Filmmaking & Screen Production

Three phones and a Sundance premiere

Sean Baker shot the feature Tangerine on three iPhone 5S handsets. The crew clipped Moondog Labs anamorphic adapters over the phone lenses to get a widescreen image, ran the FiLMiC Pro app to gain manual control of exposure and focus, and stabilised long walking shots with a cheap gimbal and a bicycle. The film premiered at the Sundance Film Festival in January 2015, was picked up by Magnolia Pictures, and played in cinemas. Its reported production cost was around one hundred thousand dollars, which is roughly what a studio film spends on catering in a fortnight.

The lesson people usually draw from this is that the gear does not matter. That is half right and it is the less useful half. Baker did not skip anything. He developed the project for months, cast it, scouted a specific stretch of Santa Monica Boulevard, scheduled it, shot it in about three weeks, cut it, mixed it, graded it, cleared the music, and delivered a master that a distributor could actually release. The phones changed one line of the budget. Everything else in the process was the same process a hundred-million-dollar film goes through, run smaller and faster by fewer people.

That process is what this lesson is about. Before you learn a single thing about lenses or lighting, you need the shape of the container those decisions sit in, because almost every avoidable disaster on a first film is a scheduling or a phase problem wearing a creative costume.

Key idea: Every film, at any budget, passes through the same five phases. Low budget does not remove phases; it removes people, so one person does more of them.

The five phases, and the decision that ends each one

Productions are conventionally divided into development, pre-production, production, post-production, and distribution. The useful way to hold them is not as five boxes of activity but as four gates. Each phase ends with a specific decision, and confusing which side of a gate you are on is what wrecks schedules.

Phase	What actually happens	The decision that ends it
Development	Idea, research, treatment, drafts of the script, attaching a producer, a director, sometimes cast; assembling money or deciding to self-fund	Green light: someone commits the money and a start date
Pre-production	Breakdown, budget, schedule, casting, crewing, location scouting and permits, design, rehearsal, insurance, equipment	The first day of principal photography arrives whether you are ready or not
Production	Principal photography: shooting the scheduled scenes, day by day, on a call sheet	Picture wrap: the last scheduled scene is shot
Post-production	Assembly, editing to picture lock, sound design, ADR, music, mix, visual effects, colour grade, titles, delivery masters	Delivery: a master and its paperwork exist
Distribution	Festivals, sales, platform release, marketing, exhibition, and the long tail	There is no end; the film simply stops earning attention

Two things about this table matter more than the rest. First, notice that pre-production ends on a calendar date, not on a state of readiness. That asymmetry is the source of most production pain. Every other phase can absorb an extra week; the shoot cannot, because the crew is booked, the location is permitted for Tuesday, and the actor flies out on Friday. Second, notice that development is the phase with no deadline, which is why most films die there. A script that is nearly ready can stay nearly ready for a decade.

The money moves through these phases unevenly, and knowing the shape saves you from a common first-film mistake. Development typically costs very little and takes the longest. Pre-production costs a moderate amount and is where the film is genuinely designed. Production is the most expensive per day by an enormous margin, often ten or twenty times the daily burn of any other phase, because you are paying a whole crew, equipment rental, insurance, catering and locations simultaneously. Post costs less per day but runs for months. Distribution, on a small film, is where you will spend money you did not budget: festival entry fees, deliverables, artwork, and travel.

Why this matters: Because production is the expensive phase, every hour of decision-making you move earlier, into pre-production, buys you back many times its cost. Preparation is not a virtue on a film set. It is arithmetic.

Who does what: departments and the chain of command

A film crew is organised as a set of departments, each with a head who answers to the director on creative questions and to the producer on money and time. Below the head sit specialists in a fairly rigid order. The structure looks bureaucratic until you have stood on a set, at which point you realise it is the only thing preventing forty people from asking the director where the extension cord is.

Department	Head	Who else is in it	Responsible for
Production	Producer, then line producer or production manager	Production coordinator, production assistants, accountant	Money, schedule, contracts, logistics, insurance
Direction	Director	First assistant director, second AD, script supervisor	Performance, staging, the meaning of every shot
Camera	Director of photography (cinematographer)	Camera operator, first AC (focus puller), second AC (slate and loading), digital imaging technician	The image: framing, lens, exposure, movement
Grip and electric	Gaffer (electric) and key grip (grip)	Best boys, electricians, grips, dolly grip	Lighting instruments and power; rigging, stands, dollies, flags
Sound	Production sound mixer	Boom operator, sound utility	Recording clean dialogue on location
Art	Production designer	Art director, set decorator, props master, construction	Everything the camera sees that is not a person
Costume, hair, makeup	Costume designer; key hair and makeup	Supervisor, standby, dressers	How the people look, and continuity of that look
Post	Editor	Assistant editor, sound designer, re-recording mixer, colourist	The film that finally exists

One role deserves more explanation than the others, because outsiders consistently misunderstand it. The first assistant director does not assist the director creatively. The first AD runs the floor. They build the shooting schedule with the producer, decide the order of shots within a day, drive the pace, call the roll, and are responsible for safety. On a well-run set the director thinks about performance and image while the AD thinks about the clock, and the two of them negotiate constantly in low voices. If you are shooting your own short with five friends, you are doing both jobs, and you will discover that the AD job silently eats the director job unless you deliberately protect it.

You will also hear the terms above the line and below the line. The line is a literal line on a traditional budget form. Above it sit the people attached before the film exists and paid as individuals: writer, director, producers, principal cast. Below it sits everything else, priced by rate and quantity: crew, equipment, locations, stock, post. The distinction matters because above-the-line costs are negotiated and below-the-line costs are calculated, which means your budget's honesty lives almost entirely below the line.

Remember: Departments exist to route questions. The right question to the right head, once, is the difference between a set that moves and a set that mills around.

The document that runs the day

Every person on a shooting crew reads one piece of paper before breakfast, and it is not the script. It is the call sheet, issued the evening before by the second AD and approved by the first AD and the producer. If you learn to read one, you can read a production.

A call sheet for a single day carries, at minimum: the date and the day number in the form Day 8 of 21; the general crew call time and the shooting call, which is the moment the camera is expected to actually roll; the location address, the parking instructions and the nearest hospital, with its address and phone number; sunrise and sunset times and the weather forecast; the scenes to be shot, in the order they are to be shot, each with its number, a one-line description, whether it is day or night, interior or exterior, the cast members in it, and its length in eighths of a page; a cast list with individual makeup, wardrobe and on-set times; the number of extras and their call; and a set of department notes, from the art department listing the props to be dressed by nine to the note that the tenant upstairs has agreed to keep her dog inside between two and four.

Two conventions on that sheet reward a closer look. Scene lengths are measured in eighths of a page, because the standard screenplay page is divided into eight vertical strips for scheduling. A scene marked 3 4/8 pages is three and a half pages, and an experienced first AD converts pages into hours from long habit. On a talky, contained drama a crew might shoot four or five pages a day; on an action sequence with stunts, a quarter of a page a day is normal. Second, the hospital line is not filler. It is there because film sets contain electricity, heavy metal on stands, vehicles, and tired people, and knowing where casualty is before you need it is the cheapest safety measure in existence.

The point: A call sheet is a schedule translated into a day. If nobody on your production is making one, nobody knows what is happening, including you.

How a shooting day actually flows

Crew call, say seven in the morning. Trucks unload, the art department dresses the set, the gaffer and key grip pre-light based on last night's conversation with the cinematographer, and the actors go into hair, makeup and wardrobe. At the shooting call the director walks the actors through the scene on the set with only the department heads watching. This is the blocking rehearsal, and it is the hinge of the whole day, because once the director and the cinematographer agree where the actors move and where the camera goes, thirty people can work in parallel for an hour.

The actors go back to makeup. Stand-ins take their marks so the crew can light the actual positions. The camera team builds the first setup. When it is ready, the first AD calls picture up and last looks, hair and makeup dart in for a final pass, and then the roll procedure runs in a fixed order that exists so that nothing is recorded before everything is recording: the AD calls for sound, the sound mixer answers that sound is speeding, the camera answers that it is rolling, the second AC steps in with the slate and calls the scene and take, and only then does the director call action. The slate matters more than it looks: the clap gives the editor a visual and audible sync point when picture and sound are recorded on separate devices, which is normal practice.

After a few takes the director asks for a different framing of the same action. That is coverage, the subject of a later lesson. Then the crew turns around, which means moving the camera and relighting for the opposite angle, and it takes real time, usually between twenty minutes and an hour. Turnarounds are why the AD groups shots by direction rather than by story order. At the end of the day the AD calls a wrap, the second AC and the digital imaging technician back up the day's media in at least two places, and the second AD issues tomorrow's call sheet. Nothing about that sequence changes if your crew is five people. Only the number of hands changes.

In short: A shooting day is a loop of block, light, rehearse, shoot, turn around. Everything in pre-production exists to make that loop faster.

What a text course can and cannot give you

Be clear-eyed about this now, because the honesty will save you time. This course can teach you vocabulary, physics, procedure, structure, and judgement about tradeoffs. It can make you legible to a crew and stop you from making expensive category errors. What it cannot do is give you the thing that only repetition gives: the feel of a room going quiet before a take, the instinct for when an actor has one more good one in them, the muscle memory of a focus pull. You get those by shooting, watching the results, and shooting again, ideally at a scale small enough that failure is cheap.

So treat every activity in this course as the actual assignment and the reading as preparation for it. And use free tools while you learn, because they are genuinely good now. DaVinci Resolve from Blackmagic Design has a free version that edits, mixes and grades to professional standard. Blender is free, open source, and includes a video sequence editor alongside its 3D and compositing tools. Kdenlive and Shotcut are free editors. Audacity handles simple audio repair. Fountain is a plain-text screenplay format you can write in any text editor and convert to industry formatting. Freesound hosts Creative Commons audio. None of these are compromises you are making because you are a student. Professionals use several of them daily.

Worth holding on to: Reading about filmmaking has a low ceiling. The ceiling on shooting a two-minute scene, cutting it, and noticing what went wrong is very high.

Common misconceptions

• Cheap cameras have made the rest of the process optional. Tangerine was shot on phones and still went through development, pre-production, a schedule, a mix, a grade, and delivery. The camera is one line in a long list.
• The first AD is the director's creative deputy. The first AD runs the floor, the schedule and safety. The relationship is a working partnership, not an apprenticeship.
• Pre-production is planning you can shorten if you are behind. Shortening prep does not save time, it moves the same decisions onto the most expensive day of the process.
• Post-production can fix it. Post can fix some things cheaply, some things expensively, and some things not at all. Sound recorded next to a running air conditioner is in the third category.
• Small crews do not need call sheets. A five-person crew needs a call sheet more than a fifty-person crew, because there is no coordinator to answer questions.

Putting it together

• Five phases: development, pre-production, production, post-production, distribution, each ending in a specific decision.
• Pre-production ends on a date rather than on readiness, and production burns money faster than any other phase.
• Crews are organised into departments with heads, so that questions route correctly and the director can think.
• The first assistant director runs the floor, the clock and safety; above the line and below the line divide negotiated from calculated costs.
• The call sheet turns a schedule into a working day, down to eighths of a page and the nearest hospital.
• A shooting day loops through blocking, lighting, rehearsal, takes and turnarounds, at any budget.
• Free tools including DaVinci Resolve, Blender, Kdenlive, Audacity, Fountain and Freesound will carry you a long way.

Sources

1. Encyclopaedia Britannica. (2024). Motion picture production. Britannica. britannica.com
2. British Film Institute. (n.d.). Education and industry resources. BFI. bfi.org.uk
3. National Film and Sound Archive of Australia. (n.d.). Collection and industry resources. NFSA. nfsa.gov.au
4. Blackmagic Design. (n.d.). DaVinci Resolve. Blackmagic Design. blackmagicdesign.com
5. Wikipedia contributors. (2024). Tangerine (2015 film). Wikipedia. en.wikipedia.org

Why the script is typed in a font from 1955

Open any produced screenplay and the first thing you notice is the typeface. Courier, twelve point, a monospaced font designed for IBM typewriters in the mid 1950s, at ten characters to the inch and about fifty-five lines to the page. Nobody chose it because it is beautiful. It survives because it is arithmetic. Monospacing means every character occupies the same width, so a page of screenplay holds a predictable quantity of speech and action, and the industry rule of thumb that one formatted page runs roughly one minute of screen time actually holds. Change the font and you break the estimate that a producer uses to price your film.

That is the whole logic of screenplay format in one example. It looks like fussy tradition and it is almost entirely functional. A screenplay is not a literary form that happens to be typed strangely. It is a technical document read by a hundred different people for a hundred different reasons: a financier reading for cost, a first assistant director reading for how many nights, a props master reading for what has to be bought, an actor reading for what their character wants, and a director reading for what the film is. Standard format exists so that all of them can extract their information from the same pages without reading each other's.

The point: Screenplay format is a measuring instrument shared across departments, not a style choice. Break it and every estimate built on it breaks with it.

The elements, and what goes in each

A screenplay is built from a small number of repeating elements, each with its own indentation. Here is a short scene written correctly, with the element names in brackets so you can see the machinery. In a real script the brackets would not appear.

INT. LAUNDROMAT - NIGHT [scene heading, also called a slugline]
A row of dryers, all still. MARA, 30s, in a wet coat, feeds coins into the last machine. Nothing happens. [action]
MARA [character]
(to the machine) [parenthetical]
I have exactly enough for this. [dialogue]
A man in a security uniform, DEV, watches from the doorway. [action]
DEV
It ate mine too. Second time this week. [dialogue]

Work through each element. The scene heading tells you three things in a fixed order: interior or exterior, where, and when. INT. or EXT., the location in capitals, then DAY or NIGHT. Those three facts drive the entire schedule, because a first assistant director groups a shoot by location and by day or night. Write a heading carelessly, say by inventing a new location name for the same room, and you have invented a new shooting unit.

