about his theory, and Eva asks him if he’d like to come over for a drink some evening. But we don’t see her say it. Godard cuts to a shot of Eva at the window, bathed in sunlight, opening and shutting her eyes as she slightly lifts and lowers her head.

As we see her, we hear the rustling of people leaving (so the class was evidently larger than three). And we hear him reply to her invitation, “That’s another story [scénario].” Are Eva and the prof looking at each other? We’re inclined to say yes, but her closed eyes and tilting head suggests someone lost in contemplation rather than engaged in conversation. Here the classic facial-expression cue is made indeterminate. We have no way of knowing if the prof’s line comes from offscreen, or is displaced from another point in time; maybe he has left the room. Such displaced diegetic sound occurs elsewhere in the film.

We don’t have to decide; our indecision is the point. By pruning away the most reliable cues, Godard wins both ways. We’re kept somewhat oriented to an intelligible action: a prof sets forth his far-fetched theory of human origins and a woman in his class invites him for coffee. This side of the balance allows us to feel that a story, however sketchy, is moving forward.

But the moment-by-moment texture of the scene allows the individual shots, gestures, and sounds to drift somewhat free. Each image takes on a more intrinsic weight, and the juxtaposition of picture and sound acquires a resonance that we usually call poetic. A shot of Eva in the sun playing with the Rubik’s cube, unanchored in time (during class? before class started?), invites us to apply metaphors, especially once we learn her name. Pascal’s thorny hair suggests not only extraterrestrials but the explosion of a nova. The silhouetted prof, detached from the mechanism he has set in motion, hints at an unknown deity watching the game play out according to his rules. Why do Godard films spawn long essays built out of erudite associations? Because the narrative progression relaxes and we can weave our own connotations out of what we see and hear.

If you don’t want to go down the expanding-association route, there’s another one open. When individual moments no longer accumulate ordinary dramatic pressure, we can savor the fugitive pleasures of the separate shots (light on face, lips by ear) and the patterns they form: flipover cuts, yellow hair and yellow facets, bookended shots of Eva at the window.

Those patterns, it should be clear, depend on our sensing a bump at every shot change, looking for a way to skip across the gap that Godard creates. The same belief that meaning and effect are born of gaps impelled Kuleshov too, and perhaps even Lloyd. If we pay attention to those gaps we can feel minds—both the filmmaker’s and ours—at work in them.

(1) Lev Kuleshov, “’50’: In Maloi Gznezdinokovsky Lane,” Kuleshov on Film, trans. and ed, Ronald Levaco (Berkeley: University of California Press, 1974), 200.

(2) V. I. Pudovkin, “The Naturschchik instead of the Actor,” Selected Essays, ed. and trans. Richard Taylor (Oxford: Seagull, 2006), 160.

(3) See Kristin Thompson, Yuri Tsivian, and Ekaterina Khokhlova, “The Rediscovery of a Kuleshov Experiment: A Dossier,” Film History 8, 3 (1996), 357-367.

(4) The sequence does begin with a long shot of the café, but it is so distant, crowded, and brief that it can’t really be said to establish the spatial relationships among the several characters we see.

(5) “Sounds of Silence,” Artforum International (April 2000): 123.

DB here:

The first quarter of the year is the biggest slump time for movie theatres. (1) Holiday fatigue, thin budgets, bad weather, the Super Bowl, and the distractions of the awards season depress admissions. If people go to the movies, they tend to catch up on Oscar nominees, and studios don’t want to release high-end films that might suffer from the competition. But screens need fresh product every week, so most of what gets released at this time of the year might charitably be called second-tier.

Ambitious filmmakers fight to keep out of this zone of death. You could argue that the January release slot of Idiocracy told Mike Judge exactly what Fox thought of that ripe exercise in misanthropy. Zodiac, one of the best films of 2007, opened on 1 March, and even ecstatic reviews couldn’t push it toward Oscar nominations. You can imagine what chances for success Columbia has assigned to Vantage Point (a 22 February bow). [But see my 4 Feb. PPPS below.]

Yet this is a flush period for those of us who like to explore low-budget genre pieces. I have to admit I enjoy checking on those quickie action fests and romantic comedies that float up early in the year. They’re today’s equivalent of the old studios’ program pictures, those routine releases that allowed theatres to change bills often. In their budgets, relative to blockbusters, today’s program pix are often the modern equivalent of the studios’ B films.

More important, these winter orphans are often more experimental, imaginative, and peculiar than the summer blockbusters. On low budgets, people take chances. Some examples, not all good but still intriguing, would be Wild Things (1998), Dark City (1998), Romeo Must Die (2000), Reindeer Games (2000), Monkeybone (2001), Equilibrium (2002), Spun (2003), Torque (2004), Butterfly Effect (2004), Constantine (2005), Running Scared (2006), Crank (2006), and Smokin’ Aces (2007). The mutant B can be found in other seasons too—one of my favorites in this vein, Cellular (2004), was released in September—but they’re abundant in the year’s early months.

By all odds, Cloverfield ought to have been another low-end release. A monster movie with unknown players, running a spare 72 minutes sans credits, budgeted at a reputed $25 million, it’s a paradigm of the winter throwaway. Except that it pulled in $46 million over a four-day weekend and became the highest-grossing film (in unadjusted dollars) ever to be released in January. Here the B in “B-movie” stands for Blockbuster.

I enjoyed Cloverfield. It starts with a sharp premise, but as ever, execution is everything. I see it as a nifty digital update of some classic Hollywood conventions. Needless to say, many spoilers loom ahead.

If you find this tape, you probably know more about this than I do

Everybody knows by now that Cloverfield is essentially Godzilla Meets Handicam. A covey of twentysomethings are partying when a monster attacks Manhattan, and they try to escape. One, Rob, gets a phone call from his off-again lover Beth, who’s trapped in a high-rise. He vows to rescue her. He brings along some friends, one of whom documents their search with a video camera. It’s a shooting-gallery plot. One by one, the characters are eliminated until we’re down to two, and then. . . .

