[From Luc Moullet’s book with Emmanuel Burdeau and Jean Narboni, Notre Alpin Quotidien (2009, Capricci). See Table of Contents]

Brigitte and Brigitte (1966)

There are no hard-and-fast rules, only useful guidelines. Every law must be bent, that’s the only obligation—and that’s precisely what can be great in cinema. Nine times out of ten, what is taught in film schools today is in fact the opposite of what was taught sixty years ago.

Two illustrations as a preamble and a warning:

  • It was once forbidden to move directly from a wide shot to a close-up. This rule is palpable in a film I like very much, Children of Paradise: it systematically uses an annoying—and seemingly forced—gradation between the shot sizes. Today, however, the transition from the wide shot to the close-up works magnificently, provided it is well managed, even if some filmmakers abuse it, starting with Sergio Leone.
  • The 180-degree rule prohibits crossing the shot-reverse shot line so that the viewer isn’t disoriented. But the truth is that it all depends on the actor. With Jeanne Moreau, Brigitte Bardot or John Wayne, you can blithely cross the 180-degree line, because the actor is known and well placed, like a pillar: he or she serves as a visual reference. It is much more difficult with beginners whom no one knows, lack as they do a stable facial reference in the frame.

Like a majority of the technical rules in use, this is also a strictly western law. The Japanese constantly break the shot-reverse shot line; for them the problem does not arise. Imagine: had Japan won the war, that would have been the end of 180 degrees!

This is true of almost all laws: they are dependent on history or geography. None of them is indisputable or eternal. You just have to be aware of the risk you are taking when you decide to apply them. Or not apply them! Conscious of this risk, I have chosen to include in the vade mecum that follows the objections that have been made or that could be made to me. Also those that will be made, no doubt.


Production, Generalities

The plumber principle (choose a title starting with A or B)

Open the phone book, all plumbers have a shop name that starts with A. They all sit at the top of the directory. Being at the top of catalogues is important, because festival catalogues play a big role. All catalogues for that matter. You can’t always do it, but it is recommended, especially for short films. Maybe there will soon be a rush of short films starting with AAA, as with plumbers.

I can already hear the first objection. My first short, Un steack trop cuit (1960), is not exactly at the top of the alphabet: an error of youth, sorry. As for the following ones, Ma première brasse (1981), Essai d’ouverture (1988), Le Ventre de l’Amérique, Le Système Zsygmondy (2000) etc., they are a bit all over; that’s because, with time, I’ve become surer of myself.

Become a specialist

Tony Gatlif controls 80% or even 90% of the market for films about Gypsies. His career is assured. Poljinsky was once the specialist of activist films. Many filmmakers are moving towards specialisation, particularly in genres like crime or erotica. Some filmmakers have a style or a brand image, for example Béla Tarr or Peter Greenaway, or formerly Jacques Tourneur with low-budget horror films. You have a label, it’s convenient, just press a button. I don’t have a speciality, it’s a weak point.

Be the producer’s most expensive project

If you are number one, the producer will take care of your film properly. If not, he is likely to get disinterested in it. I can’t really say I’ve been successful in that respect either.

Inflate the budget and schedule

Chabrol used to recommend inflating the budget and the shooting schedule: 10% over the actual cost, a few extra days… The producer will be delighted if you spend less; he will rub his hands if you finish early. Godard had even made an agreement with his producer: if he completed a film planned for 2 million in 1.8 million, he would get half of the remaining 200,000. He won’t be able to that today, but I have read this in old contracts.

Have several projects

When Chabrol is making a film, the next one is already signed. He took financial and commercial risks with Ophelia, but Landru having already been signed, he was at peace. On the contrary, had he waited for Ophelia‘s release before committing to Landru, he would have found himself high and dry for a few months or even a few years. King Vidor too alternated between personal films and very MGM films, in order to avoid trouble with producers. After The Crowd, he directed The Patsy, which is more of a producer’s film. Chabrol and Vidor have several things in common. They first married their actress, then their script girl—quite a classic itinerary.

It is also recommended to have the “bottom script,” Whatever you do, don’t pitch the script that is close to your heart right away: what will you do if the producer or the Advance on Receipts [1] says no? Offer the producer a difficult script first, then a second one if he refuses, and a third one just in case. The statistics are clear: he will usually capitulate to the third one. Same with the Advance on Receipts. Varda did it; so did I, with a phoney backup screenplay written in eight days. I knew it wouldn’t get the Advance, and I also knew that the third one would be better received. For their part, the people at the Centre for Cinema felt that if that one didn’t get the Advance, I wouldn’t hesitate to present them with a fourth script. So they asked me what I had to propose, and we agreed to select one together.

Scenes not to be filmed

Consider slipping two or three scenes into your script that the producer will remove on his own initiative, or that he will try to convince you to remove for reasons of cost or risk. It’s a red herring. You know that he will try to remove such and such scene, rather than others, which are actually more important to you. It was the same thing with the censors in the past. They used to jump on to highly pornographic scenes, ban them and leave the rest. It was a trick used by Vadim, as well as Truffaut, I think.

Being your own (reserve) producer

My first three films were produced by directors I had defended as a critic: Truffaut, Godard, Baldi. Then I found myself a little short. I don’t think I would have been able to direct Brigitte and Brigitte if I hadn’t made the decision to produce it myself. It’s indeed because I turned producer that I’ve been luckier than other director-critics like Claude de Givray or Charles Bitsch. The best of all worlds is to be able to be a producer when you’re turned down. Otherwise, you are condemned to keep waiting for a sponsor.

Buy back your own films after bankruptcy

This happened to me three days ago. In the course of my career, I’ve bought back six films that I didn’t produce. It’s crazy. Producers don’t last as long as directors. As a director, I first went to the G.T.C. lab fifty years ago. Today, nobody among those who work there knows me anymore.

Les minutes d’un faiseur de film (1983)


Writing speed(s)

I usually write very quickly. Two mornings for a short film, twenty for a feature film. Then I work on the text again. Since it is rare that one can shoot immediately, I have sometimes gone back to a manuscript forty years after it was first written. I am therefore both the fastest and the slowest screenwriter.