Action is written in the present tense, in the third person, describing only what a camera could record and a microphone could hear. This is the rule beginners break most. You may write that Mara stares at the coin slot for a long moment. You may not write that Mara remembers her mother saying the same thing, because there is no way to photograph a memory. If the information is essential, find a physical form for it: an object, a gesture, a line of dialogue. Character names are capitalised on first appearance, along with major sound effects, which is a courtesy to the sound department.

Dialogue sits under a centred character name in capitals. A parenthetical gives a brief delivery or action note and should be rare; most of what beginners put in parentheticals is either obvious from the line or is the actor's job, and a page speckled with them tells a reader you do not trust your own writing. Transitions such as CUT TO: are largely vestigial and modern scripts use few of them. So are camera directions. A spec script that says ANGLE ON or PUSH IN is doing the director's job in public, and the standard advice is to imply the shot through what you choose to describe rather than to name it.

In short: Write only what a camera can see and a microphone can hear, in the present tense, and let the format carry the rest.

Structure as a diagnostic, not a recipe

You will meet a series of structural models, and it helps to know where they come from before you decide what to do with them. Syd Field's book Screenplay, published in 1979, popularised the three-act paradigm as a set of positions: an inciting event early, a plot point turning the story near the end of the first quarter, a midpoint, a second plot point around the three-quarter mark, and a climax. Blake Snyder's Save the Cat, published in 2005, tightened this into a fifteen-beat sheet with page numbers attached. Christopher Vogler's The Writer's Journey adapted Joseph Campbell's monomyth into a screenwriting shape. The sequence approach, taught at several film schools, divides a feature into roughly eight sequences of ten to fifteen minutes, each with its own small question and answer, which is a genuinely useful frame because it descends from the reel lengths of the silent era.

Here is the honest position on all of them. These models are descriptions of what many successful mainstream films happen to do, generalised afterwards and then sold as instructions. They are extremely useful as diagnostics and quite dangerous as recipes. Used as a diagnostic, the question is: my second act sags, and the model says something should turn at the midpoint, so what could turn here? Used as a recipe, the question becomes: it is page 55, so I must insert a false victory, which produces the specific deadness you have felt in films that seem to be hitting marks rather than telling a story.

It also helps to know how much of world cinema simply does not fit. Films built on daily routine, on repetition, on an essayistic structure, on parallel unrelated stories, or on a single unbroken situation have all worked at the highest level. If your story does not want a plot point at the end of the first quarter, it does not need one. What it does need is a reason for a viewer to keep watching at minute twenty and at minute sixty, and that is the real question every structural model is a proxy for.

Key idea: Structural models answer the question of why a viewer keeps watching. Use them to diagnose a script that is not working, not to generate one that has not been imagined.

The scene, which is where scripts are actually won

Zoom in from the film to the scene, because the scene is the unit you will spend your life writing. A scene that works usually has four properties. Someone in it wants something specific and concrete, and wants it now. Something opposes that want, whether a person, an object, a rule or the character's own reluctance. Something changes by the end, so the situation at the last line is not the situation at the first. And the scene enters as late as it can and leaves as early as it can.

That last principle deserves a demonstration. Take the laundromat scene above. A beginner's version starts with Mara walking down the street, entering, looking around, choosing a machine, and putting her coat down, which takes half a page and tells us nothing we could not infer. Starting instead on the last coin going into the slot puts us at the moment of tension immediately, and it costs nothing, because the audience will assemble the walk and the entrance from context without noticing. The same applies at the end: once Dev says the machine ate his money too, the scene has delivered its change, and any further lines are the writer being reluctant to leave.

One more diagnostic, and it is brutal. Take any scene and ask what the film loses if you delete it entirely. If the answer is only that a piece of information disappears, the scene is not a scene, it is a delivery mechanism, and the information should be relocated into a scene that is doing something else as well. Good films almost never have scenes that do one job.

Why this matters: A script is not saved in the outline. It is saved scene by scene, by making each one carry a want, an obstacle, a change, and no spare running time.

Coverage, meaning number one: the reader's report

When you send a screenplay to a production company, an agency or a competition, it is very unlikely that the executive whose name is on the envelope reads it first. It goes to a reader, who produces a document called coverage. This is a standard artefact and knowing its shape changes how you write.

Coverage typically runs two to four pages and contains: the logline, one or two sentences naming the protagonist, the situation and the central problem; a synopsis of the plot in full, spoilers included, usually a page or so; a comments section, which is the reader's actual argument about whether the script works; and a grid grading elements such as premise, structure, characterisation, dialogue and production values on a scale like excellent, good, fair and poor. At the top sits a single verdict on the script and often a separate verdict on the writer: recommend, consider, or pass. Recommend is genuinely rare. Many readers will go months without writing one, because their credibility is spent every time they do.

Two practical consequences follow. First, the reader is deciding whether to keep reading somewhere in the first ten pages, so the opening of your script is doing a job that the middle is not. Second, everything in the grid is a thing you can check yourself before submitting. Can you write your own logline in two sentences? If you cannot, the problem is usually not the logline.

Coverage, meaning number two, and why the collision matters

The same word means something completely different on set. Shooting coverage is the set of angles you shoot of a scene so that it can be assembled in the edit: a master shot of the whole action, then closer sizes, over-the-shoulder angles, inserts of hands and objects, and reaction shots. A scene that has been covered can be cut in a dozen ways. A scene that has not been covered can be cut in one way, which is the way you shot it, and if that way does not work you are finished.

The collision is not a coincidence of vocabulary so much as a coincidence of meaning: both senses of coverage are about giving somebody later the material they need to make a decision. A later lesson in this course takes shooting coverage apart in detail. For now, when someone says coverage, work out from context whether they are in an office or on a floor, because the two conversations have nothing to do with each other.

Bottom line: In development, coverage is a report that decides whether your script is read further. In production, coverage is the set of angles that decides whether your scene can be cut at all.

Writing it, cheaply

You do not need to buy software. Fountain is an open plain-text markup for screenplays: you type in any text editor, using simple conventions such as a scene heading in capitals and a character name in capitals above dialogue, and a converter renders correctly formatted pages. Trelby is a free dedicated screenwriting program. Several web-based tools have free tiers adequate for a short film. Final Draft is the studio standard and costs money, and you will not need it until somebody else's workflow requires it.

Then read scripts, which is the fastest form of learning available in this entire discipline. Every awards season, studios publish screenplays of their released films for consideration, and many remain available afterwards. Read one for a film you know well, and pay attention to the gap between what is on the page and what you remember on screen. That gap is the whole of directing, and it is where the next module starts.

Common misconceptions

• Screenplay format is snobbery. Monospaced type and fixed margins make page count a reliable estimate of screen time and of cost, which is why every department relies on it.
• A good script includes camera directions so the director knows the vision. Naming shots in a spec script reads as inexperience. Imply the shot by choosing what to describe.
• Three-act structure is a rule. It is a description of a common pattern, useful for diagnosing a stalled script and harmful when followed by page number.
• Coverage always means the reader's report. On set it means the angles shot for the edit, and the two senses belong to different rooms.
• You can write a character's thoughts in the action lines. Only what a camera can see and a microphone can hear belongs there; interiority must be given a physical form.

What to carry forward

• Standard format is a measuring instrument: Courier twelve point, roughly a page a minute, read by every department for different information.
• The elements are scene heading, action, character, dialogue, parenthetical and transition, each with fixed indentation and a specific job.
• Action is present tense and camera-recordable; parentheticals and transitions should be rare.
• Structural models from Field, Snyder, Vogler and the sequence approach are diagnostics, not recipes.
• Scenes need a want, an obstacle, a change, a late entry and an early exit, and should never do only one job.
• Development coverage is a reader's report ending in recommend, consider or pass; shooting coverage is the set of angles that lets a scene be cut.
• Fountain and Trelby are free, and reading produced screenplays is the cheapest training available.

Sources

1. Fountain. (n.d.). Fountain: A markup language for screenwriting. fountain.io
2. Writers Guild of America West. (n.d.). Resources for writers. WGAW. wga.org
3. Wikipedia contributors. (2024). Screenplay. Wikipedia. en.wikipedia.org
4. Encyclopaedia Britannica. (2024). Motion picture: The screenplay. Britannica. britannica.com

Six months of improvisation before a word is written

Mike Leigh begins a film without a screenplay. He casts actors on the basis of conversation rather than audition pieces, then works with each one alone for weeks, building a character out of people that actor has actually known, until the character has a biography, a job, a way of holding a cup. Only then do the characters meet, in improvisations Leigh sets up and interrupts, and only then does the story emerge from what those people do to each other. The dialogue of Secrets and Lies, released in 1996, was arrived at this way. By the time the camera turned, the actors had lived inside their parts for months.

Almost nobody can afford that. The reason to start here anyway is that Leigh's method makes visible a thing that hurried productions hide: performance is built long before the day, and what a director says on set is usually a small correction to work that was either done or not done earlier. If the casting is wrong and there has been no preparation, no verbal formula on the day will rescue the scene.

Key idea: Directing actors is mostly casting and preparation. On-the-day direction is the last few percent, and it cannot compensate for the first ninety.

Casting is the decision

Ask working directors what most determines a performance and a striking number will say casting, before rehearsal and far before anything they say on set. The logic is unsentimental: an actor's temperament, physical presence, rhythm and age are not adjustable, and half of what an audience reads as characterisation is simply the person. Casting well means you spend the shoot refining. Casting badly means you spend the shoot compensating, and it will show.

A conventional process runs like this. A casting director puts out a breakdown describing the roles. Agents submit clients. The casting director produces a shortlist and sends sides, which are a few pages of the script, for actors to prepare. Auditions follow, now very often as self-tapes recorded by the actor at home, then callbacks for the shortlist, and finally chemistry reads that put two candidates in a room together, because two individually excellent actors can produce nothing between them.

Three practices are worth adopting even at the smallest scale. First, audition with a scene from your actual script rather than a general piece, because you are casting this character, not talent in the abstract. Second, always give at least one adjustment: let the actor do it their way, then ask for a specific change, because how an actor takes direction predicts your entire shoot better than their first read does. Third, watch the audition back rather than trusting the room. Charisma in a small room and presence on a lens are different phenomena, and every director learns this the expensive way at least once.

You may also cast non-professionals. Robert Bresson worked almost exclusively with untrained performers he called models, drilling them through many takes until self-conscious acting fell away. Vittorio De Sica cast a factory worker, Lamberto Maggiorani, as the lead of Bicycle Thieves in 1948. Non-professionals can be extraordinary at being, and are usually poor at pretending, so write to that: give them their own idiom, few lines, and situations close to their real experience.

The upshot: Cast from your own script, always give an adjustment, and watch it back on a screen. Those three habits raise the quality of a small film more than any equipment upgrade.

Playable direction: verbs, not adjectives

Here is the single most useful technique in this lesson, and it takes one minute to learn and years to stop violating. Actors cannot play adjectives. Telling someone to be sad, to be more menacing, or to make it funnier names a result you want the audience to perceive, and gives the actor nothing to do except imitate the outward signs of that result, which reads as false. Directions that work name an action the character is taking towards another person.

Result direction (unplayable)	Action direction (playable)	What changes
Be sad in this scene	Try to get through the sentence without her noticing	Gives an objective and an obstacle; the sadness leaks out instead of being displayed
Be angrier	Make him admit he was there	Anger becomes a tactic rather than a temperature
More energy	You have four minutes before the bus goes	Replaces a volume knob with a circumstance the actor can live in
Play it colder	Do not give her the satisfaction of an answer	Names what the character withholds and from whom
Make it funnier	Behave as if this is completely normal	Comedy usually comes from commitment, not from signalling the joke

This vocabulary descends from a long tradition. Constantin Stanislavski taught actors to work from objectives and obstacles rather than from displayed emotion; Sanford Meisner built an approach around genuine attention to the other actor; Uta Hagen wrote about substituting specific personal circumstances. You do not need to adopt any single system, and it is a mistake to impose one on an actor who works differently. What you need is the habit of translating what you want to see into something the character is doing to somebody else.

A few more practical rules. Direct privately and quietly, next to the actor, not across a set. Give one note at a time, because three notes at once produce a take aimed at none of them. Avoid line readings, which is saying the line the way you want it, because they produce imitation, though be honest with yourself: with a non-actor, in a hurry, a line reading occasionally is the fastest tool available and every experienced director has used it. And when a take is good, say so plainly and specifically, because actors are working without a monitor and are guessing what landed.

Remember: Turn every adjective into a verb aimed at another character. That single translation fixes more bad direction than any other single habit.

Rehearsal, when you can get it

Rehearsal on a film is not what it is in theatre, where the goal is a repeatable performance. On a film the goal is to find things and then leave them slightly unfinished, so that they can be discovered again on camera. Sidney Lumet, who described his working methods in Making Movies in 1995, was a strong advocate of the sit-down rehearsal period: cast around a table, reading and talking, then walking the scenes on a taped-out floor plan, before a single light is rigged.

You will most likely have no rehearsal budget. Three cheap substitutes are worth the effort. A table read of the whole script with the cast tells you within an hour which scenes are long and which lines are unsayable, and it is free. A separate conversation with each principal actor about who this person is, what they want in the film as a whole, and what happened to them before page one costs an evening. And a scene-specific run in a living room, days before the shoot, lets both of you find the movement before it becomes expensive.