Cloverfield exemplifies what narrative theorists call restricted narration. (Kristin and I discuss this in Chapter 3 of Film Art.) In the narrowest case of restricted narration, the film confines the audience’s range of knowledge to what one character knows. Alternatively, as when the characters are clustered in the same space, we’re restricted to what they collectively know. In other words, you deny the viewer a wider-ranging body of story information. By contrast, the usual Godzilla installment is presented from an omniscient perspective, skipping among scenes of scientists, journalists, government officials, Godzilla’s free-range ramblings, and other lines of action. Instead, Cloverfield imagines what Godzilla’s attack would look and feel like on the ground, as observed by one group of victims.

Horror and science fiction films have used both unrestricted and restricted narration. A film like Cat People (1942) crosscuts what happens to Irena (the putative monster) with scenes involving other characters. Jurassic Park and The Host likewise trace out several plot strands among a variety of characters. The advantage of giving the audience so much information is that it can feel apprehension and suspense about what the characters don’t know is happening. Our superior knowledge can make us worry about those poor victims oblivious to their fate.

But these genres have relied on restricted narration as well. Invasion of the Body Snatchers (1956) is a good example; we are at Miles’s side in almost every scene, learning of the gradual takeover of his town as he does. Night of the Living Dead (1968), Signs (2002), and War of the Worlds (2005) do much the same with a confined group, attaching us to one or the other momentarily but never straying from their situation.

The advantages of restricted narration are pretty apparent. You can build up uncertainty and suspense if we know no more than the character(s) being attacked by a monster. You can also delay full revelation of the creature, a big deal in these genres, by giving us only the glimpses of it that our characters get. Arguably as well, by focusing on the characters’ responses to their peril, you have a chance to build audience involvement. We can feel empathy and loss if we’ve come to know the people more intimately than we know the anonymous hordes stomped by Godzilla. Finally, if you need to give more wide-ranging information about what’s happening outside the characters’ immediate situation, you can always have them encounter newspaper reports, radio bulletins, and TV coverage of action occurring elsewhere.

People sometimes think that theoretical distinctions like this overintellectualize things. Do filmmakers really think along these lines? Yes. Matt Reeves, the director of Cloverfield, remarks:

The point of view was so restricted, it felt really fresh. It was one of the things that attracted me [to this project]. You are with this group of people and then this event happens and they do their best to understand it and survive it, and that’s all they know.

For your eyes only

Restricted narration doesn’t demand optical point-of-view shots. There aren’t that many in Invasion of the Body Snatchers or the other examples I’ve indicated. Still, for quite a while and across a range of genres, filmmakers have imagined entire films recording a character’s optical/ auditory experience directly, in “first-person,” so to speak.

Again, it’s useful to recognize two variants of this narrational strategy. One we can call immediate—experiencing the action as if we stood in the character’s shoes. In the late 1920s, the great documentary filmmaker Joris Ivens tried to make what he called his I-film, which would record exactly what a character saw when riding a bike, drinking a glass of beer, and the like. He was dismayed to find that bouncing and swiveling the camera as if it were a human eye ignored the fact that in real life, our perceptual systems correct for the instabilities of sensation. Ivens abandoned the project, but evidently he couldn’t get the notion out of his head; he called his autobiography The Camera and I. (2)

Hollywood’s most strict and most notorious example of directly subjective narration is Robert Montgomery’s Lady in the Lake (1947). Its strangeness reminds us of some inherent challenges in this approach. How do you show the viewer what your protagonist looks like? (Have him pass in front of mirrors.) How do you skip over the boring bits? (Have your hero knocked unconscious from time to time.) How do you hide the inevitable cuts? (Try your best.) Even Montgomery had to treat the subjective sequences as long flashbacks, sandwiched within scenes of the hero in his office in the present telling us what he did next.

Because of these problems, a sustained first-person immediate narration is pretty rare. The best compromise, exploited by Hitchcock in many pictures and especially in Rear Window (1954), is to confine us to a single character’s experience by alternating “objective” shots of the character’s action with optical point-of-view shots of what s/he sees.

What I’m calling immediate optical point of view is just that: sight (and sounds) picked up directly, without a recording mechanism between the story action and the character’s experience. But we can also have mediated first-person point of view. The character uses a recording technology to give us the story events.

In a brilliant essay on the documentary Kon-Tiki (1950), André Bazin shows that our knowledge of how Thor Heyerdahl filmed his raft voyage lends an unparalleled authenticity to the action. Heyerdahl and his crew weren’t experienced photographers and seem to have taken along the 16mm camera as an afterthought, but the very amateurishness of the enterprise guaranteed its realism. Its imperfections, often the result of hazardous conditions, were themselves testimony to the adventure. When the men had to fight storms, they had no time to film; so Bazin is able to argue, with his inimitable sense of paradox, that the absence of footage during the storm is further proof of the event. If we were given such footage, we might wonder if it was staged afterward.

How much more moving is this flotsam, snatched from the tempest, than would have been the faultless and complete report offered by an organized film. . . . The missing documents are the negative imprints of the expedition. (3)

What about fictional events? In the 1960s we started to see fiction films that presented themselves as recordings of the events as the camera operator experienced them. One early example is Stanton Kaye’s Georg (1964). The first shot follows some infantrymen into battle, but then the framing wobbles and the camera falls to earth. We see a tipped angle on a fallen solider and another infantryman approaches.

He bends toward us; the frame starts to wobble and we are lifted up. On the soundtrack we hear, “I found my camera then.”