Let us point out a danger here: if you are shooting a screenplay written at the age of twenty-two—let’s say a love story—when you are seventy, your point of view on the subject is likely to have changed considerably. There are films that have undoubtedly gestated for too long in their maker’s mind and are disappointing for that reason: Franju’s The Demise of Father Mouret or Fuller’s The Big Red One. What may have become of the Christ that Dreyer thought about for so long? We’ll never know.

Conversely, some filmmakers can only shoot in a state of urgency, without waiting. For Marguerite Duras, it was tomorrow or never. For Godard too. When, at his first meeting with the American producers of Bonnie and Clyde, he proposed starting the shoot within eight days, they did not take him seriously. They ran away. This example is probably a bit radical, but I admit I laughed when, having submitted a script on the other side of the Atlantic, the production asked me how many versions and rewrites it had undergone. A cinema masterpiece like Scarface was written in eleven days! By several people, but still…

Expedients (music, cigarettes, bicycle)

Forty or fifty years ago, I would smoke four cigarettes before I started writing. Or I would put on a record. I would play the same one over and over, to establish a continuity. The process is effective, you don’t get off the subject… But it’s also quite dangerous: anyone would go nuts listening to the same music eight hours a day. “Life Lessons,” Martin Scorsese’s short film from New York Stories, interested me because it revolved around a song that I had played in loop for a few days, A Whiter Shade of Pale. An enchanting piece of music, but rather conducive to inspiration.

At one time, I used to work on my scripts on my bicycle. I would follow a small circuit in the Alps: I would start from my grandfather’s house, take the Saulce pass, continue through the Tourtettes pass located a little higher, then come back down through the Fays pass, the Rossas pass and the Carabès pass. It’s a rather short circuit, barely a hundred kilometres long. When you ride, you draw in a thread of continuous inspiration; there is something that begins and takes shape. Once you get home, you just have to find this thread again. In truth, it’s quite easy: you think about the places you rode through and the scenes come back instantly.

Work little at a time (smart)

I can now work without expedients, but I never write for a whole day. I did it once, on December 26, 1969. I wrote thirty pages of screenplay for A Girl Is a Gun. I usually prefer working in the morning, between eight and ten. As I get older, I sometimes have insomnia. Instead of sitting around twiddling my thumbs, I work from five to seven. A lot of writers sit at their desks in the morning and stop at noon. Two hours is enough to produce something concrete and concise. You get tired after that. The rest of the day, I try to get away from work: I exercise; I have my family business, real estate and other things; I go to the movies.

Work at the last moment (The urgency principle)

This was Godard’s method when he used to write at Cahiers: instead of pointlessly ruminating for hours together and rushing like mad at the end, he would start writing at four o’clock, which was the final deadline for handing in his copy to Rohmer at six o’clock. Let us admire the economy of the process.

Working with a sparring partner

I just mentioned it: directors often stay at the table sighing for two or three hours. That’s why it’s advisable to have a collaborator, or at least someone who watches you, who forces you to work. I had a collaborator for two hours on a scene in An Overcooked Steak in 1960. I worked with Antonietta on Anatomy of a Relationship. I’d like to work with a scriptwriter someday. Gruault, for example. I also admire Jean-François Goyet and Jean Cosmos, who used to do some real research.

Given my individualistic nature, I rarely resort to collaborators. On the other hand, I give the finished script around for reading. I circulate it and ask for opinions. Always separately. And I write down the feedback. If two people have the same opinion without consulting each other, follow it.

In a script, you should avoid lazily sticking to a single basic idea (cf. The African Queen, Peeping Tom, Forbidden Games, Bicycle Thieves).

Write by weight

Take a sheet of paper and cut it into four. Write down ideas on each of the four small pieces. Then place the papers on a weighing scale. A sheet of paper weighs four grams. So a quarter of the paper weighs one gram. When you have 100 grams of script—a hundred ideas, in other words—you can begin. This is usually the method I use, with a weighing scale I bought forty-three years ago, just after the success of Brigitte and Brigitte.

Second step: the ordering. How to order your scattered ideas? Professionals use a wall on which they stick little cards that they move around, assigning one colour to outdoor scenes, another to indoor scenes; one to action scenes, another to love scenes. For my part, I don’t use a wall, but a table that I’ve had since my father’s death.

Start with your favourite sequences, reduce, eliminate

Many write page number one, then page two, then page three… Sure, why not? But there’s something better: ask yourself which sequences are your favourite and start there. Numbers twelve and twenty-seven? Very good. Then move on to the ones you like a little less. After a few days, you’ll only have leftovers to write, the sequences you don’t like. Then, either reduce or eliminate them.

I never start writing a film without knowing what the last scene will be, and even the last shot. Even at the risk of having to change it afterwards. To have a film, you need the end. Then the beginning. And then the middle. You start with five pages, then you go up to twenty. You enlarge with the same canvas, until you reach one hundred pages. Ultimately, the work gets done by itself.

The Smugglers (1967)

Balance of Drama, Management of Information

Vary the length of shots and sequences

This is a remark coming from my dual experience as a filmmaker and a viewer: a good film can fail if all the shots or all the sequences are about the same length. Although there are some interesting things in it, Mes petites amoureuses is disappointing for this reason. Once you know the average length, you know when a shot should end. Is Eustache going to cut before Daniel leaves the frame, or is he going to allow him to leave? Caught up in these calculations, the viewer is distracted from the film. You may object that this is a director’s problem, that the average viewer doesn’t pay attention to it. I rather believe that monotony plays on an unconscious level.

Rule: do not have two long sequences together.

Exception: if you decide to take a chance, be aware that such a linkage is very difficult to achieve. In a nutshell: think music.

Plan the lengths based on the locations

Allow a minimum of half a day or one day of shooting per location. Don’t try to shoot in five different locations in a single day. What if, by chance, one location requires one day + one hour? That is a bit annoying. There’s a whole plan you must come up with for that. If, on the other hand, you’re done too early on a location, you can invent a new scene to shoot in the vicinity. If you plan for it during the writing stage, so much the better. It’s fine to do it afterwards too. In any case, it’s better to decide on it before the shooting begins.

Likewise, make sure you have two or three insignificant places to alternate with the main places; two or three places that have no value in themselves and just serve as a transition. You must be able to allow the viewer to rest from time to time. There is nothing worse than being constantly stuffed with information: avoid the Boorman system like the plague.