On the day itself you get a blocking rehearsal: the actors walk the scene on the real set, with only the department heads watching, before lighting. Guard it. It is short, it is often the only rehearsal, and it is the moment when performance and staging are decided together, which is exactly what the next lesson is about.

What actors need from you on set

A film set is a hostile environment for performance. It is bright, crowded, interrupted every ninety seconds, and shot out of order, so an actor may have to play a devastating loss on the morning of day two and the cheerful scene that precedes it on the afternoon of day nine. Your job is partly to protect the conditions in which the work can happen.

Concretely: keep the set quiet and small during emotionally exposed scenes, and be willing to ask people to step away. Tell the actor where the emotional thread sits, because you have the whole film in your head and they have this morning. Give eyelines that are real, which means having the other actor present and off camera rather than a taped cross on a stand, and be genuinely reluctant to shoot a close-up against nobody. Explain the frame size before the take, since what an actor can usefully do in a wide two-shot is not what they can do in a tight single. Shoot the actor's coverage while they are still fresh if the scene is demanding, rather than saving it for the end of a long day. And say when you have got it, because an actor who does not know whether the scene worked cannot let go of it.

Two areas are not matters of taste. Scenes involving intimacy, nudity or simulated violence require prior written agreement about exactly what will happen, and since the late 2010s intimacy coordinators have become standard practice on professional productions for exactly this reason. Nothing physical should ever be sprung on a performer to get a genuine reaction; it is an old anecdote-generating habit that ended careers and, on modern sets, ends productions. Similarly, minors on set are subject to legal limits on working hours and require a chaperone and often a tutor, and those rules are not negotiable regardless of how understanding a parent is.

So what?: Protecting performance is a set of concrete logistical acts, most of which cost nothing but attention: quiet, real eyelines, frame size, order of coverage, and explicit consent.

Common misconceptions

• Directing actors means explaining the character's psychology in detail. Long analysis before a take is usually the director thinking aloud. One playable action is worth ten paragraphs.
• A good director gets emotion by manipulating the actor. Springing surprises to force a reaction is both dangerous and unrepeatable, and repeatability is exactly what coverage requires.
• Non-professionals cannot carry a film. Bresson and De Sica built careers on them, but write to what untrained performers do well: being, in their own idiom, close to their own experience.
• Rehearsal spoils spontaneity. Film rehearsal is for finding things and leaving them slightly open, which is different from fixing a performance in place.
• You can fix a miscast lead with enough takes. Casting is largely unadjustable, and a shoot spent compensating for it will read as compensation on screen.

What to remember

• Casting is the largest single determinant of performance; audition with your own scene, always give an adjustment, and watch it back.
• Result directions name what the audience should feel; playable directions name an action aimed at another character.
• The tradition behind this runs through Stanislavski, Meisner and Hagen, but no system should be imposed on an actor who works another way.
• Rehearsal on film finds things and leaves them open; free substitutes are a table read, a one-to-one conversation, and a living-room run.
• Protect performance logistically: quiet, real eyelines, stated frame size, sensible order of coverage, and clear feedback.
• Intimacy, simulated violence and minors are governed by consent, documentation and law, not by directorial instinct.

Sources

1. Encyclopaedia Britannica. (2024). Mike Leigh. Britannica. britannica.com
2. Encyclopaedia Britannica. (2024). Sidney Lumet. Britannica. britannica.com
3. Encyclopaedia Britannica. (2024). Robert Bresson. Britannica. britannica.com
4. Wikipedia contributors. (2024). Meisner technique. Wikipedia. en.wikipedia.org

The shot that gave the editor nothing to choose from

Touch of Evil opens with a bomb. A pair of hands sets a timer, plants the device in the boot of a car, and for the next three minutes and twenty seconds the camera cranes up over a rooftop, travels down a street on the border between Mexico and the United States, picks up the car, loses it, finds a couple walking, crosses a checkpoint with them, and stops. Orson Welles shot it in 1958 as a single continuous take on a crane in Venice, California, standing in for a border town. It took a full night to get.

Now consider what that shot cost Welles in flexibility. There is no other angle. If the actor fumbles at two minutes and fifty seconds, the whole thing is thrown away and the crane goes back to its start mark. There is no way to shorten the sequence in the edit, no way to cut around a weak performance, no reaction shot to hide a technical problem. Welles accepted all of that because the unbroken take was the point. Every time you decide how to shoot a scene, you are making the same trade in miniature: how much do I fix on the day, and how much do I leave for the edit to decide?

This lesson is about answering that question deliberately rather than by accident. It has three parts, and they are one problem seen from three angles: blocking (where people and camera go), coverage (which angles you shoot), and continuity (whether the pieces will fit together).

Why this matters: Every shooting decision either buys flexibility for the edit or spends it. Beginners spend it without noticing, and then discover in the edit that the scene can only be cut one way.

Blocking: bodies first, camera second

Blocking is the choreography of a scene: who is where, who moves when, and what triggers the movement. Two working philosophies exist and both are legitimate. Some directors block the actors first, letting them find the movement in rehearsal, and then place the camera to capture what emerged. Others design the shot first, because the image is the idea, and ask the actors to work inside it. Sidney Lumet worked mostly the first way; a director shooting a composed, symmetrical style works the second. What you should not do is neither, which is what happens when everyone stands where they happened to stop walking.

Good blocking is nearly always motivated by something physical. A character crosses the room because there is a kettle, a window, a coat, or a person they want to get away from. Give a scene an object or a task and the blocking half writes itself: the argument is better if one of them is packing a bag. Watch for the two-people-facing-each-other trap, where a scene about conflict becomes two heads at the same height, three feet apart, for four minutes. Vary distance, vary height, put one of them behind the other, let someone sit down mid-sentence. Height changes in particular do enormous work: a person standing over a seated person carries a power relationship you never have to state.

You should also block for the camera's convenience without letting that show. If two actors end a scene on the same side of the room, you can cover both with a small camera move rather than a full turnaround, which may buy back forty minutes. That is not a compromise, it is craft, and knowing where the turnarounds fall is exactly what separates a director who finishes the day from one who does not.

The core of it: Motivate movement with objects and tasks, vary distance and height, and know which blocking choices are cheap to shoot and which are expensive.

The shot sizes and what each one is for

You need shared vocabulary before you can plan coverage. The names vary slightly by country and by crew, but the ladder is standard.

Size	What is in frame	What it is good for
Extreme wide / establishing	Landscape or building; figures small	Geography, scale, isolation
Wide / long shot	Whole body with space around it	Physical relationships, blocking, comedy of the whole body
Medium long (sometimes cowboy)	Roughly knees or mid-thigh up	Two or three people interacting, movement still readable
Medium	Waist up	The workhorse of dialogue; gesture and face together
Medium close-up	Chest up	Conversation with more intimacy, still some body language
Close-up	Head and shoulders	Thought, decision, the smallest visible change
Extreme close-up	Eyes, hands, an object detail	Emphasis, secrets, tension; use rarely so it keeps its force
Insert / cutaway	A prop, a note, a clock; or something outside the main action	Information, time compression, an escape hatch in the edit

Two composite terms matter for dialogue. An over-the-shoulder shot frames one actor past the shoulder and part of the head of the other, which holds both people in one image and fixes their spatial relationship. A single frames one actor alone: a clean single has no part of the other person in it, a dirty single includes a sliver of them. A pair of matching over-the-shoulders cut together is the most common shape in narrative cinema, and the reason is that it keeps two people in the same room while letting you choose whose face you are on.

The line, and the rule that keeps left on the left

Draw an imaginary line through the two people in a conversation and extend it in both directions. The 180-degree rule says: keep the camera on one side of that line for the whole scene. Do that and Mara stays on the left of frame looking right, Dev stays on the right looking left, in every shot, and the audience assembles a stable room without thinking about it. Cross the line for one shot and both actors appear to swap sides and to look the same direction, and the viewer briefly loses the geography. In an action scene, crossing the line makes two people running towards each other appear to be running the same way.

This is a convention, not a law of physics, and it can be crossed on purpose. There are four honest ways. Move the camera across the line within a shot, so the audience travels with you. Put one shot exactly on the line, a neutral head-on angle, and cross on the far side of it. Let a character move, which redraws the line legitimately. Or cut away to something else first and re-establish. Yasujiro Ozu ignored the rule for decades, staging conversations in frontal shot and reverse shot that cross the axis routinely, and his films are perfectly readable, which should tell you that the rule describes an audience habit rather than an absolute.

Its smaller cousin is the 30-degree rule: when you cut between two shots of the same subject, change the camera angle by at least thirty degrees or change the image size substantially. Cut between two nearly identical framings and the subject appears to jump slightly, because the audience reads it as the same shot with a piece missing rather than as a new view. Jean-Luc Godard broke this constantly in Breathless in 1960, and the resulting jump cuts became a style, but he broke it knowingly.

In short: Screen direction is the audience's map of the room. Keep it consistent by default, cross the line only by one of the four legitimate routes, and change angle by at least thirty degrees when you cut.

Choosing a coverage plan

Now the practical question. You have a three-page kitchen scene, two actors, and one shooting day. What do you shoot?

The traditional answer is the master plus coverage pattern: shoot the entire scene wide from beginning to end, so you have a continuous version that always works, then move in for matching over-the-shoulders, then matching singles, then any inserts. Its virtues are real. You can always fall back to the master, the actors get to play the scene whole, and the editor has a complete safety net. Its costs are also real: it is slow, it means performing the scene many times, and it can tempt an editor into a mechanical wide-then-close pattern.

A more economical approach, common on smaller films, is to break the scene into two or three chunks and cover only what each chunk needs. Perhaps the first page is a wide two-shot because the physical business matters; the middle is over-the-shoulders because the conversation tightens; and the last exchange is singles because it is the turn of the scene. This is faster, gives a stronger point of view, and requires you to be right, because you have not shot the alternatives.

Whichever you choose, three habits pay for themselves every time. Shoot overlaps: begin each take a few lines before the point you expect to use and end it a few lines after, so the editor has somewhere to cut. Get a clean version of every line from every angle you intend to use, because the one line you skipped is the line the edit needs. And shoot at least one deliberate escape hatch per scene, an insert of a hand or a cutaway to something outside the conversation, because that shot will save you from an unmatchable continuity problem you have not discovered yet.

Bottom line: Shoot fewer angles than you think and shoot them more completely. Half a dozen setups covering the whole scene beats fifteen setups that each miss two lines.

Continuity: the part nobody thinks about until the edit

The scene is shot out of order over five hours. The wide was at ten in the morning with a full glass of water and an actor's jacket buttoned. The close-up was at three in the afternoon, the glass was half empty, the jacket was open, and the actor picked the glass up with the other hand. Cut those together and the audience sees the water level jump and the jacket heal itself. On a professional set the person preventing this is the script supervisor, who watches every take against the script, notes which lines were covered from which angle, marks the director's preferred takes, records the exact physical action so that closer sizes can be matched, and keeps a lined script showing what coverage exists for each line.

Categories of continuity worth checking, in rough order of how often they go wrong: the position of props and how full or empty they are; which hand does what and on which word; costume details such as buttons, sleeves and collars; hair and makeup, especially anything wet, sweaty or bleeding; the state of the environment, including doors, curtains, weather and light direction; and eyeline consistency, which is the one that actually breaks scenes because it destroys the geography rather than just being noticed.

If you have no script supervisor, do these four things. Photograph the set and each actor before the first take of each setup, with a phone, and again after any break. Keep a written note per take of the take number, whether it was good, and any action that must be matched. Ask actors to repeat the physical action on the same word every time, and tell them explicitly that this is why. And prefer to shoot the wider size first so the closer sizes have something to match to, since it is far easier to match a close-up to an existing master than the reverse.

One consolation. Audiences forgive small continuity errors constantly, because they are watching faces and following a story, not auditing water levels. What they do not forgive is spatial confusion: not knowing who is where, or which direction someone is going. Spend your attention on geography and eyeline first, and on the water glass second.

Remember: Continuity errors of geography break a scene; continuity errors of detail merely annoy. Protect eyeline and screen direction above everything else.

Common misconceptions

• More setups mean better coverage. Coverage quality is about completeness and options, not count. Six complete angles beat fifteen partial ones.
• The 180-degree rule cannot be broken. It can be crossed by moving the camera, by a neutral on-the-line shot, by character movement, or by cutting away. Ozu ignored it for a career.
• A master shot is always required. It is a safety net with a real time cost. Many well-shot scenes have no continuous master, but you should skip it on purpose, not by accident.
• Continuity is the script supervisor's problem. On a small crew there is no script supervisor, so it is the director's problem, and photographs plus written take notes are the fix.
• You can fix crossed screen direction in the edit. Occasionally a cutaway rescues it. Usually you are choosing which of two unmatchable halves to lose.

Pulling it together

• Every shooting choice trades on-the-day certainty against flexibility in the edit; Welles paid a large flexibility price for the Touch of Evil opening on purpose.
• Block bodies with motivated movement, vary distance and height, and know which staging is cheap or expensive to cover.
• The shot-size ladder runs from extreme wide to extreme close-up, with over-the-shoulders and singles as the dialogue workhorses.
• The 180-degree rule preserves screen direction; there are four legitimate ways to cross the line, and the 30-degree rule prevents jump cuts.
• Choose between master plus coverage and chunked coverage deliberately; always shoot overlaps, complete lines, and one escape hatch.
• Continuity is protected by photographs, take notes, repeated physical action on the same word, and shooting wider sizes first.