The emergence of portable equipment and cinema-verite documentary seems to have pushed filmmakers to pursue this narrational mode in fiction. One result was the pseudo-documentary, which usually doesn’t present the story as a single person’s experience but rather as a compilation of first-person observations. Peter Watkins’ The War Game (1967) presents itself as a documentary shot during a nuclear war, and it contains many of the visual devices that would come to be associated with the mediated format—not only the flailing camera but the face-on interview and the chaotic presentation of violent action. There’s also the pseudo-memoir film, pioneered in David Holzman’s Diary (1967). Later examples of the pseudo-documentary are Norman Mailer’s Maidstone (1971) and the combat movie 84 Charlie MoPic (1989). (4)

As lightweight 16mm cameras made filming easier, directors adapted that look and feel to fictional storytelling. The arrival of ultra-portable digital cameras and cellphones has launched a similar cycle. Brian DePalma’s Redacted (2007), yet another war film, has exploited the technology for docudrama. A digital equivalent of David Holzman’s Diary, apart from Webcam and YouTube material, is Christoffer Boe’s Offscreen (2006), which I discussed here.

Interestingly, Orson Welles pioneered both the immediate and the mediated subjective formats. One of his earliest projects for RKO was an adaptation of Heart of Darkness, in which the camera was to represent the narrator Marlowe’s optical perspective throughout. (5) Welles had more success with the mediated alternative, though in audio form. His “War of the Worlds” radio broadcast mimicked the flow of programming and interrupted it with reports of the aliens’ attack. The device was updated for television in the 1983 drama Special Bulletin.

Sticking to the rules

Cloverfield, then, draws on a tradition of using technologically mediated point-of-view to restrict our knowledge. Like The Blair Witch Project (1999), it does this with a horror tale. But it’s also a Hollywood movie, and it follows the norms of that moviemaking mode. So the task of Reeves, producer J. J. Abrams, and the other creators is to fit the premise of video recording to the demands of classical narrative structure and narration. How is this done?

First, exposition. The film is framed as a government SD video card (watermarked DO NOT DUPLICATE), the remains of a tape recovered from an area “formerly known as Central Park.” This is a modern version of the discovered-manuscript convention familiar from the nineteenth-century novel. When the tape starts, showing Rob with Beth in happy times, its read-out date of April plays the role of an omniscient opening title. In the course of the film, the read-outs (which come and go at strategic moments) will tell us when we’re in the earlier phase of their love affair and when we’re seeing the traumatic events of May.

Likewise, the need for exposition about characters and relationships at the start of the film is given through a basic premise. Jason wants to record Rob’s going-away-party and he presses Rob’s friend Hud into service as the cameraman. Off the bat, Hud picks out our main characters in video portraits addressed to Rob. What follows indicates that Hud will be amazingly prescient: His camera dwells on the characters who will be important in the ensuing action.

Next, overall structure. The Cloverfield tape conforms to the overarching principles that Kristin outlines in Storytelling in the New Hollywood and that I restated in The Way Hollywood Tells It. (Another example can be found here.) A 72-minute film won’t have four large-scale parts, most likely two or three. As a first approximation, I think that Cloverfield breaks into:

*A setup lasting about 30 minutes. We are introduced to all the characters before the monster attacks. Our protagonists flee to the bridge, where Jason dies. Near the end of this portion, Rob gets a call from Beth, and he formulates the dual goals of the film: to escape from the creature, and to rescue Beth. Along the way, Hud declares he’s going to record it all: “People are gonna know how it all went down. . . . It’s gonna be important.”

*A development section lasting about 22 minutes. This is principally a series of delays. Rob, Hud, Lily, and Marlena encounter obstacles. Marlena falls by the wayside. They are given a deadline: At 0600 they must meet the last helicopters leaving Manhattan.

*A climax lasting about 20 minutes. The group rescues Beth and meets the choppers, but the one carrying Rob, Hud, and Beth falls afoul of the beast. They crash in Central Park, and Hud is killed, his camera recording his death at the jaws of the monster. Huddled under a bridge, Rob and Beth record a final video testimonial before an explosion cuts them off.

*An epilogue of one shot lasting less than a minute: Rob and Beth in happier times on the Ferris wheel at Coney Island—a shot left over from the earlier use of the tape in April.

Next, local structure and texture. It takes a lot of artifice to make something look this artless. The imagery is rich and vivid, the sharpest home video you ever saw. The sound is pure shock-and-awe, bone-rattling, with a full surround ambience one never finds on a handicam. (6) Moreover, Hud is remarkably lucky in catching the turning points of the action. All the characters’ intimate dramas are captured, and Hud happens to be on hand when the head of Miss Liberty hurtles down the street.

Bazin points out that in fictional films the ellipses are cunning gaps, carefully designed to fulfill narrative ends—not portions left out because of the physical conditions of the shoot. Here the cunning gaps are justified as constrained by the physical circumstances of filming. When Hud doesn’t show something, it’s usually because it’s what the genre considers too gross, so the worst stretches take place in darkness, or offscreen, or strategically shielded by a prop when the camera is set down.

Mostly, though, Hud just shows us the interesting stuff. He turns on the camera just before something big happens, or he captures a disquieting image like that of the empty Central Park carriage.

At least once, the semi-documentary premise does yield something evocative of the Kon-Tiki film. Hud has to leap from one building to another, many stories above the street. He turns the camera on himself: “If this is the last thing you see, then I died.” He hops across, still running the camera, but when a rocket goes off nearby, a sudden cut registers his flinch. For an instant out of sheer reflex, he turned off the camera.

Overall, Hud’s tape respects the flow of classical film style. Unlike the Lady in the Lake approach, the mediated POV format doesn’t have a problem with cuts; any jump or gap is explained as a moment when the operator switched off the camera. Most of Hud’s “in-camera” cuts are conventional ones, skipping over a few inconsequential stretches of time. There are as well plenty of hooks between scenes. (For more on hooks, go here.) Hud says: “I’ll walk in the tunnels.” Cut to characters walking in the tunnels. More interestingly, visible cuts are rare, which again respects the purported conditions of filming. Cloverfield has much longer takes than any recent Hollywood film I know. I counted only about 180 shots, yielding an average of 24 seconds per shot (in a genre in which today’s films average 2-5 seconds per shot).