Introduce the characters little by little

Not all the main characters have to be in the shot from the start. The viewer needs time to understand who is who. DeMille was good at this. In Reap the Wild Wind (1942), the main actors appear after one, then two, then eighteen minutes. In The White Rose (1923), Griffith marvellously distributes characters and information. In The Taste of Others, Agnès Jaoui juggles with a large number of characters.

(Sidebar: it is not possible to have more than four important characters in a short film. Beyond that, the viewer has no time to get a handle on things).

Exception to the rule: going against the viewer, building the film on disorder. This is a perverse technique, that of the modern film, which requires a certain science. In Heaven’s Gate, the viewer watches the main actors from very far away: Cimino manages to express the loneliness of the individual in general, and of the main character in particular, through the world and through the battle. The film is not all that good, but this element plays in his favour. It’s the opposite in Captain Conan: it’s impossible to get a handle on the characters, even if it’s clear that Tavernier wanted to show soldiers lost in battle, to highlight the “Fabrice in Waterloo” aspect of things…

The principle of delayed entrance is obviously easier to apply with a known actor. Philippe Torreton wasn’t very well known at the time of Captain Conan, so that must have certainly played a part. In Sergeant York, Gary Cooper appears at the sixteenth minute, at the end of an expertly orchestrated crescendo. Someone is standing in the background; the viewer doesn’t know if it is Gary Cooper. Then he turns around: it’s indeed him! The principle of delayed entrance is excellent for the glamour of the character and the actor. Above all, it should not be systematic. Hawks was incidentally the first to take the opposite route: in Red River, John Wayne appears immediately.

Two extreme cases to wrap up:

  • Which is the film where the viewer almost never sees the main character, except for a quarter hour at the end, all the while having the impression of having seen only him? Carol Reed’s The Third Man, with Orson Welles as the mystery man.
  • Do you know how long Peter Lorre is actually on screen in M? I recently did a study. Result: twelve minutes! You don’t see him right away: you see his shadow, you hear his voice, his sniffling… Lang’s genius lies in making us believe that Lorre is indeed the main actor. It’s a challenge, to make a character speak without showing him.

Repeat character names twice

Truffaut used to say: you have to repeat information. The calculation is difficult. Some viewers absorb quickly, others are slow. How can you find a way for the latter to understand without the former getting annoyed over hearing the same thing again? It is usually said that a character’s name must be repeated twice. Or, even better: don’t give it a name.

The law of two-fifths

The high priest of this law—which, for my part, I use too little—was Cecil B. DeMille. He often had his scripts restart after two-fifths or 45% of the film. A typical example: Fool’s Paradise (1921), which begins in Texas, then moves on to Siam without any transition or apparent logical reason, or only with a very artificial one. The film can restart in a more plausible way as well. This is the case with Male and Female (1919): a family leaves its living room, goes on a cruise, then is washed up on a desert island. The same is true of Madam Satan (1930), a great film in which a couple is suddenly transported into a zeppelin that hosts a fashion show, then goes up in flames. Or how to go from domestic vaudeville to disaster cinema.

This was quite common in the 1930s: when a film wasn’t dynamic enough, it was made to branch off into a completely different track. In Made for Each Other (John Cromwell, 1939), the romantic story suddenly changes direction after an hour and ten minutes: someone falls ill and the couple flies across the country. Will they crash? I’d bet the decision didn’t come from the director, but from RKO, who must’ve demanded something more to spice up the film. Same thing with History Is Made at Night (Frank Borzage, 1937), a rather accomplished romantic comedy featuring Charles Boyer. The ending depicts the sinking of the Titanic. A surprise is guaranteed. It has nothing to do with James Cameron’s film. The viewer of his Titanic knows what awaits him. Here, not at all.

The law of the seventh reel

Variant of the law of two-fifths. Truffaut once said that the crucial point is the seventh reel. That is to say, after about an hour. If it’s a DeMille film, that’s about the two-fifths point. The idea is to restart around the midpoint. Sometimes a little before, sometimes a little after. Clouzot’s The Wages of Fear is excellent on this front. For an hour, the action is confined to a sort of truck depot; almost nothing happens. Then, in a car! The film is outdated, but the dramatic conception is quite good: assemble everything indoors and then take it outside.

I did not obey this law in Origins of a Meal. The very idea of the film has a force, which prevents it from going elsewhere. On the other hand, I obeyed it in The Smugglers: at one point, the film restarts with the help of typical adventure film music. I should pay attention to this law more often. But since my films are much shorter than those of Cecil and François, I feel less concerned by this problem.

Ma pemière brasse (1981)


Division of labour

On location shoots, I got into the habit of picking up the crates and carrying them to the van at the end of the day. I usually pick up the heaviest material. If there are fifteen of us and each one carries two suitcases, everything goes much faster. Otherwise, it can take a good hour to pack up. But be careful: this kind of intervention can be misinterpreted by stagehands, especially the older ones. It’s best to ask their permission. They feel devalued and are afraid of being called to order by the producer, if by chance the latter catches the director holding two crates in his hands.

I’ve seen some very beautiful things, the star picking up the tripod… I can’t see Tom Cruise doing the same. On one of my last films, I even saw the sound engineer carrying the camera and the tripod! Remember that the rivalry between the sound engineer and the cameraman can often be very intense. The former likes to stand as close as possible to the edge of the frame, four or five centimetres away. So all it takes is a bit of improvisation in the pan shot to get the two of them arguing over questions of principle. The sound engineer has to anticipate the cameraman’s reflexes and the cameraman has to make the sound engineer’s life easier instead of fighting with him.

To move a dress or a prop, it had to be sometimes determined whether it was the prop master’s job or the assistant’s. The director had to intervene, which is always annoying.

On small films, the division of labour is less rigid today than it was in the past. There is often more solidarity within a team of five or six than within a team of twenty.

Prepare two découpages (one of them an express version), especially for outdoor scenes

Prepare two découpages in case of a difficult scene, especially in the mountains. A decoupage of thirteen shots—that’s my usual rhythm—and a decoupage of seven shots, to be able to deal with the most pressing situations: rain, thunderstorm, unexpected clouds… If there is a little time, you can even have fun planning a decoupage in three shots. I like it.