Sources

1. Reich, J. (2017). Exploring movie construction and production: Chapter 5, What is directing? Open SUNY / Milne Publishing. milnepublishing.geneseo.edu
2. Yale University Film Studies. (n.d.). Film analysis: Cinematography. Yale. filmanalysis.yale.edu
3. Yale University Film Studies. (n.d.). Film analysis: Editing. Yale. filmanalysis.yale.edu
4. Wikipedia contributors. (2024). Touch of Evil. Wikipedia. en.wikipedia.org

One block in Bedford-Stuyvesant

For Do the Right Thing, released in 1989, Spike Lee took a single block of Stuyvesant Avenue in Bedford-Stuyvesant, Brooklyn, and turned it into the whole world of the film. The production cleaned and repainted the block, worked with residents and community organisations to clear it for the summer of the shoot, and built two of the film's central buildings, Sal's pizzeria and the Korean grocery opposite it, inside vacant lots. The production designer Wynn Thomas pushed reds, oranges and warm browns through the street so that the image itself would feel like a heatwave.

Notice what that decision bought. Because everything happens on one block, the audience learns the geography in ten minutes and then knows exactly where every character is for the rest of the film, which is why the last act lands. Because the block was designed rather than found, the colour temperature of the film could be controlled. And because the buildings were built, the production could light them, shoot inside them, and change them. Design and location are not decoration. They decide what the story is able to do.

Key idea: The art department and the location choice determine what the audience knows, what the crew can physically achieve, and what the film feels like, before a single lens is chosen.

What a production designer actually controls

The production designer is responsible for everything in the frame that is not a person: sets, locations as dressed, props, vehicles, graphics, signage, and the colour and texture of all of it. They usually work directly with the director from early prep and with the cinematographer constantly, because a wall colour is a lighting decision and a lighting decision changes a wall colour.

Under the designer sits an art department. The art director manages the practical execution and the budget. The set decorator chooses and places everything that is not built, from sofas to curtains. The props master handles anything an actor touches or carries. A construction department builds and paints. A graphic designer produces the newspapers, labels, letters and signs, which is a job that people forget exists until a scene requires a convincing 1974 electricity bill.

The designer's toolkit is worth knowing even if you never use it at scale. Mood boards and reference images establish a visual language quickly with a director who cannot yet describe what they want. A colour palette, deliberately limited, keeps a film coherent and lets a single deviation carry meaning. Floor plans and elevations, drawn to scale, let the cinematographer plan camera positions before anything is built. A set model or a rough digital model answers the question of whether the camera can physically get where the shot requires, which is the question that most often kills a beautiful design.

Design as characterisation

Here is the exercise that turns design from decoration into storytelling. Take a character and choose three objects that could belong to no one else. Not a lamp, a mug and a chair; those are furniture. A hospital lanyard still on the hook by the door, a stack of unopened post from the same sender, a single expensive knife in a drawer of cheap ones. Each one is specific enough to imply a life. An audience reads a room in about two seconds and then stops looking, so those two seconds have to carry information rather than generic set dressing.

The same discipline applies to what you deliberately leave out. Empty space says something: a room with nothing personal in it tells a story about a person who has not unpacked, or does not intend to stay. And the state of things matters more than the things: worn, repaired, half-finished, too new. A brand-new kettle in a lived-in kitchen is a small lie the audience notices without being able to say why.

Props deserve a technical note. Anything an actor handles in close-up is a hero prop, and you want more than one of it, because it will break, get lost, or need to be in two states at once. Anything that gets destroyed needs multiples, one per take, which is a scheduling constraint as much as a budget one. And every prop is a continuity object: which hand, which pocket, how full, and where it was left at the end of the take.

What matters here: Specific objects imply lives; generic objects imply nothing. Choose three things that could belong to nobody else, and get the wear right.

Built sets against real places

Question	Built set	Practical location
Cost	High up front: design, materials, construction, stage rental	Low to moderate: fee, permit, insurance, cleaning
Control of light	Total; ceilings and walls fly out, lamps rig overhead	Limited; windows, low ceilings, and the sun moving
Sound	Quiet by design, no traffic, no neighbours	Whatever the world is doing that day
Camera access	Walls remove; a dolly can go anywhere	Doorways, corridors and furniture constrain everything
Credibility	Risk of feeling like a set if detail is thin	Free authenticity: real wear, real proportions, real depth out of windows
Schedule	Available all day, every day, unchanged	Available Tuesday until four, and the owner wants it back as found

Almost every small film uses locations, and the sensible hybrid is to shoot real places and build only what a real place cannot give you. A common approach is to find a location with the right bones and then dress it heavily: change the paint, replace every soft furnishing, remove three quarters of what is there, and add your three specific objects. Removal is usually the biggest single improvement, because real rooms are far more cluttered than film rooms and clutter reads as noise on a screen.

Scouting: what to check before you fall in love

A location scout looks at photographs; a technical scout brings the department heads and answers the questions that decide whether shooting there is possible. Go through this list every time, in this order, because the first four items have killed more shoots than any aesthetic problem.

1. Sound. Stand still and listen for two full minutes with your eyes closed. Traffic, aircraft paths, a refrigerator, a hum from a fluorescent fitting, a school playground at eleven, building works next door starting Monday. Hard floors and bare walls create reverb that no editing will remove. This is the item beginners skip and regret.
2. Power. How many circuits, how many amps, where is the board, and can you run cable to it without crossing a doorway that people need? A domestic ring circuit will trip if you plug in enough lights.
3. Light and the sun. Which way do the windows face and where will the sun be at your shooting times? An east-facing kitchen looks completely different at nine and at three, and a scene shot across those hours will not match unless you control it.
4. Access and logistics. Parking for vehicles, a route to carry equipment, a lift or stairs, somewhere for cast and crew to sit that is not the set, toilets, and somewhere to feed people.
5. Ceiling height and wall access. Can lights get above the actors? Do walls or windows allow a light outside shining in? Low ceilings are the most common physical constraint in real houses.
6. Permission and neighbours. Who actually owns it, who can sign, what does the neighbour above do for a living, and is there a dog?

Then get the paperwork right. A location agreement or release, signed by someone with authority, states the dates, the hours, the fee, what you may alter, and that the production has permission to film and to exploit the resulting footage. Public filming often needs a permit from a council, city film office or police, particularly if you use a tripod on a pavement, park vehicles, or affect traffic. Your insurer will want to see the address, and the location owner will very often want a certificate of insurance naming them. That paperwork is covered properly in a later lesson; for now, understand that a location without a signed agreement is a location that can be withdrawn at nine on the morning of the shoot.

The point: Scout with your ears and a power meter, not only with your eyes. Sound and electricity decide what is shootable; the view decides only what is pretty.

Design on almost no money

Six strategies do most of the work on a small film. First, control the background rather than the whole room: the audience sees what is behind the actor, so change that and leave the rest. Second, shoot with a longer lens and a wider aperture so the background falls soft, which converts a mediocre room into texture. Third, use practical lamps, meaning lights that exist inside the scene, because a table lamp both dresses the frame and lights the actor. Fourth, remove ruthlessly, then add three deliberate things. Fifth, control colour with the cheapest surfaces you can change: bedding, curtains, cushions, a painted wall, a tablecloth. Sixth, borrow rather than buy, and photograph everything before you move it so you can put the room back exactly as it was, which is the habit that gets you invited back.

One warning about the internet's favourite advice. Coloured gels and haze can make a cheap room look styled in a still frame, and they are genuinely useful. They are also the fastest way to make a serious scene look like an advertisement. Ask what the colour means in the story before you buy the gel.

Bottom line: Background, lens, practicals, subtraction, cheap colour, and borrowing. That is nearly the whole art department on a short film.

Common misconceptions

• Production design means building sets. Most design work is selection, subtraction and dressing of real places, plus colour and texture decisions made with the cinematographer.
• A real location is automatically cheaper. Permits, insurance, parking, restoration, lost time to noise and low ceilings all cost, and sometimes exceed the cost of a simple build.
• Set dressing should fill the frame. Real rooms are cluttered and clutter reads as visual noise. Removing three quarters of what is there is the usual first move.
• Location scouting is about how a place looks. Sound, power, sun path, access and ceiling height decide whether you can shoot there at all.
• A verbal yes from the owner is enough. Without a signed location agreement stating dates, hours and permission to use the footage, your film can lose its location or its release.

The short version

• Do the Right Thing built its world on one designed block, which fixed the geography, the palette and what the film could do.
• The production designer controls everything in frame that is not a person, working through art director, set decorator, props, construction and graphics.
• Design characterises through specific objects and their state of wear, and through deliberate absence.
• Built sets buy control of light, sound, access and schedule; locations buy authenticity and low up-front cost.
• Scout for sound, power, sun path, access, ceiling height and permission, in that order, and sign a location agreement.
• On a small budget: control the background, use longer lenses, light with practicals, subtract, recolour cheap surfaces, and borrow.

Sources

1. Yale University Film Studies. (n.d.). Film analysis: Mise-en-scene. Yale. filmanalysis.yale.edu
2. Wikipedia contributors. (2024). Production designer. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (2024). Do the Right Thing. Wikipedia. en.wikipedia.org
4. British Film Institute. (n.d.). Filmmaking resources and guidance. BFI. bfi.org.uk

The lens that was built for the moon

For Barry Lyndon, released in 1975, Stanley Kubrick wanted interiors lit only by candles. No film stock of the period could expose a usable image at those light levels through a normal lens, so the production acquired Zeiss Planar 50mm lenses with a maximum aperture of f/0.7, originally made for NASA to photograph the far side of the moon, and had them adapted for a motion picture camera by Ed DiGiulio at Cinema Products. The cinematographer John Alcott then shot candlelit scenes in which depth of field was measured in centimetres and an actor leaning forward went soft.

Every term in that paragraph is a variable you will set on every shot you ever take: how much light the lens admits, how sensitive the recording medium is, and how much of the world is in focus at once. This lesson makes those variables legible, so that when a shot looks wrong you can name what to change instead of guessing.

Key idea: Exposure and focus are not settings to get right and forget. They are the controls you use to decide what the audience can see and where they look.

Exposure: three controls and a fourth you forget

The amount of light that reaches the sensor is set by three things. Aperture is the size of the hole in the lens, written as an f-number: f/1.4, f/2, f/2.8, f/4, f/5.6, f/8, f/11, f/16. That sequence is not arbitrary. Each step is one stop, and one stop halves or doubles the light. Smaller numbers mean a bigger hole and more light. Cinema lenses are often marked in T-stops instead, which are measured rather than calculated and so match across different lenses.

Shutter controls how long each frame is exposed. Stills cameras express this in seconds; motion picture work usually uses shutter angle, inherited from a rotating disc with a slice cut out of it. A 180-degree shutter exposes each frame for half the time available, which at 24 frames per second means one forty-eighth of a second, and it produces the motion blur audiences read as normal. Sensitivity, marked as ISO or gain, is how much the signal is amplified; raising it brightens the image and adds noise.

The fourth control is the one beginners forget: the amount of light in the room. If a shot is underexposed, the professional response is usually not to open the aperture, which changes the depth of field, or to raise the ISO, which adds noise. It is to add light or to move what light exists. Conversely, outdoors in daylight you will often have too much, and the answer is neutral density filtration, which is grey glass that removes light without changing colour, letting you keep the aperture and shutter you actually want.

Control	Brightness effect	The side effect that matters
Aperture (f or T stop)	One stop doubles or halves light	Changes depth of field: wide open isolates, stopped down keeps more in focus
Shutter angle	Halving the angle halves the light	Changes motion blur: narrow angle looks staccato, wide looks smeary
ISO / gain	Doubling ISO doubles apparent brightness	Adds noise; every camera has a native ISO where it performs best
Light in the scene	Direct and unlimited	Costs time, power and gear, and is almost always the right answer anyway
Neutral density filter	Removes light in stops	Lets you shoot at a wide aperture in bright sun without overexposing

Judge exposure with tools, not with your eye on a small screen in daylight. A waveform monitor plots brightness across the frame and tells you where highlights are clipping. False colour maps exposure levels to colours so you can see instantly whether skin is sitting where you want it. Zebras stripe areas above a set level. Most cameras that shoot in a log profile hold highlight detail well and recover shadows badly, so the general practice is to protect the highlights and keep skin tones a little above middle grey.

The upshot: Aperture, shutter and ISO all change brightness, but each pays for it in a different currency: focus depth, motion blur, or noise. Choose which currency you are willing to spend before you turn a dial.

Focal length: what it really changes

Here is the misunderstanding that causes more bad-looking footage than any other. People say wide lenses distort faces and long lenses flatten them. That is a description of the result, not the cause, and the difference matters enormously in practice.

Perspective is set by where the camera is, not by the lens. A face looks distorted in a wide shot because you had to get close to fill the frame, and being close makes the nose much nearer to the lens than the ears. A long lens flattens a face because you had to stand far away, so nose and ears are almost the same distance from the camera. What the focal length actually controls is how much of the world fits in the frame from wherever you are standing. Once you internalise this, framing becomes a two-step decision: choose the distance for the perspective and the psychological relationship you want, then choose the focal length that frames it.

Focal length (on a Super 35 or APS-C sensor)	Field of view	Typical use
12 to 18mm	Very wide	Small spaces, exaggerated depth, disorientation; faces distort badly up close
24 to 28mm	Wide	Environment with a person in it; handheld work; interiors
35mm	Slightly wide	The most common all-purpose lens for scenes with movement
50mm	Roughly normal	Neutral; two-shots and medium coverage
85mm	Short telephoto	The classic close-up lens: flattering, isolating, soft background
135mm and up	Long	Compression, spying from distance, stacking layers of a street

Those numbers depend on sensor size. The same 50mm lens frames much wider on a full-frame sensor than on a Micro Four Thirds sensor, because a bigger sensor captures more of the projected image. This is the crop factor, and it means that focal length numbers are only meaningful once you state the format.