The digital palimpsest

We could find plenty of other ways in which Cloverfield adapts the handicam premise to the Hollywood storytelling idiom. There are the product placements that just happen to be part of these dim yuppies’ milieu. There are the character types, notably the sultry Marlena and the hero’s weak friend who’s comically a little slow. There’s the developing motif of the to-camera addresses, with Rob and Beth’s final monologues to the camera counterbalancing the party testimonials in the opening. There’s the final romantice exchange: “I love you.” “I love you.” The very last shot even includes a detail that invites us to re-view the entire movie, at the theatre or on DVD. But let me close by noting how some specific features of digital video hardware get used imaginatively.

I’ve already mentioned how the viewfinder date readout allows us to keep the time structure clear. There’s also the use of a night-vision camera feature to light up those spidery parasites shucked off by the big guy. Which scares you more—to glimpse the pinpoint eyes of critters skittering around you in the dark, or to see them up close in a sickly green light?

More teasing is the fact, set up in the first part, that this video is being recorded over an old tape of Rob’s. That’s what turns the opening sequence of Rob and Beth in May into a prologue: the tape wasn’t rewound completely for recording the party. Later, at intervals, fragments of that April footage reappear, apparently through Hud’s inadvertently advancing the tape. The snippets functions as flashbacks, showing Rob and Beth going to Coney Island and juxtaposing their enjoyable day with this horrendous night.

Cleverly, on the tape that’s recording the May disaster something always prepares the audience for the shift. For instance, when Jason hands the camera over, we hear Hud say, “I don’t even know how to work this thing.” Cut to an April shot of Beth on the subway, suggesting that he’s advanced fast forward without shooting. Likewise, when Rob says, “I had a tape in there,” we cut to another April shot of Beth. As a final fillip, the footage taken in May halts before the tape ends, so we get the epilogue showing Rob and Beth on the Ferris wheel in April, emerging like figures in a palimpsest.

No less clever, but also a little poignant, is the use of the fallen-camera convention. It appears once when Beth has to be extricated from her bed. Hud sets the camera down by a concrete block in her bedroom, which conceals her agony. More striking is the shot when the camera, dropped from Hud’s hand, lies in the grass, and the autofocus device oscillates endlessly, straining to hold on his lifeless face.

In sum, the filmmakers have found imaginative ways of fulfilling traditional purposes. They show that the look and feel of digital video can refresh genre conventions and storytelling norms. So why not for the sequel show the behemoth’s attack from still other characters’ perspectives? This would mobilize the current conventions of the narrative replay and the companion film (e.g., Eastwood’s Iwo Jima diptych). Reeves says:

The fun of this movie was that it might not have been the only movie being made that night, there might be another movie! In today’s day and age of people filming their lives on their iPhones and Handycams, uploading it to YouTube. . . .

So the Dead Zone of January through March yields another hopeful monster. What about next month’s Vantage Point? The tagline is: 8 Strangers. 8 Points of View. 1 Truth. Hmmm. . . . Combining the network narrative with Rashomon and a presidential assassination. . . . Bet you video recording is involved . . . . See you there?

PS: At my local multiplex, you’re greeted by a sign: WARNING: CLOVERFIELD MAY INDUCE MOTION SICKNESS. I thought this was just the theatre covering itself, but I’ve learned that no recent movie, not even The Bourne Ultimatum, has had more viewers going giddy and losing their lunch. You can read about the phenomenon here, and Dr. Gupta weighs in here. My gorge can rise when a train jolts, but I had no problems with two viewings of Cloverfield, both from third row center.

Anyhow, it will be perfectly easy to watch on your cellphone. But we should expect to see at least one pirate version shot in a theatre by someone who’s fighting back the Technicolor yawn, giving us more Queasicam than we bargained for.

(1) The only period that rivals this slow winter stretch is mid-August to October, when genre fare gets pushed out to pick up on late summer business. [Added 26 January:] There are, I should add, two desirable weekends in the first quarter, those around Martin Luther King’s birthday and Presidents’ Day. Studios typically aim their highest-profile winter releases (e.g., Black Hawk Down, 2001) for those weekends.

(2) Joris Ivens, The Camera and I (New York: International Publishers, 1969), 42.

(3) André Bazin, “Cinema and Exploration,” What Is Cinema? Vol. 1, trans. and ed. Hugh Gray (Berkeley: University of California Press, 1967), 162.

(4) Not all pseudodocumentaries present themselves as records of a person’s observation. Milton Moses Ginsberg’s Coming Apart (1969) presents itself as an objective record, by a hidden camera, of a psychiatrist’s dealings with his patients. Like a surveillance camera, it doesn’t purport to embody anybody’s point of view.

(5) Jonathan Rosenbaum, Discovering Orson Welles (Berkeley: University of California Press, 2007), 28-48.

(6) For Kevin Martin’s informative account of the film’s polished lighting and high-definition video capture, go here (and scroll down a bit). For discussions of contemporary sound practices in this genre, see William Whittington’s Sound Design in Science Fiction (Austin: University of Texas Press, 2007).

PS: Thanks to Corey Creekmur for correcting two slips in my initial post!

PPS 28 January: Lots of Internet buzz about the film since I wrote this. Thanks to everyone who linked to this post, and special thanks for feedback from John Damer and James Fiumara.

Some people have asked me to comment on the social and cultural implications of Cloverfield’s references to 9/11. At this point I think that genre cinema has dealt more honestly and vividly with the traumas and questioning around this horrendous event than the more portentous serious dramas like United 93, World Trade Center, and the TV show The Road to 9/11.