Cost-effective shots: between forty and seventy seconds

Allow an average of 40 minutes of work per shot. The duration is about the same for shots of five or thirty seconds. As a result, if you only shoot five-second shots, the shooting may take a long time. That’s why it’s good to know that the cost-effective duration, in terms of production, is in the range of forty to seventy seconds. Beyond one minute, the shot poses too many problems for the actor’s performance. If you have completed five one-minute shots in a day, you can consider yourself satisfied. However, this regularity may be tedious (see above).

Don’t get tied up with a single goal

You have to have a goal when you film a shot, but you also have to think about everything that can come corrupt it. Too often, the director only thinks about what the actor is doing. But not about the background and the meaning it can have. Or about the supporting roles and gestures in the background (see below).

Stay calm

The director is often feverish, especially when he is just starting out. I can present a film in front of a thousand people without flinching, but I always have some apprehensions when starting on a film. As Lubitsch used to say: dear director, don’t worry if you shit in your pants when you start shooting. I shit my pants every day.

There are times when you feel that everything is going wrong and that it is better to liquidate an actor, but that is not always possible. And in the rushes, it’s great…

At other times, you think everything is running smoothly, but what you thought was great may turn out to be rubbish.

So you have to keep a certain distance and not get angry then and there. See how it looks, while keeping at least an apparent calm. Over four or five weeks of shooting, your staff will be reassured that they have a serene director at their side, who respects his work plan and is reluctant to raise his voice. Sometimes you should not hesitate saying “what am I going to do?” You can share your uncertainties with the sound engineer or the cinematographer, but not with the actors.

The director must not talk to the assistants to cameramen, sound engineer, electrician

Talk, yes, but not about the film. About the rain or the good weather, as an alternative. Most importantly, do not talk to the electrician without going through the cinematographer: watch out for the short circuit! There is an order to be respected. On the other hand, the director can have an intimate conversation or sexual intercourse, for some, with the assistant cameraman. But the director must absolutely not talk to them about work. Unless there is an urgent technical problem.

Make room for silence after gags

There are often more or less improvised gags during my shoots. Jokes that make the cameramen, and sometimes the sound engineers, laugh. It’s a bit tricky. Even if it’s a good sign to make the technicians laugh, it’s best to pick phlegmatic cameramen. I interview my wife at one point in L’Empire de Médor (1986): “Why did you marry your husband? —Because he protects me well from dogs.” The crew laughed. So you have to calculate, or forewarn. There are techniques for controlling laughter: pursing your lips, pinching yourself… You can always drown out unwanted laughter with music or background noise, but it is not ideal.

Pay attention to the calculus of laughter during the general conception of a film, see to it that the gags aren’t mashed together. Same thing if you want to make people laugh during a post-screening debate: calculate the pauses. These are music-hall techniques that it can be a good idea to master.

As I have already said, the viewer must be put at ease. On this subject, you can refer to my article on bravura sequences (in Piges choisies): I explain that spectacular films must end with something rather soft, a “slow curtain fall,” thanks to which the return to the norm will not be too brutal for the viewer. Typical examples: the endings of Samson and Delilah, Land of the Pharaohs or Ruby Gentry. Some directors favour the ecstatic final shot, but the rupture is such that the return to reality can be dizzying. A danger.

The director must change tables at each meal

You have to take turns. Even if crew members see through your calculation, they will be happy to see you have lunch with the electrician one day, leaving the star alone in his/her corner for once. This is especially true for large-scale shoots, where technicians and actors often tend to form separate clans.

It remains to be seen whether it is really a privilege to share the director’s table. Nothing is less certain: the greatest filmmakers are often quite detestable in private. See Pialat, or Godard. Having said that, you have to know how to handle Godard. My brother, who played in Les Carabiniers, told me how. You have to make him laugh, light up the child in his heart. He has the soul of a child, and it’s excellent for him to come back to it.

On the other hand, bad directors are often very pleasant dining companions. Alain Robbe-Grillet, in my opinion the worst filmmaker, was a delicious companion; you had to fight to sit next to him. He put all his genius into his life and nothing into his films. Similarly, Claude Sautet’s films are terrible, except Hello Smile! (1955). But in life, magnificent. I’ve had excellent relationships with bad directors, Louis Daquin, Jean-Paul Le Chanois, Jean Dréville… Great friends: we would tap each other on the shoulder, we were on first-name terms. Either you’re a living being, or you’re a director, it’s difficult to combine the two. Renoir managed it, even if he was quite hypocritical. His principles were effective, but rather twisted.

It is said that, during trips, no crew member wanted to sit next to Naruse because he never opened his mouth. There are times when I don’t open my mouth: when I am shooting, I think about things. On the other hand, when I’m travelling, there’s no problem, because I act as a guide. I know all the places: in a kilometre, you will see such and such church, lean over to see such and such thing… I think I’m better than Naruse in geography. As for the rest, I don’t know.

A Girl Is a Gun (1971)


Director-cinematographer relationship

In “Jaundice” (in Piges choisies), I talk about the case of cameramen having command over spineless directors who don’t quite know what they want. It also happens, more simply, that the cameraman comes to fill a gap.

Sometimes, the conflictual relationship that the director has with the cameraman can paradoxically help the film. I had problems on Anatomy of a Relationship with a cameraman who started performing stunts, zooms that I didn’t want at all. It served me well: against him and for the film.

There are some technical details that hardly interest me. I don’t know everything; I can’t talk about the choice of filters. So I sometimes give the cameraman carte blanche. When I ask him for photography that looks like Far from the Madding Crowd (made by Nicholas Roeg, 1967), I give him the cassette. If he respects this precise model, that’s perfect.

In any case, this is indisputable: you have to trust the cinematographer, for the simple reason that a director shoots for four, five weeks a year, and a cameraman for twenty or thirty weeks. He has more practice. You have to take his experience into account.


The backlight has several significant values. They must be studied. The main ones are myth and anguish, the valorisation of an enigmatic or all-powerful character. But backlighting also allows you to consign what is unknown or suspicious to the shadows, for example a murderer.

There are then several technical solutions. You can use real backlighting or reduced lighting. You can film the whole frame with the correct amount light, except for the actor you are interested in. If you go for a shot-reverse shot setup, there will almost inevitably be backlighting; you might as well take advantage of it and play on it. For this reason, backlighting is very important in Westerns. It’s very sunny in the Rockies, I stayed there for two weeks without seeing a cloud. Or, on the contrary, there are clouds for very long periods.