Depth of field, the range of distances that appear acceptably sharp, gets deeper as you close the aperture, as you use a shorter focal length, and as you move further from the subject. It gets shallower as you open up, use a longer lens, get closer, or use a larger sensor. Shallow focus is not automatically better, despite what the internet believes. It isolates a face, but it also throws away the world, and a film in which every shot has a creamy background has thrown away the world fifteen hundred times.

Remember: Distance sets perspective; focal length sets how much you see. Choose where to stand first.

Frame rate and shutter angle

Cinema records and projects at 24 frames per second, a standard fixed in the late 1920s when sound required a constant speed. Broadcast in some regions runs at 25, and much video work uses 30 or 60. The number you shoot at determines the number the audience sees only if you play it back at the same rate. Slow motion works by recording more frames per second than you will play back: shoot at 60 and play at 24 and the action takes two and a half times as long, smoothly, because you have real frames rather than duplicated ones.

Shutter angle then controls how each of those frames looks. The 180-degree convention gives motion blur that matches audience expectation. Narrow the shutter towards 45 degrees and each frame is crisper, so fast movement stutters and shatters; the beach landing in Saving Private Ryan used narrow shutter angles for exactly this effect. Widen it and movement smears. Note that narrowing the shutter also loses light, which is why the effect is easier to achieve outdoors.

Two practical traps. First, artificial lights running on alternating current pulse at the mains frequency, and a shutter that does not divide evenly into that frequency produces flicker or rolling bands. In a 50Hz country, 25fps with a 180-degree shutter is safe; in a 60Hz country, 24 or 30fps behaves. If you see bands, change the shutter first. Second, a very high frame rate demands proportionally more light, because each frame is exposed for a shorter time.

Why this matters: Frame rate is a recording decision and playback rate is a presentation decision. Slow motion is the gap between the two, and shutter angle decides how each frame feels within it.

Moving the camera, and meaning it

The vocabulary is small. A pan rotates the camera horizontally on a fixed point; a tilt rotates it vertically. A dolly or track moves the whole camera through space. A crane or jib lifts it. Handheld puts it on a shoulder, with the operator's body in the image. A Steadicam or a gimbal isolates the camera from the operator's steps, producing smooth movement through space that a dolly cannot follow. A zoom changes focal length during the shot.

The distinction that teaches you the most is dolly against zoom. A dolly changes your position, so the relationships between foreground and background genuinely change: you move past things. A zoom only magnifies, so the whole image scales up with the perspective frozen. That is why a zoom feels like looking harder and a dolly feels like walking closer. Combine them in opposite directions, dollying in while zooming out, and the subject stays the same size while the background appears to rush, which is the effect Hitchcock used on the stairwell in Vertigo in 1958 and which is now named after it.

The practical rule is that movement should have a reason, and there are only about four good ones: to follow a subject, to reveal information at a chosen moment, to change the relationship between the viewer and the subject, or to build or release tension through pace. A move with none of those is decoration, and slow creeping pushes on static conversations are the most common beginner tic in existence. Note also that a moving camera is a scheduling decision. Laying track, rehearsing a Steadicam route, or repeating a gimbal move until it is clean costs setups you could have spent on coverage.

In short: Dolly moves you; zoom magnifies. Move the camera to follow, reveal, change relationship or pace, and count the cost in setups.

What the expensive camera actually buys

Be honest about gear. A recent phone shoots a sharp, well-corrected image in good light, and Tangerine proved the rest. What a cinema camera buys you is a set of specific advantages: more dynamic range, meaning it holds detail in a bright window and a dark corner in the same frame; a better codec with more colour information, so the image survives grading; interchangeable lenses; internal neutral density; proper audio inputs; and an ergonomic body that a focus puller and an operator can actually work with. Those are real, and none of them is resolution. A well-exposed, well-lit HD image beats a badly lit 8K image every single time, and audiences have never once left a cinema complaining about pixel count.

If you are choosing where to spend limited money on a first film, the order that most working people would give you is: sound recording first, then light, then lenses, then the camera body. That ordering is the entire argument of the next two lessons.

Common misconceptions

• Wide lenses distort faces. Proximity distorts faces. The wide lens simply lets you get close enough to fill the frame, which is where the distortion comes from.
• Shallow depth of field means cinematic. It isolates a subject and discards the world. Many of the most admired images in cinema are deep-focus compositions.
• Raise the ISO if it is dark. Adding or moving light is usually the correct answer; ISO and aperture both charge you in noise or in focus depth.
• A zoom is the same as moving the camera closer. A zoom magnifies with perspective frozen; a dolly changes your position and therefore the relationships between planes.
• Higher resolution is the main thing a better camera gives you. Dynamic range, codec quality, lens options and ergonomics matter far more than pixel count.

What you now know

• Exposure is set by aperture, shutter and ISO, plus the light in the room and neutral density filtration; each pays in a different currency.
• Waveform, false colour and zebras replace guessing; protect highlights when shooting log.
• Camera distance sets perspective, focal length sets field of view, and sensor size makes focal length numbers meaningful only with a format attached.
• Depth of field deepens with a smaller aperture, a shorter lens and greater distance.
• Slow motion is a mismatch between recording and playback rates; shutter angle decides motion blur, and mains frequency can cause flicker.
• Dolly changes position, zoom changes magnification, and camera movement should follow, reveal, change relationship or set pace.
• Cameras buy dynamic range, codec, lenses and ergonomics, not cinematic quality; spend on sound and light first.

Sources

1. Yale University Film Studies. (n.d.). Film analysis: Cinematography. Yale. filmanalysis.yale.edu
2. Reich, J. (2017). Exploring movie construction and production: Chapter 6, What is cinematography? Open SUNY / Milne Publishing. milnepublishing.geneseo.edu
3. American Society of Cinematographers. (n.d.). Craft and technique resources. ASC. theasc.com
4. Wikipedia contributors. (2024). Barry Lyndon. Wikipedia. en.wikipedia.org

The cinematographer who put a star's eyes in shadow

In the opening scene of The Godfather, released in 1972, Don Corleone sits behind a desk while a man asks him for a favour. The light comes from almost directly overhead. It falls on Marlon Brando's forehead, his cheekbones and the bridge of his nose, and it leaves his eyes in shadow. Gordon Willis, the cinematographer, kept the image well under the exposure levels Paramount was used to seeing in dailies, and the studio was nervous enough about it to complain. Willis, who later carried the nickname the Prince of Darkness, held the line, and the look became one of the most imitated in American cinema.

Take the shot apart and it teaches the whole subject. The light has a direction, and the direction hides the part of a face we read most, which makes a man unreadable. It has a quality, in this case fairly hard, so shadows have edges. It has an intensity relative to everything else in the frame, which is what makes the room feel enclosed. And it has a colour, warm, which places us indoors in a private world. Those four properties are all you can change about any light, ever.

Key idea: Lighting is not making things visible. It is deciding what is visible, how sharply, at what brightness relative to everything else, and in what colour.

The four properties, one at a time

Intensity is how much light arrives, and the thing to internalise is that it does not fall off gently. The inverse square law says that doubling the distance from a source quarters the light at the subject, which is a loss of two stops. That single fact is a shaping tool. Put a soft source close to an actor and the far side of their face falls off dramatically, giving you contrast without a second light. Pull the same source back and the light becomes even across the room, which is what you want for a wide shot with people at different distances.

Direction is where the source sits relative to the subject and the camera. Front light flattens and hides texture. Side light, a key at roughly forty-five degrees, carves out form and is the workhorse of portraiture. Top light does what Willis did to Brando. Underlight reads as unnatural because almost nothing in ordinary life lights us from below, which is why it has been horror shorthand since the silent era. Backlight separates a subject from the background and, with nothing on the front, produces silhouette.

Quality means hard or soft, and it is decided by one thing: the apparent size of the source as seen from the subject. A small source makes hard-edged shadows; a large one wraps around and makes soft, gradual ones. The sun is enormous, but at ninety-three million miles it is a small dot in our sky, so direct sunlight is hard. Put a cloud in front of it and the whole sky becomes the source, which is why overcast light is soft. You therefore soften light in exactly three ways: use a physically bigger source, put diffusion between the source and the subject to spread it, or bounce it off something large such as a wall, a ceiling or a white board. Moving a source closer also makes it apparently bigger, and therefore softer, as well as brighter.

Colour is measured in kelvin. Domestic tungsten bulbs sit around 2700 to 3200K and read as warm orange; midday daylight sits around 5600K and reads as neutral or blue; shade and overcast skies go higher still. Your camera's white balance decides which of these it calls white, and everything else in the frame shifts relative to that choice. Mixing sources of different temperatures in one frame is not automatically an error. A warm lamp inside a cool blue window is one of the most reliable images in cinema, and it happens simply by white balancing for one and letting the other go.

The point: Bigger and closer is softer. Doubling distance costs two stops. Those two sentences will solve most lighting problems you meet in your first year.

Three-point lighting, and why professionals rarely use it as taught

You will be taught the three-point setup: a key light as the main source, a fill light on the opposite side to lift the shadows, and a backlight to separate the subject from the background. It is a useful teaching device and it is the standard for interviews and portraits.

It is not how most narrative scenes are lit, and knowing why makes you better faster. Real rooms rarely contain three sources arranged in a triangle around a person. Narrative lighting normally starts from a question, which is where is the light in this room actually coming from? A window, a lamp on the table, a television, a streetlight through blinds, an open fridge. The answer becomes the key, and its position is then fixed by the story rather than by a diagram. Everything after that is subtraction and control, not addition.

The most underused instrument on a small set is a piece of black cloth. Adding fill light raises the shadow side; taking away light from the shadow side deepens it. This is negative fill, and a large black flag placed just outside the frame on the dark side of an actor will do more for a face than a second lamp. On location, where you often have far too much bouncing light coming off white walls, the art of lighting becomes mostly the art of blocking light: flags, cloth, curtains half drawn, a door closed.

How much contrast? Compare the key side to the fill side in stops. A difference of one stop, roughly a two to one ratio, is gentle and modern; two stops is dramatic; three or more and the shadow side is going black. There is no correct ratio, only a ratio that matches the scene: a comedy in a kitchen and a confession in a stairwell should not have the same one.

Worth holding on to: Ask where the light comes from in the story, make that the key, then shape with flags and bounce rather than reaching for another lamp.

A kit that costs almost nothing

Item	Rough substitute	What it does
Bounce board	White foam board from a stationery shop	Turns one hard source into a large soft one; fills shadows without power
Negative fill	Black fabric, a duvet cover, cardboard painted matte black	Deepens the shadow side and adds shape
Diffusion	Baking parchment at a safe distance, a white shower curtain, a bedsheet	Enlarges a source and softens shadow edges
Practical lamps	Household lamps with brighter bulbs, on dimmers	Motivated light that also dresses the frame
Paper lantern	A cheap paper shade with a suitable bulb	A large soft omnidirectional source that can hang just out of frame
One adjustable LED panel	Entry-level bicolour panel	A controllable key or an accent when nothing in the room will serve
Clamps, stands, sandbags, gaffer tape	None; buy these	Rigging and safety; an unweighted stand falling on an actor is the classic set injury

Note what is not on that list: a big lighting package. A window, a bounce board and a black cloth will light a two-person scene beautifully, and this is a genuinely standard professional approach rather than a compromise for beginners.

Three situations, three recipes

Daylight interior. Turn every domestic overhead light off; ceiling fixtures are the enemy of a good image because they light from directly above and flatten everything. Place your actor so the window is to one side of them, not behind the camera. Bounce a board into the shadow side from just out of frame, and if the contrast is still too low, replace the bounce with black. If the sun swings round and the window blows out, diffuse the window itself with a sheet or net curtain. Watch the clock: this setup will not match in four hours.

Day exterior. Direct midday sun is a small hard source almost overhead, which is the least flattering light available and the reason everybody squints. Three fixes, in order of cost. Turn the actors so the sun is behind them, which gives a free rim light, and bounce a board into their faces from the front. Move into open shade, such as the shadow side of a building, where the source becomes the whole sky. Or shoot in the hour after sunrise or before sunset, when the sun is low, warm and soft enough to use directly. Bring neutral density filters, because you will otherwise be forced to shoot at f/16.

Night exterior. This is the hardest thing on this list and the place where small productions most often produce unusable footage. The failure mode is trying to light the whole scene. Do the opposite: find or place a few motivated sources, a streetlight, a shop window, a car headlight, a phone screen, and stage the action around them, letting the rest go black. Expose so that faces sit a stop or two below middle grey rather than crushed into noise, and give the frame something bright in it so the darkness reads as a choice. If you can, shoot a night interior instead.

Remember: On location the good light is usually already there. Your job is to find it, stand the actor in it, and take away what ruins it.

Safety, briefly but seriously

Lighting is the department that puts hot metal on tall poles above people. Sandbag every stand. Run cables along walls and tape them down, never across a doorway. Do not exceed a circuit: a domestic ring circuit trips when you overload it, and finding out mid-take costs an hour. Tungsten instruments get hot enough to burn skin and to ignite paper, so keep diffusion at a distance and never leave a lamp unattended against fabric. Wear gloves for hot fixtures. If you are wiring anything unusual, do not; hire someone qualified. This is the first appearance of a theme that gets a full lesson later.

Common misconceptions

• More lights mean better lighting. One motivated source plus negative fill outperforms four instruments arranged without a reason.
• Soft light comes from a special kind of lamp. Softness is apparent source size. Any hard source becomes soft if you bounce it, diffuse it, or make it bigger and closer.
• Three-point lighting is how films are lit. It is an interview and portrait template. Narrative lighting starts from where the light in the room would actually be.
• Underexposing makes a scene look moody. Noise makes a scene look cheap. Moody scenes are usually correctly exposed on a small bright area with everything else deliberately dark.
• You must turn the room lights on to shoot indoors. Domestic ceiling fixtures light from directly overhead and flatten faces. Turning them off is often the first improvement.