The two most intriguing post-9/11 films I know are by Spielberg. The War of the Worlds gives a really concrete sense of what a hysterical America under attack might be like, warts and all. (It reminded me of a TV show I saw as a kid, Alas, Babylon (1960), a surprisingly brutal account of nuclear-war panic in suburbia.) Spielberg’s underrated The Terminal reminds us, despite its Frank Capra optimism, that the new Security State is run by bureaucrats with fixed agendas and staffed by overworked people of color, some themselves exiles and immigrants.

I think that Cloverfield adds its own dynamic sense of how easily the entitlement culture of upwardly mobile twentysomethings can be shattered. Genre films carry well-established patterns and triggers for feelings, and a shrewd filmmaker can channel them for comment on current events—as we see in the changing face of Westerns and war films in earlier phases of Hollywood history.

On this point, Cinebeats offers some shrewd responses to criticisms of Cloverfield here.

Finally: In the new Creative Screenwriting an informative piece (not available online) indicates that the initial logline for Vantage Point on imdb is misleading. Screenwriter Barry Levy planned to present the assassination from seven points of view, but reduced that to six. As for my speculation that video recording/replay would be involved, a production still seems to offer some evidence. Shall we call it the Cloverfield effect? The same issue of CS has a brief piece on the script for Cloverfield.

PPS 30 January: Shan Ding brings me another story about the making of Cloverfield, and Reeves is already in talks for a sequel, says Variety.

PPPS 4 February: A recent story in The Hollywood Reporter offers a nuanced account of how Hollywood is rethinking its first-quarter strategies. Across the last 4-5 years, a few big releases have done fairly well between January and April; a high-end film looks bigger when there is less competition. The author, Steven Zeitchik, suggests that the heavy packing of the May-August period and the need for a strong first weekend are among the factors that will encourage executives to spread releases through the less-trafficked months. I hope, though, that tonier fare won’t crowd out the more edgy, low-end genre pieces that bring me in.

PPPPS 8 February: How often has a wounded Statue of Liberty featured in the apocalyptic scenarios of comics and the movies? Lots, it turns out. Gerry Canavan explains here.

DB here:

James Quandt started it.

The indefatigable Senior Programmer of the Cinémathèque Ontario emailed me in early 2004 to ask if I had any thoughts on the aspect ratios of Godard films. He attached an essay which eventually appeared in the gorgeous anthology, For Ever Godard. Reading a Quandt essay is like eating a ripe nectarine, tangy and nourishing. So you should find the original and indulge yourself. (1)

You might be asking what the term aspect ratio means. It refers to the ratio of the width to the height of the film image. The image was fairly square in the early silent era, then became roughly standardized at 4 x 3, or as the pros say, 1.33:1. Sound filming made the format a tad more horizontal, at 1.37. Anamorphic widescreen (CinemaScope and its brethren) was more or less standardized at 2.35 (more recently 2.40). Various non-anamorphic, or “flat” aspect ratios have appeared since the early 1950s. The US has favored 1.85, Europe has been known to use the squarer 1.66, and some films, like E. T., are designed for 1.75. Widescreen TVs are set at 16:9, or about 1.78:1, so that’s likely to be a common proportion in the future. We discuss aspect ratios at more length in Film Art: An Introduction (pp. 183-185 in the newest edition).

Filmies care about aspect ratios because shot composition matters. Sometimes the print is “hard-matted,” with the correct proportions given as black bars at the top and bottom of the frame, like video letterboxing. Here’s an example, from a 16mm print of Godard’s 1972 Tout va bien. It is hard-matted to 1.66. (The original film is in color.)

If the image isn’t hard-matted, the projectionist must insert an aperture plate that will mask the image properly. But what plate? Should she set it for 1.37? That’s a very rare option nowadays, and many theatres aren’t really designed to show it. Typically, if the print doesn’t indicate, the US projectionist will fall back on 1.85. Nowadays, if a Hollywood film isn’t in Scope, the projectionist is expected to use that ratio. Some shots will be problematic if the projectionist includes more than the 1.85 format allows. Here’s a full-frame film strip from The Hudsucker Proxy, where you can see that a chunk of the set is blocked or missing in the bottom area, and a microphone peeks into the frame from the top. (2)

Like many other movies, the films made by Godard since the mid-1970s show up at the projection booth without hard matteing. So at what ratio do we show them?

A great many careful viewers have voiced their views on the Internets, and I’ve learned a lot from the discussions here and here. In part this blog entry is an effort to introduce readers to this debate.

Moreover, this apparently film-wonkish question has wider implications. It can teach us a fair amount about how film images work, and the implications of any masking, matteing, or cropping of an image—especially on DVD. So if you’re interested in Godard, keep reading. If not, skip to the final section, “Relationships: The fundamental question,” where I talk about some artistic effects of cropping any film image.

It’s a just image, not just an image

James Q was mounting one of his typically ambitious retrospectives, this time on JLG, and so his essay posed a question that had long been ignored.

A disturbing discovery of the retrospective was how frequently the full-frame compositions of Godard’s late films have been ignored and overruled. Many of the prints are clearly marked by the lab with the widescreen ratios of 1.66 or (the almost standard) 1.85, and their subtitles are printed in the frame at the height indicated by those standards. Our meticulous projectionist Kate Mackay experimented with whole reels of films, showing them first in 1.33 and then in the prescribed wider screen ratio, revealing the violence done to the compositions when shown the latter way.

James found that several films, including Passion, Je vous salue Marie (Hail Mary), Nouvelle Vague, Hélas pour moi, and For Ever Mozart, looked “abjectly constricted” in 1.85. So James wrote the man himself.