Choose: sun or no sun

Imagine that you are shooting a sequence with a lot of shots. You start shooting without the sun, then the sun comes in. Do you go on? A sequence shot is obviously one solution. Or a match cut with a scene set in a different location. That way, when you return to the previous frame, the viewer won’t pay attention to the change in brightness. But this is not always possible.

Sun or no sun: it’s a headache that encourages a filmmaker to shoot in regions where the light changes little. For example, in the Midi region of southern France, even though I remember waiting for an hour for the sun to go away for a shot in A Girl Is a Gun. In Ecuador, there is a permanent cover: the sun is always present, but behind tropical or equatorial clouds. You can see it on the light meter: from six o’clock in the morning until noon, the light rises very regularly and then falls again until six o’clock in the evening. That is ideal. Except for cameramen: most of them prefer regions that allow more light effects.

No more than two “golden hour” shots per day

Twilight lasts for thirty to forty minutes. In Brazil or Zaire, it’s even less: ten minutes at most, from five to six to five past six.

One of my students at the Fémis had planned seven “golden hour” shots for one day of filming. I insisted that it was impossible and advised him to line up two shots for the first day and two more the next day. It makes no sense otherwise.

(A brief digression on the Fémis or ex-IDHEC. As it is not easy to pass the school’s entrance exam, there was a time when students considered themselves stars. They would come to class a quarter hour or even half an hour late. A debutant would want to do everything in three days of shooting: outdoor shots, indoor shots, indoor shots with and without light, blurred shots, tracking shots. He would try to make a visiting card, to prove to himself that he was capable of accomplishing everything in every field. But this is not possible. And it is to the detriment of the film. I remember ticking off students, who have made a career since then, because I told them that you can’t do everything. I was unpopular: I was reproached for discouraging tracking shots).


Friendly tip: avoid studio effects in a neorealist film. Self-evident, perhaps, but sometimes forgotten. A Steadicam suddenly appearing in a biblical film can be surprising. Because, you see, there was no Steadicam at the time. There was no cinema either, of course, but the viewer doesn’t think about it if the film is successful. A camera movement could remind us of it. The Ten Commandments, for example, looks like a futuristic film, even though common opinion is more likely to situate Moses in the past than in the future. This is a problem. Unless the work is a self-conscious, ironic comedy, something that DeMille was absolutely unaware of.

The discrepancy between the artificial and the natural is related to what some call “the tracking shot of Kapo,” In a nutshell: avoid discordant effects. A great director could possibly have taken the liberty of zooming into the barbed wire of a concentration camp. Abel Gance, or perhaps Samuel Fuller. It takes a genius to make these effects work. There are lighting effects and filters in Bicycle Thieves, or in Michael Powell’s films, which are at the limit of what is tolerable.

No white at the bottom of the picture

This was the only recommendation I made to my cameraman on Land of Madness. Bergman knows a thing or two about it, the man whose films were mainly distributed abroad. It’s a provision to be made with subtitles in mind. Yellow-coloured subtitles got Bergman off the hook, but those can’t be made any more. Down with progress.

Remove a colour (which then appears suddenly)

On A Girl Is a Gun, I had planned to eliminate yellows for half an hour and then have it appear at some point. It’s an attempt at an effect. A character could be primarily associated with one colour. If he changes colours, it’s because there’s something new. Godard did a remarkable work with the fluorescent colours of video in The Power of Speech (cf. “The Cosmic Film,” in Piges choisies).

Shortly after the birth of colour films, when the tendency was to really sock it with Technicolor, the opposite reflex may have existed: going for black-and-white in colour. That was Dreyer’s plan for his film about Christ. It was probably with this in mind that Stevens, when he shot a Christ story, played a lot with black and white, even though it was a colour film. Cecil’s film about Jesus, The King of Kings—not a very good one, by the way—was elaborately worked over: only shades of black. Christ is a character that inspires a certain sobriety.

TV blacks

Dark areas in colour films, gradations of blacks work fine in cinema, but hardly on television. They become “totally dark,” I saw The Blind Owl without colour, I saw absolutely nothing. There could even be a kind of TV censorship: “Don’t do things like that, it won’t work.”

In Africa, darkness is frowned upon. The viewers get the feeling that there is no picture, they feel robbed and break the seats. The Carthage Film Festival, to whom I had sent a copy of Origins of a Meal, had arbitrarily cut out the little black frames between the shots. “We can’t keep them, the audience will be furious,” I was told.

I had the same problem in Peru, where I had brought a trailer of Vecchiali’s The Strangler. When the distributor saw that there were a lot of dark areas, he refused to buy the film. To buy a film with dark areas is to buy a film that is partially invisible. Similarly, in 1910, theatre owners protested when actors’ bodies were cut off at the thighs or the chest, going so far as to ask for a discount on prints.

Le Prestige de la mort (2006)


How to fill it

Whatever the format, widescreen or 4:3, you must know what you are putting into the frame. An old practice, inspired in particular by Victorian art, was to fill it to the maximum. The champion at this was Cecil B. DeMille, with the VistaVision of The Ten Commandments: not a single corner of the picture without a character or an object.

The modern approach is to isolating a thing. This becomes much more striking from 1953 onwards, with the invention of CinemaScope: sometimes a character stands alone in a corner of the screen—to emphasise the loneliness of man—and sometimes, on the contrary, he appears in the middle of others. Between one and the other, variations and alternations can have a lot of effect.

Nicholas Ray is probably the filmmaker who has emphasized the widescreen the most. Not in his blockbusters, but in his films featuring real-life characters: Rebel Without a Cause, Bitter Victory or Bigger Than Life.

Batches of two or three shots

For those who don’t know the problem: A is in one frame, B is in another, you film eight shots on A, then eight on B, in a total break with the chronology. This is what we call “batching” (‘faire les axes’). Should you do it? Should you not? A serious dilemma.

American studio cinema—and Austrian-born filmmakers in particular—used to batch shots together. Either out of interest, like Fritz Lang, with his little sketches and his angles, or out of necessity, like Edgar Ulmer. The possibility of doing an end-to-end rehearsal without a camera probably made things much easier.

That said, the practice is tricky, especially if the reactions of the characters are important. You must preferably have super-professional actors. You absolutely can’t do it with Claude Melki, who improvises constantly and whose reactions are unpredictable. But I don’t see Lang filming with Melki, or even with Jean Abeillé.