Looking back

• Willis lit Brando from above and left his eyes dark, which is a demonstration that lighting decides what an audience can read.
• The four controllable properties are intensity, direction, quality and colour, and nothing else about a source can be changed.
• The inverse square law makes distance a shaping tool: close sources fall off fast, distant ones spread evenly.
• Hard or soft is decided by apparent source size, so bounce, diffuse, enlarge or move closer to soften.
• Motivated key plus subtraction beats the three-point diagram for narrative work, and negative fill is the cheapest shaping tool there is.
• Daylight interior: window to the side, bounce or black. Day exterior: backlight and bounce, open shade, or low sun. Night exterior: motivate a few sources and let the rest go.
• Sandbag stands, tape cables, respect circuits and heat.

Sources

1. Yale University Film Studies. (n.d.). Film analysis: Cinematography. Yale. filmanalysis.yale.edu
2. American Society of Cinematographers. (n.d.). Craft and technique resources. ASC. theasc.com
3. Wikipedia contributors. (2024). Gordon Willis. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (2024). Three-point lighting. Wikipedia. en.wikipedia.org

Two arms above the head for nine hours

A boom operator spends a shooting day holding a carbon fibre pole above their head with both arms, positioning a shotgun microphone roughly a foot above the actor's hairline, angled down at the mouth, and about an inch outside the top edge of the frame. They hold it there through the take without drifting, and they follow the dialogue by rotating the pole so the microphone points at whoever is speaking. It is genuinely exhausting, and every part of the position is calculated.

Why so close? Because of the same inverse square law that governs light. Halve the distance between a microphone and a mouth and the dialogue arrives about six decibels louder, while the traffic outside, the fridge in the corner and the room's reverberation stay exactly as loud as they were. Distance is the whole game. A four hundred pound microphone one metre away will produce worse dialogue than a hundred pound microphone thirty centimetres away, every single time.

Now the uncomfortable part. Audiences tolerate a soft, grainy, imperfectly exposed image without complaint, and many admired films are visually rough. They do not tolerate dialogue they have to strain to hear. Bad sound reads instantly as amateur and, unlike a bad image, it cannot be graded into acceptability afterwards. This is why the standard advice for a first film is to spend on sound before spending on camera.

Key idea: Dialogue quality is set by the distance from the mouth to the microphone. Everything else in this lesson is in service of getting that microphone closer.

Microphones and what each one is for

A microphone's polar pattern describes which directions it listens to. An omnidirectional capsule hears everything equally. A cardioid hears mostly in front. A hypercardioid or supercardioid, which is what a shotgun tube produces, narrows that further, at the cost of a small rear lobe that picks up whatever is directly behind it, which is why you point a shotgun down at an actor from above rather than up at them from below, where the rear lobe collects the ceiling and the air conditioning.

Microphone	Strength	Weakness	Best used
Shotgun on a boom	Directional, natural, close, full-bodied dialogue	Needs an operator; its narrow pattern misbehaves in echoey rooms	Exteriors and most dialogue coverage
Hypercardioid on a boom	Smoother in reverberant interiors than a long shotgun	Less reach; picks up more room	Small rooms with hard surfaces
Lavalier (radio mic)	Very close to the mouth regardless of framing; survives wide shots	Clothing rustle, radio interference, thin sound, hiding it takes time	Wide shots, movement, crowds, documentary
Plant microphone	Hidden in the set, near where an actor will be	Fixed; only works if the actor hits the mark	Static scenes at a table or a car dashboard
Camera-mounted microphone	Nothing to carry	It is exactly as far from the actor as the camera is	Reference and scratch audio only

The professional habit worth copying is to record a boom and a lavalier at the same time on separate tracks. The boom sounds better; the lavalier saves you when the boom cannot get in, when a plane goes over, or when a line is delivered facing away. Two tracks give the editor a choice, and choice is the same currency coverage buys in the picture.

Levels, formats and sync

Set your levels so that the loudest moment peaks around twelve to six decibels below full scale. That headroom exists for the shout you did not expect. Digital audio does not distort gracefully like tape: once a signal clips, the peaks are gone and no plugin recovers them. Record at a sample rate of 48 kHz and a bit depth of 24, which is the standard for film and gives you room to raise a quiet recording later without dragging up noise.

Most productions record sound on a separate device from the camera, which is called dual system, because a dedicated recorder has better preamplifiers and proper inputs. That means picture and sound must be lined up afterwards. There are two ways. Professional sets feed matching timecode to both devices so the editing software aligns them automatically. Everyone else uses a clapperboard: the visible clap and the audible clap mark the same instant, and software can also sync by comparing the recorder's audio with the camera's scratch track. Whichever you use, always record a scratch track on the camera as well, because it is what lets automatic sync work and what saves you when a file goes missing.

And monitor with closed-back headphones on your head for every take. Meters tell you about level; only headphones tell you that a motorbike went past, that the actor's shirt is rustling, or that the boom is picking up a hum you will never remove. The mixer who is not wearing headphones is not doing the job.

Bottom line: Peak around minus twelve, record 24-bit at 48 kHz, always run a camera scratch track, and never take the headphones off.

Controlling a location

Half of production sound is a routine you perform before the first take. Walk the room and switch off the refrigerator, the freezer, the air conditioning, the boiler, the extractor fan and any computer with a fan. Tie a note to the plug of anything you unplug, because a warm freezer full of somebody's food is how a production loses a location permanently. Stop ticking clocks. Close windows, and if that is impossible, work out the traffic light cycle down the road and shoot in the gaps.

Then deal with reverberation, which is the problem beginners cannot name and cannot fix later. Hard floors, bare walls and glass make speech bounce, and a reverberant recording can never be made dry again. Soften the room: lay a rug outside the frame, hang a moving blanket or a duvet behind the camera, open a wardrobe full of clothes, put a coat over the back of a chair. Get the microphone closer, since closeness raises the direct sound relative to the reflected sound. Choose your locations partly on this basis, which is why the previous lesson put sound at the top of the scouting list.

Watch wardrobe and props too. Nylon and stiff waterproof fabrics are loud. Bangles and necklaces rattle. Leather shoes on a wooden floor overwhelm dialogue, and the fix is to tape foam under the soles or to lay a rug where the camera cannot see it. Paper is astonishingly loud near a microphone. Costume and sound should have this conversation in prep rather than at the first take.

Finally, record room tone: thirty to sixty seconds of the room doing nothing, with everybody present and silent, at every location and after any significant change. It sounds like nothing and it is one of the most useful files you will ever record, because an editor uses it to fill gaps, to smooth between takes and to hide edits. A scene without room tone has audible holes where the background disappears, and that alone tells a viewer that something is wrong even if they cannot say what.

So what?: Silence the room, deaden its reflections, dress the actors for quiet, and record room tone every time. None of these costs money and all of them cost less than the alternative.

The expensive alternative: ADR

If dialogue is unusable you can replace it. Automated dialogue replacement brings the actor into a studio to watch the scene and re-perform their lines in sync. It is a real craft, used routinely on large films for both technical and creative reasons, and it is the reason many crowded action scenes sound so clean.

For a small production, though, treat it as a last resort, and know what it costs before you rely on it. You need the actor's time again, which may mean months later and after a haircut. You need a quiet room and a microphone. You need the performance to match the picture frame by frame. And you need the replaced line to sit convincingly in the same acoustic space as the lines around it, which requires reverb matching and is the step that most often gives it away. Every hour of care spent on set is worth several hours of ADR you will not have to schedule.

The highest-leverage two hundred pounds on a first film

A workable location sound kit is genuinely cheap: a used shotgun or hypercardioid microphone, a suspension mount to isolate it from handling noise, a windshield if you go outdoors, a boom pole (a painter's extension pole works while you save up), an XLR cable, a handheld two-channel recorder, and a pair of closed-back headphones. Add a second person to hold the boom, and that person is more valuable to your film than any lens.

For post, Audacity is free and handles simple repair and level work, and DaVinci Resolve's free version contains a full audio workstation. Freesound provides Creative Commons effects. But understand the ordering: those tools rescue very little. What they mostly do is polish something that was recorded properly in the first place.

Common misconceptions

• An expensive microphone fixes bad sound. Distance dominates. A modest microphone close to the mouth beats a costly one across the room.
• The camera's microphone will do for dialogue. It is exactly as far away as the camera, so it records the room rather than the actor.
• Noise can be removed in post. Broadband noise reduction trades noise for artefacts, and reverberation cannot be undone at all.
• Room tone is optional. Without it, an editor cannot fill gaps or smooth between takes, and the background audibly disappears at every join.
• ADR is a routine fallback for a small film. It costs actor availability, a studio, precise sync and convincing reverb matching, and it usually sounds worse than what you failed to record.

What to carry forward

• Halving microphone distance gains about six decibels of dialogue while the background stays the same; closeness is the whole strategy.
• Audiences forgive a rough image and do not forgive dialogue they must strain to hear.
• Boom for quality, lavalier for coverage, both at once on separate tracks; camera microphones are for reference only.
• Peak around minus twelve decibels, record 24-bit at 48 kHz, use a clapperboard or timecode, and keep a camera scratch track.
• Silence appliances, tame reverberation with soft materials, quiet the wardrobe, and record room tone at every location.
• ADR is expensive and rarely as good; on-set care is much cheaper than replacement.
• A used microphone, a pole, a recorder, headphones and a second person is the highest-value spend available on a first film.

Sources

1. Reich, J. (2017). Exploring movie construction and production: Chapter 8, What is sound? Open SUNY / Milne Publishing. milnepublishing.geneseo.edu
2. Yale University Film Studies. (n.d.). Film analysis: Sound. Yale. filmanalysis.yale.edu
3. Wikipedia contributors. (2024). Production sound mixer. Wikipedia. en.wikipedia.org
4. Audacity Team. (n.d.). Audacity: Free audio editor and recorder. audacityteam.org

Fifty-one percent

Walter Murch, who edited Apocalypse Now and The Conversation, set out in his book In the Blink of an Eye a list of six things a cut can honour, ranked by importance, with percentages attached. Emotion, meaning whether the cut serves what the audience should be feeling, he weighted at 51 percent. Story, whether it advances the narrative, 23. Rhythm, whether the cut falls at the right instant, 10. Eye-trace, whether it respects where the viewer is already looking, 7. The two-dimensional plane of the screen, meaning the 180-degree convention and screen direction, 5. And the three-dimensional continuity of the physical space, whether people and objects genuinely match across the cut, 4.

Read that list twice, because the ordering is deliberately provocative and it is the most useful thing in this lesson. Physical continuity, the thing beginners agonise over, is last and worth four percent. Emotion is worth more than the other five combined. Murch's point is not that continuity is unimportant; it is that when they conflict, and they conflict constantly, you sacrifice the lower one. A cut that matches perfectly and kills the feeling of the scene is a bad cut.

Criterion	Weight Murch assigns	The question it asks
Emotion	51	Does the cut serve what the audience should be feeling here?
Story	23	Does it advance the narrative?
Rhythm	10	Does it fall at the right instant?
Eye-trace	7	Does it respect where the viewer is already looking?
Two-dimensional plane	5	Does screen direction hold?
Three-dimensional space	4	Do the physical details match?

Key idea: When criteria conflict at a cut point, serve the higher one. Emotion outranks continuity by an order of magnitude.

The workflow, in the order it actually happens

Editing begins with clerical work that determines whether the creative work is possible. Ingest means copying media off cards to at least two drives, verifying the copies, and never formatting a card until both copies are verified. Organisation means a folder and naming convention you can still understand in six months, and bins in the project arranged by scene. Sync pairs the separately recorded audio with the picture. If your computer struggles with the native files, generate proxies, which are small low-quality copies you edit with and swap back for the originals at the end.

Then you watch everything, all of it, and log it. Mark the takes and the pieces of takes you might use, and write short notes on why. The instinct to start cutting immediately is strong and costs you the film, because you end up building from the three takes you happen to remember rather than the ones that are best.

The creative sequence usually runs: a string-out or selects reel, which is the good material for a scene laid end to end in order; an assembly, the whole film in sequence, always too long, often by half; a rough cut, in which structure is fought over and scenes get removed; a fine cut, in which the film is tuned frame by frame; and picture lock, the moment the edit stops changing so that sound, music and colour can be finished against a stable timeline. Picture lock matters administratively, because every frame changed after it costs the sound and colour departments real work.

The grammar of a cut

Continuity editing is a set of conventions for joining shots so that the audience perceives continuous action rather than a series of separate images. Four are worth knowing precisely.

Match on action. Cut in the middle of a movement rather than before or after it: a hand reaching for a door in the wide, then continuing to reach in the closer shot. The viewer's attention is on the movement, and the movement carries them across the join. This is the single most invisible cut available and it is why you shoot overlaps.

Eyeline match. A shot of someone looking, followed by a shot of what they see. The audience builds the spatial relationship for free, and it works even if the two shots were filmed in different countries a year apart, which is the Kuleshov principle in practical clothing.

Shot and reverse shot. The standard dialogue pattern, alternating between complementary angles across the line, which works only if screen direction has been kept.

The split edit, also called a J-cut or L-cut, where sound and picture do not change at the same moment. Bring the next scene's sound in before its picture and the transition feels smooth and forward-leaning. In dialogue, cut to the listener before the speaker finishes, and the scene stops feeling like a tennis match. If you learn one editing technique this year, learn this one; it is the difference between a scene that plays and a scene that alternates.

Why this matters: Most amateur dialogue editing cuts exactly at the end of each line, on both picture and sound. Splitting the audio and the picture apart is the cheapest possible upgrade to the feel of a scene.