Disturbed by some oddly cropped compositions in Éloge de l’amour, which result in seemingly unintentional beheadings and concretions, I consulted Godard by fax about the aspect ratio and he confirmed that it was indeed, as stated, 1.66 (rather old-fashioned in its own way). That he occasionally still seems to be jamming a 1.33 composition into a frame that cannot accommodate it suggests his instinctual preference for the open image.

I couldn’t help James much at the time, but I did send him a couple of frames that favored squareish compositions and that came from 35mm prints. Other frames we reproduced, at 1.37, in both editions of Film History: An Introduction. The still that pretty much settles the matter for me is the gorgeous shot of Nathalie Baye and Johnny Hallyday at the top of this entry. Here’s the image as it is on a 35mm print.

Downsize that to 1.66 without losing those eyes!

Later I sent James another killer example, drawn also from a 35mm print of Detective. (It’s in the new Film Art, p. 46.)

Here’s what it would look like in one try at 1.66 matteing.

I say “one try” at a matted version because I didn’t take as much off the top as a normal aperture plate would; I didn’t want to slice into the hands and the gun. Not only is the 1.37 image preferable (we get to see Claude Brasseur’s slumping posture) but 1.66 looks, as James says, jammed. The 1.37 ratio lets Godard load information in the very top of the shot, as we’ll see often in the examples to come. And, needless to say, at 1.85 the shot would make no sense.

Soon after James and I had our exchange, Godard—perhaps prodded by James’ query—sent a diagram to Cahiers du cinéma. (3) (Thank you, Craig Keller, aka evillights, who called attention to it on this thread.) The Cahiers editors report that Godard has asked that Notre musique be shown in 1.37. His photomontage lines up two shots from the film and arranges them according to the three major “flat” ratios, and for each one he supplies a tart annotation.

For the close-up of the woman, the captions translate as: 1.37 person. 1.66 character. 1.85 satellite slave. For the long shot of the street, we get: 1.37 Proof of Serbian bombing. 1.66 Proof diminished by Europe/ USA. 1.85 Extermination of proof (Milosovec acquitted).

Pretty strong evidence that JLG doesn’t like cropping the classic format. But these remarks are about Notre musique. What about the other films? Apart from the evidence onscreen and on the film strip, we can add one thing. Evidently he shoots at 1.37, but there’s also evidence that in the late stages of postproduction he seems to preserve that ratio. Here, for example, is a sheet of color timing instructions for Nouvelle Vague. (4) Godard has pasted in a frame for each shot in the sequence, and alongside he notes how much red, green, and blue he wants. The frames he mounted are 1.37.

Recently our Cinematheque has been holding a Godard retrospective, and I’ve taken the opportunity to revisit the aspect ratio issue. As an archival venue, we can screen at any ratio, even the squarish silent and early-sound ones. Our projectionist Jared Lewis has run the Godards at 1.37. They look fine.

Jared pointed out to me that one other factor leans toward screening them in the 1.37 format: the thickness of the spaces between frames. In a modern “full-frame” film like Hudsucker Proxy, there is very little space between the frames. The line separating one from another is quite thin. That tends to make the frames squarer, closer to 1: 1.2, as I mention in endnote 2.

In a classic sound film, there is often more space between the frames. Usually that space is black, but I can’t resist showing what it looked like in Technicolor.

This frame, from Renoir’s Golden Coach, shows the characteristic silver frame surround (and silver soundtrack) of a true Tech print. Nifty, huh?

Anyhow, the sort of thick spacing between frames that we get usually find in Godard prints, and that’s visible in the Baye/Hallyday frame above, favors the classic ratio. The thickness of these spacers is similar to what we find in a modern film that was explicitly designed for 1.37 screening, Hans-Jürgen Syberberg’s Parsifal (1982).

This array, Jared points out, gives different frame proportions than one would find in a print hard-matted to 1.66 or 1.85.

One more wrinkle. On the film strip, Godard’s frames aren’t all the same dimensions. Here are two from Je vous salue Marie; note that the first is taller, with narrower spacers, than the second.

Both, however, would be appropriately shown at 1.37.

How then are we to explain Godard’s saying the films should run at 1.66? Perhaps, as one of the online commentators has suggested, Godard assumed that 1.66 is the closest that most commercial venues can come to 1.37. Perhaps too he was just being contrary–that is, just being Godard.

Fortunately, some DVD producers seem to recognize his full-frame aesthetic. The UK version of Detective is full-frame and preserves my nifty shots. Also, the Cahiers du cinéma discs for Prénom Carmen, Hélas, and so on are at 1.37. Bowing to Godard’s wishes, Wellspring’s version of Notre Musique announces that it is presented “in its original theatrical aspect ratio of 1.33.” Although purists may say that virtually no theatres showed it that way, we should appreciate the gesture.

Relationships: The fundamental question

Even if you’re not that interested in Godard, everybody should be aware of what video cropping can do to the film image. I’m not talking about panning and scanning, that process which begins with a widescreen film, typically one of an aspect ratio 1:2.40, and extracts a 1.37 frame out of it for video purposes. This is deplorable, but most of us are alert to it. What’s more interesting is the sort of thing that happened when a film is cropped inaccurately, either in projection or for DVD.

My example will be from the first reel of Godard’s Éloge de l’amour/ In Praise of Love, which is available on DVD in a full frame version from Optimum in the UK and in a cropped version from New Yorker in the US. I won’t be focusing on the quality of each transfer, though the Optimum one looks superior to me.

Nor will I do a detailed narrative account, because I find the characters and their interactions still fairly baffling. I’m always amazed that critics can praise a Godard film without ever getting down to explicating what’s literally happening in a scene. They write as if these films were telling their stories straightforwardly. Without help from the presskits, could journalists discern even the sketchy plots they refer to? A great deal of the fascination of Godard’s late works comes from his refusal of the most elementary forms of exposition–picking out characters, explaining their relations, and the like. There is always a story, but it’s about three-quarters hidden, and this seems to me to require a lot more analysis than people tend to give it.