Cameramen appreciate it when you batch shots together, the actors less so. It takes formidable memory to guess your partner’s sixth or eighth line! That’s why I advocate a middle path: instead of filming all the shots in one axis one after the other, or, on the contrary, chaining the shots together like tying shoelaces—right, left, right—I do it in pairs (two shots on one axis, then two shots on the other).

TV margins

A film is never shown on TV in the exact format. Some areas are not guaranteed, on the right and left of the picture: “PHILIPPE NOIRET” can become “LIPPE NOIRE,” “BERTRAND TAVERNIER” “RAND TAVERN,” The examples are countless, I wrote an article on this. In the same way, elements get lost, pixels appear less. So you should work out not only your frame, but also a sub-frame for TV. It’s a headache. What’s more, the 1.85 aspect ratio is shown in 16/9 (1.78), and the 1.37 ratio in 14/9 (1.55). Enough to tear your hair out. All films calculated to down till the last detail—for example, Jean-Pierre Lajournade’s La Fin des Pyrénées—are ruined when they are shown on TV.

You should even plan for margins for theatre screenings, unless you wish to cut the heads off some of your actors.

Doing the Dutch angle (or refusing horizontal framing)

The exercise is difficult, few filmmakers know how to master it. Hitchcock is the best at it, thanks to his storyboards, his ability to anticipate everything. Kazan, on the other hand, does a terrible job in East of Eden, particularly in a scene featuring Jo Van Fleet and James Dean with ten shots-reverse shots filmed from a Dutch angle. Duvivier used the Dutch angle like a madman: in a twenty-minute stretch of Life Dances On, every shot is canted. It used to be artistic; today it rather makes you laugh: canted shots, blurs, flashbacks…

So be careful. Don’t use the Dutch angle for more than one or two shots, and above all avoid canting the shots during editing: it’s a recipe for disaster.

Le Fantôme de Longstaff (1996)

Casting and Matters of Performance

Choose an actor from the character’s social class

Even if actors like to play the opposite of what they are, you have to use your reason. Montand is ridiculous as a professor in Delvaux’s One Night… A Train. It’s not his cup of tea at all. In any case, Montand shouldn’t have been walking around with a cup of tea… Katharine Hepburn in the role of a washerwoman or a traffic constable? No thanks.

Beware of discrepancies in general. A vaudeville actor dreaming of playing Jesus Christ or a head of state? The result is likely to be shocking. You can of course bet on it, suggesting that Jesus Christ was a vaudeville hero in his own way. Why not? But the risk remains. Likewise, avoid taking the easy way out of casting based on reputation.

Study the six actors in one

Carefully consider what an actor’s right cheek (1) or his left cheek (2) expresses, what the viewer feels when he is facing front (3), when he is facing the back (4), when he is heard but not seen (5), when he does voiceover (6). Work on the actor, observe him in front of a mirror, have him turn around to get a clear idea of the effect produced by each of these possibilities. The back is crucial: Raimu, Jannings, Marlon Brando in A Streetcar Named Desire, all massive actors use it a lot.

Two personal memories:

  • Sabine Haudepin, with whom I worked on La Comédie du Travail, has cheeks that are very dissimilar: she doesn’t like them, precisely because of their asymmetry. However, this can be a great advantage, which is one of the reasons why I enjoy filming with her.
  • I had planned the decoupage of Longstaff’s Ghost around the face of Italian actress Laura Morante. I had to modify it after she pulled out, and the actress finally cast—Iliana Lolic—had a face full of contradictions: different cheek, different emotion, different shot. So be careful: a storyboard can prove useless, even harmful, if there is a change of actor or actress before shooting begins.

Avoid the charismatic actor who disappears quickly

It is such a truism that I will limit myself to recalling the canonical counter-example: the death of Janet Leigh halfway through Psycho (cf. “Maps and Habitats“).

Think of the actors in the background

An exercise: how to film five characters in the same frame, two in the foreground and three in the background? Above all, don’t focus exclusively on the first two. The other three are just as important: you have to think about them, or else eliminate them altogether. You can very well decide that they will be doing nothing. But they must visibly be doing nothing. It’s a conscious choice.

I applied it in Les Sièges de l’Alcazar. It was fun. There were five people in the frame, including one at the centre. Strict instructions for him: not to move or even blink for a minute. That’s quite difficult. But the effect is guaranteed: the viewer will believe that the character is blind or paralytic. He will laugh.

Contrary to popular belief, a fixed actor can be very expressive. How often have we seen a supporting actor gesticulating like a madman in the foreground, with Cary Grant next to him doing nothing? The laughter then changes sides: it is the gesticulator who becomes ridiculous. (Note: always plan for moments when an actor does nothing).

Combining underplaying with overplaying

I tried it out in The Litre of Milk: one actor overplays, the other underplays, which corresponds to the opposition of two worlds. In the audience’s eyes, the actor who doesn’t perform has an edge over the actor who does. Think of Robert Mitchum. Think of Marie-Christine Questerbert in Anatomy of a Relationship: with lesser experience, she got the viewer’s sympathy more than I, who worked a little too much on my effects.

This is in line with what I said earlier: you shouldn’t cut too quickly; you have to go all the way to the end of the shot. You never know what might result from a difference or a mismatch in performances. Let it roll.

The actors should not deliver everything at once

I will stick to the case of the famous French actor Gérard Depardieu. At the beginning of a film he stars in, the viewer is usually a little surprised. He may even wonder how an actor with so little involvement could have acquired such reputation. Then with each passing scene—the third, the fourth…—Depardieu delivers more and more. In the fifth, the viewer is convinced, he can’t take his eyes off the actor anymore. An excellent strategy: deliver a little at first, then gradually build up.

Diction: the viewer doesn’t like an actor who can’t be understood

A peculiar consequence of this, and good news for me, is that the audience prefers old actors to young ones.

Conversely, the strength of an actor like Yves Afonso, the pleasure he can instil in the viewer, stems precisely from the fact that the latter understands nothing of what he is saying. He does it on purpose, and his lines generally have little importance in themselves.

Keeping the actors busy

In order to bring some spontaneity, you can make an actor perform while eating, walking or smoking. You can make him stutter, or put a matchstick in his mouth. The payoff can be considerable: if a filmmaker decided to feed her throughout a film, even Sophie Marceau could be a good actress!