Rhythm, which is not the same as pace

Beginners hear pace and think shorter shots. That is not it. Rhythm is where the cut falls relative to the internal beats of the performance and the scene, and a slow film with excellent rhythm feels faster than a fast film with bad rhythm.

Three practical handles. First, cut on the breath. Actors breathe before they speak, and cutting to a listener during that intake feels natural, while cutting on the word itself feels mechanical. Second, hold after the line. The most valuable frames in most dialogue scenes are the second and a half after someone finishes speaking, when they have to live with what they have said, and beginners systematically cut them off. Third, vary the interval. A scene in which every shot lasts about four seconds develops a metronome quality the audience feels as monotony even if they cannot identify it.

Then there is the diagnostic that fixes more scenes than any other: watch the cut with the sound off, and then listen to it with the picture off. The picture-only pass shows you where the images are repetitive or the geography is unclear. The sound-only pass shows you where the performance dies, because dialogue rhythm survives without images and reveals every hesitation you left in. Almost every over-long scene is diagnosed correctly in one of those two passes.

The upshot: Cut on breaths, hold after lines, vary the interval, and diagnose with picture-only and sound-only passes.

Structure: the film you did not shoot

The largest gains in editing come from removing things rather than from refining them. Certain patterns recur so consistently that they are worth checking on every project.

The first scene often goes, because films are commonly written with a warm-up the audience does not need. Whole scenes whose only job was to deliver information often go, because the information turns out to be inferable, and the audience is quicker than the writer assumed. Explanations placed before an event often work better after it, or not at all. And a scene that is not working sometimes needs its ending cut off, not its middle tightened, because the point is made two lines before the writer stops.

To do this you need distance, and distance is a technical problem. Leave the cut alone for a week. Show it to three people who have not read the script, and do not explain anything before or during the screening; the explanation you feel compelled to give is precisely the information the film is failing to convey. Watch their bodies rather than their faces, because people fidget honestly and comment politely. Ask what they were wondering at minute twelve rather than whether they liked it.

One trap deserves a warning: temp love. Editors cut with temporary music, usually taken from other films, and everyone falls for it. By the time your composer delivers something original, the team has heard the temp track two hundred times and the new score sounds wrong purely because it is unfamiliar. Cut with less temp music than feels comfortable, and try any scene you believe needs music with none at all before you accept that judgement.

Remember: Editing is mostly subtraction. The first scene, the explanatory scene, and the last two lines of a scene are where the film is usually hiding.

Doing it for nothing

The free version of DaVinci Resolve is a full professional editing, sound and colour application, and its limitations are irrelevant to a short film. Kdenlive and Shotcut are free, lighter, and run happily on modest machines. Blender includes a video sequence editor if you are already in that ecosystem. What actually constrains you is storage and patience: video files are large, you need at least two copies of everything, and an underpowered machine turns a creative process into a waiting process, which is what proxies exist to solve.

Common misconceptions

• Good editing means invisible continuity. Murch ranks physical continuity last. A perfectly matched cut that kills the feeling of a scene is the wrong cut.
• Faster cutting means better pace. Rhythm is where cuts fall relative to performance beats. Short shots cut carelessly feel slower than long shots cut well.
• You should start cutting as soon as the footage arrives. Watching and logging everything first is what stops you building the film from the takes you happen to remember.
• Temp music helps you evaluate a scene. It mostly makes weak scenes feel finished and sets up a fight with the composer later.
• A scene that is not working needs tightening. More often it needs its last two lines removed, or the whole scene removed.

The takeaway

• Murch ranks a cut's obligations: emotion 51, story 23, rhythm 10, eye-trace 7, screen plane 5, physical continuity 4.
• Ingest to two verified drives, organise, sync, proxy if needed, then watch and log everything before cutting.
• The sequence runs string-out, assembly, rough cut, fine cut, picture lock, and lock exists so sound and colour can finish.
• Match on action, eyeline match, shot and reverse shot, and split edits are the working grammar; split edits transform dialogue.
• Cut on the breath, hold after the line, vary the interval, and diagnose with picture-only and sound-only passes.
• Structural gains come from cutting scenes, especially first scenes and information-only scenes; get distance and watch strangers watch it.
• Resolve, Kdenlive, Shotcut and Blender are free; storage and two backups are the real constraint.

Sources

1. Reich, J. (2017). Exploring movie construction and production: Chapter 7, What is editing? Open SUNY / Milne Publishing. milnepublishing.geneseo.edu
2. Yale University Film Studies. (n.d.). Film analysis: Editing. Yale. filmanalysis.yale.edu
3. Wikipedia contributors. (2024). Walter Murch. Wikipedia. en.wikipedia.org
4. Blackmagic Design. (n.d.). DaVinci Resolve. Blackmagic Design. blackmagicdesign.com

A projector motor and the back of a television

Ben Burtt built the lightsaber out of two accidents. The first was the idling motor of an old film projector at the University of Southern California, which put out a steady hum with a slight wobble in it. The second was electrical interference: walking a microphone with a poorly shielded cable past the back of a television picture tube made the cable pick a buzz straight out of the air. Burtt layered the two, then played the result through a speaker while swinging a microphone past it, so the Doppler shift gave the hum its swing. That was 1977, and it is still the noise people make with a cardboard tube.

The same film gave him the TIE fighter, which is an elephant call layered with a car passing on wet tarmac, and R2-D2, which is Burtt's own voice interleaved with an ARP 2600 synthesizer. The trivia is fun. The principle underneath it is the reason this lesson exists: essentially nothing you hear in a finished film was recorded at the moment you watched it happen.

Take an ordinary shot. A woman walks across a kitchen, sets a mug on a counter, and says one line while rain falls outside. In the finished film you hear four footsteps, a coat rustling, ceramic on stone, the line, rain, a fridge, and a distant road. On the day, the microphone recorded the line and a room full of compromise. The footsteps were performed later by someone in different shoes on a slab of tile. The mug was a different mug. The rain is from a library. The fridge was switched off for the take and its hum added back in the mix. Nothing about that is deception; it is the only way to get a soundtrack in which each element can be controlled.

The core of it: A soundtrack is built, not captured. Production sound gives you dialogue and a reference. Everything else arrives later, one layer at a time.

The layers, and why they are kept apart

Post-production sound is organised into layers, and each layer lives on its own group of tracks that can be balanced independently. These groups are called stems.

Layer	What it contains	Where it comes from
Dialogue	Every word spoken on camera	The production recording, cleaned and edited
ADR	Lines re-performed to picture	An actor in a quiet room, later
Group and walla	Crowd murmur, background chatter, a pub	A group of voices recorded specially, or a library
Foley	Footsteps, cloth movement, props handled by characters	Performed in sync to picture on a Foley stage
Hard effects	Doors, cars, gunshots, phones, weather events	Libraries, field recording, or design from scratch
Backgrounds	The continuous sound of each place: room, street, forest	Room tone from the location, plus library ambience
Music	Score and any music heard within the story	A composer, a licensed track, or a source recording

Separating them is not tidiness. It is a delivery requirement with a specific commercial reason. To sell a film into a territory that dubs rather than subtitles, you must supply a music and effects track, universally called the M and E: the entire soundtrack with the dialogue removed and nothing missing. If your character's footsteps only exist because they happened to be audible on the dialogue microphone, they vanish the moment the dialogue stem is muted, and the dubbed version has a woman crossing a kitchen in total silence. That single requirement is why Foley is performed even for sounds that were perfectly audible on set.

It also buys you creative control. Every element on its own fader means the rain can come down four decibels under the line and come back up after it, which is a decision, rather than being welded to the dialogue at whatever level the day gave you.

Foley, and the man it is named after

Jack Foley worked at Universal during the changeover to sound in the late 1920s, when a film's effects had to be performed live while the picture ran. He and his team walked, opened doors, poured drinks and shuffled paper in front of a microphone in time with the projection. The craft kept his name. A modern Foley stage has pits set into the floor filled with gravel, sand, concrete, wood and leaf litter, a wall of shoes in every weight and sole, and a heap of props that includes celery for breaking bones and a leather jacket for wing beats.

Why perform footsteps when a library has thousands? Because a library footstep is a generic person on a generic floor. A Foley artist watches this actor, in this shot, carrying this much tension, and matches their weight, their hesitation, the drag of the left foot. Footsteps are characterisation, which is why the sound of someone approaching a door tells you their mood before they open it. Cloth movement does the same job more quietly. It is also the element that most reliably makes an amateur soundtrack feel finished, because its absence reads as a person floating half a centimetre above the floor.

On a short film with no stage, you can do a great deal with a quiet room, a duvet over a table for a dead acoustic, a microphone on a stand at knee height, and a laptop playing the scene on mute while you perform to it. Record each pass separately. Do the shoes, then the cloth, then the props. Trying to perform all three at once is how you end up with one unusable take.

So what? Foley is the cheapest professional upgrade available to a no-budget film, because it needs a quiet room and patience rather than money.

Dialogue first, always

The order of work in post sound is fixed, and it is fixed because each stage depends on decisions made in the last. Dialogue edit, then backgrounds, then hard effects, then Foley, then music, then the mix. If you build a beautiful ambience bed before the dialogue is cut, you will rebuild it, because the dialogue edit changes which takes are used and therefore what the background has to smooth over.

A dialogue edit means the following, in order. Split each character onto their own track, so a single equalisation setting can serve one voice throughout rather than fighting a compromise across all of them. Remove clicks, lip smacks, chair creaks and off-mic noise in the gaps. Fill every gap you have created with room tone from that location, because a hole of pure digital silence is more noticeable than the noise it replaced. Cross-fade every join, even by a few frames. Then, and only then, decide what needs ADR.

Noise reduction deserves a warning. Software that learns a noise profile and subtracts it is genuinely useful in small doses and instantly recognisable in large ones: voices acquire a hollow, underwater, swirling quality that is worse than the hiss you removed. Take out three or four decibels, not fifteen. Audacity includes a workable noise reduction; the professional tool most editors reach for is iZotope RX, which is not free. Reverberation, as the production sound lesson said, is not removable at all in any meaningful sense.

Building a place the audience believes in

Every location needs one continuous background that never stops for the length of the scene. It is the single most overlooked element in student films, and its absence is why cutting between two shots of the same room can feel like cutting between two different films. Lay the ambience across the whole scene, under everything, at a level you barely notice, and the scene knits together.

Perspective is the next step. A sound made in the room and the same sound made in the next room are not different volumes; they are different sounds. The distant one loses its high frequencies, gains reverberation, and arrives slightly later. If you take a clean effect and simply turn it down to put it in the hallway, it sits in the mix like a small close sound rather than a distant one. Roll off the top end and add a little reverb, and the audience places it down the corridor without being told.

Off-screen sound is also where a low budget quietly buys production value it cannot afford to shoot. You cannot stage a street festival. You can absolutely have your two characters argue in a kitchen while a festival goes past outside the window, and the audience will build the festival for you. Sound is the only department that can create things which were never in front of the camera, and it is why the sound of a war is more affordable than the sight of one.

The mix, in numbers

Mixing is the stage where every element is balanced, placed and processed into one coherent soundtrack. Four technical facts govern it.

Headroom. Keep peaks around minus six to minus three decibels full scale on the master, never touching zero. Digital clipping is permanent damage, not a warm distortion.

Loudness is not peak level. Perceived loudness is measured in LUFS, loudness units relative to full scale, averaged over time. Broadcast standards fix it: the European Broadcasting Union's R 128 recommendation targets minus twenty-three LUFS, and the American ATSC A/85 standard targets minus twenty-four. Streaming platforms normalise uploads towards their own house level, commonly reported near minus fourteen LUFS, which means a track you crushed to be loud simply gets turned down, and it arrives quieter and flatter than a track that was mixed with dynamics intact. Mastering loud is a self-defeating strategy on every platform a first film will actually reach.

Dialogue rules the mix. A common working method is to set the dialogue at a comfortable level and never move it, then build everything else underneath. If music and dialogue fight, the music loses. If an effect obscures a line, the effect ducks. Every experienced mixer has a story about a director who wanted the music louder and an audience who could not hear the plot.

Check on what people use. Mix on decent headphones, then listen on laptop speakers, then on a phone speaker held at arm's length, then in a car if you have one. A mix that survives a phone speaker is a mix that survives anything, because the phone has no bass at all and reveals every place where you relied on low frequencies to carry weight. Check in mono as well: if a sound disappears when the channels are summed, you have a phase problem that will bite you in a cinema.

Bottom line: Protect headroom, respect loudness standards rather than fighting them, keep dialogue intelligible, and audition the mix on the worst speakers your audience will use.

Silence, and why loud things need it

Loudness in a film is entirely relative. An explosion at maximum level after two minutes of a similarly loud action sequence is a dull thud, and the same explosion after four seconds of near silence is enormous. Because the range from quietest to loudest is fixed by the medium, the only way to make something feel big is to make what precedes it small. This is why the most striking use of sound in many films is a removal: the moment the score drops out, or the world goes muffled inside a character's head.

That last technique is worth naming. Point of audition is the sound equivalent of point of view. When you filter and dull everything except a character's breathing, you have put the audience inside their body, and you have done it without a line of dialogue explaining that they are overwhelmed. It costs nothing but an equaliser and a decision.

Music: what it can do, and what it gets used to hide

Music in a film is either diegetic, meaning it exists in the world of the story and the characters could hear it, or non-diegetic, meaning it is scoring the film from outside. Both are useful, and the moment a film moves between them deliberately, a car radio swelling into a full orchestra, is one of the reliable pleasures of the form.