Anyhow, in studying Éloge de l’amour‘s video versions, I learned that there can be a big difference between tiny numbers. For instance, the Optimum version is prepared at 1.35:1. No big deal between this and 1.37:1, surely? Except that the New Yorker version seems to have started from a 1.37 frame. Even though it’s cropped on the top and bottom, it consistently supplies a tad more information on the right and left edges, and these extra bits are visible in side-by-side comparison. First, a 35mm frame.

Needless to say, the projector’s aperture plate won’t preserve everything in the physical frame; at a minimum it masks off the curved corners. But if we look at the two video versions, there are some surprises.

The 1.37 version of this shot is of course much closer to the overall composition of the original. But more areas of the window frame (on the left) and the painting (on the right) are visible in the widescreen version than in the full-frame one. Did going for 1.35 shave off those areas? Moreover, New Yorker’s cropping is at 1.77, for all intents and purposes the same as 1.75. But to achieve this wide frame, the transfer of some shots seems to have been optically stretched a little. In some upcoming examples the faces are a bit plumper and the surroundings a bit more horizontally spacious.

Okay, maybe I’m splitting hairs. So let me assume that the UK DVD preserves a reasonable amount of the 35mm original. I want to consider some effects of the cropping we get in the US DVD. Some are obvious, some more subtle, and all go beyond this individual case to suggest the results of overcropping any movie.

(1) Of course we lose the top and bottom. In the full-frame shots from Hudsucker Proxy, no problem; the filmmakers are counting on the projectionist to mask the frame. But in Godard the cropping makes us lose stuff. Godard likes to frame heads pretty high in the shot, and this means that we often lose part of them.

Heads are trimmed in movies all the time, and it doesn’t much matter in close views. But in Éloge, Godard is composing long shots with heads quite high up. He will even daringly chop off heads himself. This is partly a strategy to conceal who is present, to block our recognizing characters by their faces. It also has the effect of activating areas of the frame that aren’t usually so important. We have to strain to see partially visible things, tucked away in bits of the shot.

In the example below, I submit, the original composition creates a tension among three centers of interest: the two very visible paintings and the almost indiscernible face of the art dealer standing by the rear window. That tension is lost when the 1.77 cropping lops off the head in the background. Significantly, the man offscreen left is talking about how classic painting displayed “relationships” (rapports)–presumably both personal and pictorial. “That’s the fundamental question.”

The framing of the assistant in the foreground, incidentally, shows that spotting a decapitation in a video version doesn’t necessarily mean that Godard wanted every head to be seen.

In a later scene, we strain to see the older man’s face as he bends over Bruno. As he speaks Picasso’s immortal line, his profile scrapes the very top of the shot, but not in the cropped version.

In most film shots, the upper half of the frame harbors what we look at first, so we’re probably most likely to notice when something goes missing there. But actually, the area at the bottom of the frame is important too, especially as part of Godard’s all-over approach to composition.

Throughout early scenes of the film, Godard’s compositions favor the art works and minimize the humans trafficking in them. So the picture (by Delacroix? Matisse?) on the coffee table is foregrounded when the art collector signs the papers proffered by a mostly unseen woman, but it vanishes in the cropped version.

Likewise, the old man on the bed can rub his glasses fretfully at the very bottom of the 1.37 format, but that performance detail goes for naught in the 1.77 format.

More generally, even when we scan the top half of the frame for major information, we tend to take for granted that people are anchored to a ground plane, the earth or the floor or whatever. Often, of course, film shots don’t show us this ground. But the material at the bottom of a distant view can weight the shot, providing a sense of gravity. Here, the dealer peering over his balcony is minimally tied to the patio ground (as minimally as he was visible in the earlier shots when his head grazed the upper edge, I suppose). But in the 1.77 version he floats free.

(2) The top and bottom zones include the corners of the frame as well, but I single them out for special mention because I like them so much. Again, we don’t expect key information to be tucked there, but it can happen—in Godard, in Tati (a big influence on the late Godard), and even in one remarkable shot in Lumet’s recent Before the Devil Knows You’re Dead. In the first reel of Éloge de l’amour, the best example I can find comes with the long-shot of the woman, turned from us, standing at the window. In the lower right corner of the shot sits a woman’s photograph on a table. The 1.77 frameline chops it off and makes it less segregated for our notice: we lose the spacing that separates it from the other objects on the table. Since the voice-over is meditating on memory, the photo adds an overtone to the shot, but less clearly in the cropped version.

Maybe it matters, maybe not; but it’s a lot harder to see in 1.77 than in the 1.37 transfer. Something similar happens with the businessman’s hand in the lower left of this shot.

(3) Cutting off top and bottom alters the shot scale. All other things being equal, cropping not only eliminates; it enlarges. Figures come closer to us. A medium shot becomes a medium close-up. All of the examples so far indicate this to some degree, but it comes across clearly in these variants.

Again, note the stretching. It seems that someone decided that the image had to be 1.75 and instead of cropping it, stretched the 1.66 one. Yikes!

(4) Overambitious cropping changes the compositional dynamics. In reducing information, it reorganizes the composition. Rudolf Arnheim (I blogged about his achievements here) suggested that we consider a picture as a field of vectors and forces, pushes and pulls, balance and imbalance, rival centers of attention. By changing the framing we change the relation of the figures to the edges, and this can alter the composition.

The clearest examples come from the sort of reframings we find when a Super-35mm film is rendered in home video versions in both 2.40 and 1.37.

In this shot from 8mm (the movie, not the gauge), the cashier questioned by Cage looks more isolated in the Scope framing, while in the full frame they seem closer together and he seems to press in on her. Cropping can change a lot.
Now consider this comparison.