Dialogues, especially in movement (and not necessarily face to face)

Two examples to meditate on at home: the dialogue between James Stewart and Margaret Sullavan, facing away from each other, in The Shop around the Corner (1940); Louis de Funès in a swimming pool, trying to hold a conversation while sitting in a small rubber dinghy spinning endlessly, in The Brain (1968).

Work on overlaps

American comedy, especially Hawks, excelled at making two actors speak at the same time. Welles was also very good at this game. Come to think of it, maybe he was the best.

The aces of overlapping dialogue know the golden rule: even if five or six lines collide, the few important pieces of information must be perfectly audible and understandable to the viewer. Hence the need for a great deal of planning, lots of rehearsals and probably also lots of takes.

Les Sièges de l’Alcazar (1989)

Direction of Actors


I am often criticized for not giving enough instructions. My attitude is similar to Chabrol’s: the moment I choose actors before shooting, and even when I write for them, the essential thing is accomplished. In some cases, I have made five to ten films with them, so I don’t need to give them a lot of instructions. Maybe a little something…

Chabrol orients actors, but doesn’t direct them too much. When Karina asked him what her lines were, Godard replied: “Listen, don’t make me sweat, you have a mouth to talk, go ahead!” Beware of directors who over-direct: On the Waterfront, which earned its actors several Oscars, remains a model of excessive overplaying.

I believe that, in a film, you don’t direct most of the actors, but that you direct some of them.

Rehearsal diplomacy

Actors like to rehearse movements in their entirety. But the cameraman, for his part, wants to be sure that what will be said or done will match his marks, his lights. So he could be confused if the actors don’t yet know exactly how far they will move. Whom do you favour, the technicians or the actors? In principle, the actors. But be diplomatic.

The desire for the Italian Run

An Italian Run is a barebones rehearsal. Actors have been wanting it more and more for the last twenty years. This is understandable, but the risk is the formation of automatisms that won’t work at all on location. How do you redo a scene rehearsed in vitro, around a table, in a roubine? It’s a problem I’ve encountered on several occasions. My recommendation is as follows: rehearse in a bare location, empty of any object that could establish a misleading reference. Or, more radical and simpler: rehearse a scene that you won’t shoot.

Say “Action” or “Go” based on the film’s tone (without a megaphone)

In France, directors say “Action,” This is a problem since the actor does not necessarily act. But when he hears this word, he is tempted to force himself, to act in too strong a mood. I recommend a “Go,” or a very feeble “Action,” And always tell the actors to begin, not on “Action,” but two seconds later. It’s very good for continuity.

Pay attention to line beginnings

The problem of line relays is a very delicate. It’s customary to begin a shot with a few words from the previous one so that the actor is in the right tone. The actor must know that what is said at the beginning will not be retained during editing. He must be placed in the right ambiance, and at the same time, he must know where the break is. Otherwise, you won’t get a clean cut at the beginning or the end of the shot. There must also be a determined starting position. Otherwise, there is a big risk of a mix-up.

Think about secondary actors

A remark drawing from my own experience as an actor: they will happily ignore you when you are a secondary actor or an extra. It’s usually the first assistant’s job to take care of you. He does, more or less, but sometimes he is very busy. Without any instructions, you find yourself stranded, you don’t know what to do. That’s very bad.

I, for one, can address the director directly if push comes to shove. But if he doesn’t know that I’m a filmmaker myself—I experienced this with Sergio Corbucci—or if it’s an unknown person who calls out to him, it can cast a chill.

Compliment after takes

Make a precise remark, say what interested you. Find a kind word. Slip in a “bravo,” congratulations. It’s necessary, but not always easy. Renoir’s technique is well known today: after giving the greatest compliments in the world, he would always ask for an extra take, “a backup take, just for me, for the pleasure of it…,” You have to vary; the actor shouldn’t know your tricks.

This is not a strength of mine: I often forget to congratulate the actors. After a shot, I mull over it, I’m not absolutely sure of myself. As I usually cast good actors, I tend to assume that they are good. Don’t be like me.

No more than three objections (only one for beginners)

If you have a problem with an actor, take him aside. Don’t criticize him in front of others: his sensitivity is likely to be hurt. And don’t make more than three objections: beyond that, he will get confused. I’ve seen Clouzot in action: he was always finding things to criticize, sometimes making five or ten objections. He would stop the actor in the middle of a rehearsal, or even in the middle the shot. A real nightmare.

Here’s what I recommend: go over the entire take to identify one or two problem areas. Not ten—get another actor otherwise! Do one rehearsal of the take, focusing on the main points, and then possibly another one to study the secondary points. But not all at once!

There are countless techniques to handle this kind of situation. The list would be very long, it will be the subject of a second volume. Truffaut had an excellent trick. He would tell the little-known actor preparing to reply to Depardieu, “Try to push Gérard a little, I feel he’s lousy this morning.” The little actor doesn’t panic any more, he thinks that Depardieu is to blame rather than him, he becomes relaxed.

Do not eliminate close-ups

Fatal mistake: going to an actor to tell him that, all things considered, you have decided not to shoot the close-ups planned for him. He will think that you find him deplorable, the consequences on his morale could be disastrous. It’s better to shoot the close-ups, even if means not retaining them during the editing.

It is important to know that most actors live on the number of close-ups and lines of dialogue. One actress turned down a role I had planned for her on the pretext that she didn’t have a single line of dialogue: you can, however, do fascinating things without talking. Television, on the other hand, often pays according to linage.

Give replies off-screen

A great actor, even when he is not being recorded, should give replies off-screen. Otherwise, the co-actor is likely to be ruffled. I’m sure Depardieu does it when he performs with a beginner. It’s a basic form of courtesy, which has more or less become part of cinema’s mores. Likewise, the co-actor in the scene should be there in case of a close-up without lines, in order to help the eyeline (classical American system used doubles: this is not so good).

Have the actors wear hats

When filming outdoors, the actor must wear a wide hat to protect himself from the sun. He shouldn’t arrive white and end up black. Professionals know this, they love to wear a hat.

Anatomy of a Relationship (1976)


Avoid credits when the film is underway

I can’t stand credits that cut into the action, or those that take place during a car trip: it’s of no interest. A certain snobbery has the credits go on for eight to ten minutes: it’s stupid.

Avoid small gestures at the end of a shot

In view of editing, each shot must end with a specific gesture. If a character in the background attracts the viewer’s attention, you must cut to him: a potential problem.

Sonic ambiance

In view of sound mixing, you have to think about sound from the shooting stage onwards. Consider recording compatible sounds right after the take. When the sun is out, the crew would want to go on as quickly as possible. That’s why the sound engineer must know how to get working between takes, in order to record the ambient sound immediately. And above all, don’t wait until the end of the day: wherever you are, the sound at six o’clock is not at all like the sound at ten o’clock.

Note the particularity of each sound recording

This requirement corresponds to a simple truth: human beings perceive the difference between visual takes better than between sound takes. Is it because silent cinema preceded the talkies? Yes. And no, because today’s filmmakers haven’t experienced the time of the silents. In any case, the acuity of the ear is less than that of the eye, except in the case of the blind, who can guess an image from the noise accompanying it. They would make the best sound editors in the world.

Begin the music after the emotion

Music can come to echo an emotion produced by the visual: not on the visual, but a little afterwards.

Too few filmmakers think about it. And too many are those that lean on music. This is quite apparent with the English, with Reed (The Third Man) and particularly with David Lean: Rachmaninov for A Brief Encounter, the music of Maurice Jarre, The Bridge on the River Kwai… Some films have had great success thanks solely to their music and nothing else, for example Dances with Wolves. Dead Man, Jarmusch’s film with Mitchum, and La Patellière’s films lean on music as well.

Background music can mask sound flaws

I shot a scene in The Litre of Milk on a road. I realised too late that there was gravel on this road. Or they had been added since my previous visit. So I had to ask my brother to compose some music. Moral: the more the gravel, the more money my brother makes.

When shooting in Paris, you are forced to use equalizing music because the background noise is too loud. Or you can do like Godard and stake everything on reality. But it is a provocative effect that he is the only one, or almost the only one, to have pulled off.

In the studio

For dubbing, voiceover etc., you must speak standing up: you breathe better, it’s much better. I never record a voiceover sitting.

You also need to be careful while dubbing: your studio voice will inevitably be very different from your voice recorded outdoors.

Lapel mic

You distinctly hear what each character says, you are close to their mouth…: there are several advantages. The disadvantage is that the sound you get is hollow, because it is recorded right against the chest. Lapel microphones can work with two people in a room, but not at all outdoors. But a boom pole is not always easy to use.


A character leaves the frame from the left and re-enters from the right. It’s common to do so in the visuals, but if you do it with sound, it is likely to come across as a gag. So you have to be careful with Dolby.

Similarly, Dolby gives the impression of a disproportionate volume of sound in a small room: eighty-six square feet echo like a palace!

L’Empire de Médor (1986)

After the Film

Attentions and gifts

The director should not be all “work, work,” He should give small gifts from time to time, for example to production assistants. When you start working with a new producer, it is advisable to visit their office, to study each other’s personalities discreetly, to try to establish friendly relationships.

Truffaut taught me a lot on this front: sending a postcard to the actors for New Year or a little note to Doniol, our editor at Cahiers, when his child was born. I would never have thought of it on my own. Truffaut was unbeatable in this respect. At the homes of friends and actors, I’ve found a number of postcards he had written. The use of cards has somewhat reduced; today people write much less.

Testers (or guinea pigs)

For Land of Madness, as I had gotten a head start with the editing, I was able to organise a few screenings with a small range of testers. Two or three, no more. Truffaut even went so far as to organise screenings for just one person. Once the film is finished, everyone should immediately get a tester: I invite one to lunch, the editor invites another… It is imperative to prevent the testers from talking to each other.

Don’t screen the film too much

With video, it has become rather difficult not to screen a film. A Girl Is a Gun sold rather well, on its title alone (for more details, see “Heights and Chances” – next chapter). Contrary to what is generally said, the ideal is to never show your film. Or to tell the potential buyer that it’s not quite finished: he’ll think you’re going to improve on it. An excellent technique.

Get a laugher

You just have to go around during screenings to spot them. If the laugher laughs on purpose, that’s good. If he spontaneously applauds, it’s even better for the film. Jean Domarchi was formidable: he would laugh continuously, sometimes right from the credits, especially at Lubitsch’s films. Jean-Pierre Léaud too: very good laugher.

Have an accomplice

A “baron” is the commercial term for an accomplice when you sell something, for example on the street. He rushes to buy your product, making other potential buyers want to buy too. My wife won’t step into a restaurant if no one is eating there…

The baron may or may not be a volunteer. An involuntary baron is even better. If a well-known distributor—possibly paid for this purpose, but not necessarily—rushes up to you at the end of a screening to tell you how much he is interested in your film, the curiosity of others will increase tenfold! Upon hearing the viewers’ laughter at a Cannes screening, a distributor took me to the exit: he wanted to sign right away, before others could make an offer. I seem to remember that he was a little disappointed when he saw the film six months later (cf. “The Maoists of the Centre du cinéma,” in Piges choisies).

Note: the baron can be a laugher, or the laugher a baron, but it is advisable to split up the tasks.


A critic who defends a film because a colleague has said good things about it.

There is no shortage of conformists in the profession. They must be spotted, and those who think for themselves must be invited to the first screenings. The conformists must wait for the last screening. Imagine a conformist coming alone to the premiere: disoriented, the poor guy could go completely haywire and hate the film!

It’s fun to fool the critics. Alas, people don’t always think about it.

Say bad things about the film during debates

(I shouldn’t say this, as it’s a trick to pull on the audience).

It may be a good idea to attack your film yourself during a post-screening debate. The viewer will automatically try to reassure you: it’s not that bad, you are wrong, in fact there are good things in such and such scene that you don’t like… Faced with those who praised Hallelujah, a prestigious and much-admired film, King Vidor would answer that he had bungled a certain number of things. John Ford takes the cake: “Stagecoach? Are you sure it was I who made it?”

Another advantage: this strategy tempers directors’ obsession with praising their own films. Even when they pretend to downplay it. The truth is that you shouldn’t step on their toes too much.



[1] [Translator’s Note]: Advance sum given to filmmakers by the National Centre for Cinema (CNC) against the film’s prospective receipts. For more details on the CNC’s film funding, refer to Moullet’s essay “The Maoists of the Centre du cinéma.”