Godard’s original shot (here from 35mm) keeps the painting’s upper horizon, the darker, frothy waterline as a kind of backboard, halting the water’s recession into the distance. Graphically, the water on the right center becomes a negative space for the two figures, with the boy counterbalanced by the tip of the skiff.

But the cropped version loses the distant waterline, creating an infinite stretch of space top to bottom, and the boy’s head seems to float more freely. Most starkly, the skiff, by losing its shadow, seems to have swung more toward us.

It’s worth noting that Godard himself is a mean hand at radical cropping. I’ll forebear from rambling on about what his original framing above does to the original, Manet’s Seine at Argenteuil (1874), but it could constitute a lesson in how framing changes effect and meaning.

Several factors come into play when we look at this shot from the two DVD versions.

The woman’s face is off-center in both images, but it looks more off-center in the 1.77 transfer. In fact, despite the extra bits on left and right, it is measurably more off-center, because of this transfer’s optical stretching. Yet I’d argue in addition that the cropping of the frame has squeezed the pictorial elements into a stronger horizontal to-and-fro, giving a sense that she has been pushed more out of the middle. You can see it more markedly if we crop it more drastically, and it may help to hide the others when you look at this.

This effect is akin to what happens in the cropping of the 8mm example: the spatial relations have reorganized in relation to the frame edges. Rapports again.

(5) Overcropping can affect the way we experience the time of the shot. Before you call the men in the white coats, I hasten to say that cropping is purely a spatial effect, but in cinema space is bound up with time.

We’ve seen that Godard manipulates the vertical dimension of the frame to an unusual degree, and the effect on time becomes apparent in one scene of Éloge in which Bruno talks with an older man, in a sort of casting session for his project. First we see Bruno alone, and as he walks to the window the old man comes in, his back to us. We presume it’s a man by the bulk, the gait, and the fedora, making its appearance in the upper right corner.

Or does a man come in? In the 1.77 version, at the corresponding point in the shot, we can’t tell it’s a man until the figure comes further into the room.

Godard’s reliance on the upper part of the frame allows us to discern the caller sooner in the 1.35 version. Seconds, even split-seconds, matter in cinema. Insofar as cropping affects the timing of a shot’s unfolding, it affects our experience.

( 6) Cropping affects perspective, the perceived distances and volumes of objects in the visual array. Blowing up the center of an image creates a flatter, more friezelike space than we discern in the original. This becomes evident in a later phase of the scene I just mentioned. After Bruno leaves the shot, the old man is left standing in the office.

The 1.77 image looks more like it was shot with a long lens than does the full-frame version. The result recalls the sort of perpendicular telephoto framings so common in the 1970s, in films like The Parallax View and The Conversation (below).

Godard has said that he preferred 30-40mm lenses for much of Sauve qui peut (la vie) because a focal length of 50mm (and presumably one longer than that) will “destroy perspective.” (5)

Many of these differences wouldn’t matter in most films, which aren’t composed as meticulously or as daringly. Hollywood images aren’t typically as dense as those in late Godard. (I must do a blog some day on fussbudget filmmakers like him.) But even if these niceties seem negligible, I think you’ll grant that the film would be much more compromised by being shown in a 1.85 ratio, the squarest option available in most commercial theatres today.

Critical discussions of Godard’s late films have treated them as poetic meditations, and that seems partly right to me. Yet few critics ask how they manage to create their lyrical, associative quality. I think, as I hope to show in a future blog, this has to do with his treatment of narrative (naturally) and his layout of scenes. But even before we get there, I think that we find in the very texture of his images (let alone his sounds) a daring decentering of faces and bodies—the usual nodes of our attention. If he often blocks the flow of our glance, it’s in order to rechannel it to unexpected areas and textures, crannies and gaps, within the image. And so we want all those areas and textures, along with the crannies and gaps, available to our eyes and minds.

(1) James Quandt, “Here and Elsewhere: Projecting Godard,” in For Ever Godard, ed. Michael Temple, James S. Williams, and Michael Witt (London: Black Dog, 2004), 126-139.

(2) Geek note: You may notice that this “full-frame” image isn’t itself in the 1.37 ratio. It’s very square. The reason is that many 1.85 frames will be exposed in the camera at a ratio of 1: 1.2! I believe this was standardized for the Panavision cameras of the 1970s and afterward, though I’d appreciate more information about this. See the entry on Panavision cameras in American Cinematographer Manual, fifth ed., ed. Charles G. Clarke (Hollywood: American Society of Cinematographers, 1980), 104. See also Rob Hummel, “Comparison of 1.85, Anamorphic and Super 35 Film Formats,” American Cinematographer Manual, eighth ed., ed. Rob Hummel (Hollywood: ASC Press, 2001), 24-29.

(3) Jean-Luc Godard, “Formats,” Cahiers du cinéma no. 591 (May 2004), 78.

(4) This image is taken from Jean-Luc Godard par Jean-Luc Godard, vol. 2: 1984-1998, ed. Alain Bergala (Paris: Cahiers du cinéma, 1998), 199.

(5) Jean-Luc Godard, “Propos rompus,” in Godard par Godard vol. 2, 466.

Thanks to Suzy Buenger and Nancy Marshall for identifying the Manet painting for me.

PS 15 Dec: And thanks to James Quandt, Michael Kerpan, and Yogesh Raut for a name correction I’m too embarrassed to specify further.

PS 7 April 2008: The issue is raised anew with Gus van Sant’s Paranoid Park. It’s designed to be shown at 1.37, with more than a few shots reminiscent of Godard. Joe Beres explains here.

PPS 25 January 2009: Ranjit Sandhu provides a lively and detailed discussion of aspect ratios and matteing strategies, along with remarks on Godard’s frames.

PPPS 21 Sept 2009: Thanks to editor John Olivio for a correction on the 1.78 aspect ratio.

JLG par JLG (1995).