The process is a spotting session: the director and composer watch the cut and agree exactly where music starts and stops, to the second. What most beginners get wrong is quantity. A short film with wall-to-wall music has no dynamic range left for the moment that matters, and, worse, continuous music tells the audience that the film does not trust its scenes.

The honest diagnostic is uncomfortable and worth running: play any scene you believe needs music without it. If the scene works, the music is a choice and you can make it well. If the scene collapses, the music was carrying it, and no score will make a scene that does not work into one that does. It will only make the failure louder and more expensive.

Clearing music without a lawyer

This is where student films get taken off the internet. A recorded piece of music carries two separate rights, and you need permission for both.

The composition is the written work, the notes and lyrics, controlled by the songwriter and publisher. Permission to sync it to picture is a synchronisation licence. The recording is the specific performance captured on a specific master, controlled by the label. Permission to use that recording is a master use licence. Buying the song on a streaming service or owning the CD grants you neither of these things.

The most common misunderstanding follows directly. A composition can be old enough to be out of copyright while the recording of it is new and fully protected. A 1901 piano piece is free to arrange and perform yourself; a 1998 concert recording of that same piece is not free to use. Public domain applies to the work, not to somebody's performance of it.

Three legitimate routes exist for a film with no money. Commission someone: a composition student will often score a short for a credit and a copy of the film, and you should still sign a one-page agreement granting the production the right to use the music in all media, worldwide, in perpetuity, because a festival or platform will eventually ask you to warrant exactly that. Use openly licensed music: Creative Commons licences are real permissions, but read the conditions, since BY requires you to credit the creator in a specified way and NC forbids commercial use, and a festival that charges admission or any distribution deal will count as commercial. CC0 material carries no conditions at all. Or record it yourself.

Doing it with free tools

DaVinci Resolve's Fairlight page is a full digital audio workstation attached to the same timeline you edited on, which removes the round trip that used to make small-film sound work painful. Freesound holds hundreds of thousands of Creative Commons licensed recordings, each with its licence stated on the page. Audacity is excellent for cleaning, converting and processing individual files, and genuinely poor for sound design to picture, because it does not display video, so you cannot see what you are syncing to. The Free Music Archive lists music with its licence terms attached. Blender's video sequencer will handle basic audio if you already live there.

The realistic minimum for a decent-sounding short is smaller than it looks: a clean dialogue edit with room tone filling every gap, one continuous ambience per location, footsteps wherever feet are visible and matter, three or four hard effects for the moments the story turns on, and a mix checked on three different speakers. That is a weekend of work and it will lift a film further than any camera upgrade.

Common misconceptions

• The production recording is the soundtrack. It is one layer of about seven, and on a dubbed release it is the only layer that gets thrown away.
• Noise reduction can rescue a bad recording. Three or four decibels of it is invisible; fifteen makes voices sound underwater. Reverberation cannot be removed at all.
• Public domain music is free to use. The composition may be, but the recording is a separate copyright with its own owner.
• A louder mix is a stronger mix. Platforms normalise loudness, so a crushed mix just arrives quiet and flat.
• Foley is a luxury for big productions. It is the single cheapest thing that makes a soundtrack feel professional, and its absence is what makes characters seem to float.
• Music will fix a scene that is not working. Play the scene silent. If it collapses, the problem is in the scene.

Where this leaves us

• A soundtrack is assembled from dialogue, ADR, group, Foley, hard effects, backgrounds and music, kept on separate stems.
• Stems exist partly for control and partly because a music and effects track is required for dubbing into other languages.
• Work in order: dialogue edit, backgrounds, hard effects, Foley, music, mix. Fill every dialogue gap with room tone.
• Give each location one continuous ambience, and place distant sounds with filtering and reverb rather than volume alone.
• Keep peaks near minus six decibels full scale, respect loudness normalisation rather than mastering loud, and let dialogue win every fight.
• Loud needs quiet before it. Muffling everything but a character's breathing is point of audition, and it costs nothing.
• Music needs a synchronisation licence for the composition and a master use licence for the recording; public domain covers only the composition.
• Resolve Fairlight, Freesound, the Free Music Archive and Audacity cover a short film's needs; Audacity cannot show you picture.

Sources

1. Yale University Film Studies. (n.d.). Film analysis: Sound. Yale University. filmanalysis.yale.edu
2. Wikipedia contributors. (2024). Sound design. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (2024). Foley (filmmaking). Wikipedia. en.wikipedia.org
4. Music Technology Group, Universitat Pompeu Fabra. (n.d.). Freesound: a collaborative database of Creative Commons licensed sounds. freesound.org
5. Audacity Team. (n.d.). Audacity: free, open source, cross-platform audio software. audacityteam.org

A green Mississippi summer that had to look like a drought

The Coen brothers shot O Brother, Where Art Thou? in Mississippi in the summer of 1999. The script is set in a parched, dust-blown Depression-era South. What Roger Deakins had in front of his camera was thick green foliage, wet with humidity, in full leaf. Repainting a state was not available. So the production sent the entire film through a digital intermediate: the negative was scanned, graded shot by shot in a computer at Cinesite, and recorded back out to film for release prints. The greens were pulled down towards yellows and browns, and the film that reached cinemas in 2000 looks like a hand-tinted postcard.

It is generally cited as the first feature graded digitally end to end, and the process it proved is now simply how films are finished. The reason to start here is not the milestone. It is the scale of the change: colour stopped being a chemical property of the print and became a decision made on a timeline, per shot, per region of the frame, reversible until the moment you export.

That power has a shadow, and you have already heard it on set. Somebody looks at a badly lit shot and says they will fix it in the grade. Some things really can be fixed there. Others cannot be fixed there at any price, and knowing which is which is most of this lesson.

Key idea: Grading moves and reshapes the information that exists in the file. It cannot invent information that was never recorded.

Correction and grading are two different jobs

Colour correction is technical: making each shot look right and making all the shots in a scene look like the same place at the same time of day. Colour grading is creative: giving the film a look. Correction is invisible when done well and glaring when skipped, because an audience will forgive a strange palette and will not forgive the actor's face changing colour between the wide shot and the close-up.

The order of work is fixed, and doing it out of order is the most common way to waste an evening.

1. Conform. Relink the edit to the full-quality original media instead of the proxies, and confirm every shot points at the right file.
2. Balance each shot. Set black level, white level and white balance so the shot is neutral and correctly exposed.
3. Match within the scene. Pick the best shot in the scene as the reference and bring every other shot to it.
4. Secondaries. Adjust specific regions or specific colours: darken a blown window, lift a face out of shadow, take the green out of a wall.
5. The look. Apply the creative grade over a scene or the whole film, once the underlying material already matches.
6. Deliver. Render the masters and viewing copies the world will ask for.

Applying a look before balancing is what produces the effect where a stylish grade appears to slide about from shot to shot. The look is not the problem; the unbalanced material underneath it is.

Scopes, because your eyes are unreliable witnesses

Human vision adapts. Stare at a slightly blue image for ninety seconds and your visual system will decide it is neutral, which means the second hour of a grading session systematically drifts. Add an uncalibrated laptop screen, a lamp behind you, and a coloured wall reflecting into the display, and the honest position is that you cannot trust what you see. Colourists therefore work to instruments, and check with their eyes.

Scope	What it shows	What you use it for
Waveform	Brightness plotted across the width of the frame	Setting black and white levels; spotting clipped highlights and crushed shadows
RGB parade	Three waveforms, one per colour channel, side by side	White balance: if the three channels sit at different heights in a neutral area, there is a colour cast
Vectorscope	Hue as angle and saturation as distance from the centre	Judging saturation, and keeping skin tones along the diagonal skin tone line
Histogram	How many pixels sit at each brightness	A quick read of overall exposure distribution

The vectorscope's skin tone indicator is the one that surprises people. Skin from any human being on earth, of any complexion, falls along roughly the same line on that display, at about the eleven o'clock position. Complexion changes how far out along the line the trace sits, not the angle. So if your actor's face is pulling away from that line, you have a colour cast on their face, and no amount of squinting is a substitute for looking.

Two things are still worth spending on before technique. Grade in a dim room with neutral grey or white walls behind the monitor, and turn off every coloured light. Then calibrate the display, or at the very least grade on the same screen consistently and check the result on two other devices before you call it finished.

What matters here: Set levels and balance with the scopes. Use your eyes for the creative decision, and only after you have removed the reasons they might be lying.

Log, raw, and what a lookup table actually is

Most cameras above the entry level can record a logarithmic picture profile, usually branded with the maker's name for it. Log footage looks flat, washed out and low in contrast, and that is the point: it distributes the sensor's dynamic range across the recorded values evenly so that highlights and shadows survive to be pushed around later. It is not an image. It is a container for one.

To view it, you apply a transform that maps log values back to the display standard, which for ordinary high definition delivery is Rec. 709, the ITU recommendation that fixes the colour primaries, white point and transfer characteristics of HDTV. That transform is often supplied as a lookup table, a LUT, which is simply a file mapping input colour values to output colour values.

Here is the confusion worth clearing up, because it costs people whole projects. There are two kinds of LUT and they are not interchangeable. A technical LUT converts one colour space to another and is a correct, neutral operation. A creative LUT is somebody's look, sold or shared as a file. Dropping a creative LUT onto unconverted log footage produces a mess, and dropping any LUT on at the start and then grading on top of it means you are grading an already crushed image with less to work with. Convert first, balance second, look last.

A blunt piece of advice for a first film: if you do not intend to grade seriously, do not shoot log. An eight-bit camera recording log has too few values to survive being stretched back out, and you will see banding in skies and on walls. A well-exposed standard profile from a camera you understand beats a log file you cannot process.

What a grade cannot do

Clipped highlights are gone. When a sensor pixel saturates, it records its maximum value and nothing about how far past it the real scene went, so a blown window is a flat white shape and pulling it down yields grey. Crushed blacks are the same failure at the other end. Focus cannot be corrected. Motion blur cannot be removed. Mixed lighting, a face lit half by a tungsten lamp and half by daylight through a window, can be improved with a tracked window and a great deal of patience, and cannot be made to look as if it were lit properly. Noise can be reduced, at the cost of detail, and the good temporal noise reduction is one of the features the free version of Resolve leaves out.

What a grade does extremely well is exposure within the recorded range, white balance, matching shots, controlling where the eye goes by darkening the edges of the frame or lifting a face, and building a consistent palette across a film. That list is enough to transform a short film. It just does not include rescue.

Titles, and the small legal block at the end

Titles are graphics, and beginners consistently make two mistakes with them. The first is placement. Broadcast convention keeps text inside a title safe area of roughly the central ninety percent of the frame, and important action within an action safe area of around ninety-three percent, because displays and projectors crop the edges by varying amounts. Your editing software will draw both zones on request. The second mistake is duration: read the title aloud twice at a comfortable speed, and that is the minimum time it needs to be on screen.

The end credits are also a legal document. They carry the copyright notice with the year and the owner, the production company, and every contractual credit you promised anyone, and you promised more than you remember. Keep a running credit list from day one of pre-production, because reconstructing it from memory eight months later is how people get left out, and being left out of the credits is one of the few things that reliably ends a working relationship.

Delivery: the files somebody will actually ask you for

A festival, a broadcaster and a platform want different things, and the professional habit is to make one high-quality master and derive everything else from it.

Deliverable	Typical form	What it is for
Master	A mezzanine codec such as ProRes 422 HQ or DNxHR HQ, at the finishing resolution and frame rate	The archival copy everything else is made from
Viewing copy	H.264 or H.265 in an MP4 container	Submissions, links, general viewing
Cinema copy	A Digital Cinema Package: JPEG 2000 image sequence plus audio, in a defined folder structure	Projection in a cinema; many festivals request one on selection
Subtitles and captions	SRT sidecar files, and sometimes burned-in versions	Accessibility, and any festival outside your language
Stills and artwork	High-resolution frames or unit photography, plus a poster	Programmes, press, platform pages

Two practical notes. DCP-o-matic is free, open source software that will build a valid Digital Cinema Package on an ordinary computer, which removes what used to be a genuine barrier for short films. And subtitles matter more than filmmakers expect: a large share of viewers watch with them switched on by choice, and a festival programmer in another country cannot consider a film they cannot follow.

In short: Finish once, to a high-quality master, and generate every other format from it. Never re-encode a compressed file into another compressed file.

Archiving, which is the part everyone skips

When the film is delivered, put three things somewhere safe: the master file, the camera original media, and the project files with any custom assets. Two copies, in two physical places, on drives you actually spin up once a year. Hard drives left in a drawer for five years fail at a rate that surprises people, and the cloud counts as one of your two copies only if you are paying for it and would notice if the payment lapsed. Films get remastered, recut and resubmitted years later, and the ones that cannot be are usually the ones whose only copy was on a laptop that died.

Common misconceptions

• We can fix it in the grade. Only within the recorded range. Clipped highlights, crushed blacks and soft focus are permanent.
• A LUT is a look. Some are technical colour space conversions and some are creative looks. Applying the second without the first produces a mess.
• Log footage is always better. On an eight-bit camera it bands badly when stretched, and it is only better if you are going to grade properly.
• You can judge colour on a laptop in a bright room. Your eyes adapt within seconds and the screen is not calibrated. Use the scopes.
• Grading is applying a preset. Balancing and matching shots comes first, and it is the part the audience actually registers.
• Exporting the same file at higher quality improves it. Re-encoding a compressed file only adds artefacts. Always go back to the master.

Summing up

• O Brother, Where Art Thou? in 2000 is generally cited as the first feature graded digitally end to end; the process is now standard.
• Correction makes shots right and matched; grading gives the film a look. Correction first, always.
• Work in order: conform, balance, match within the scene, secondaries, look, deliver.
• Trust the waveform, parade and vectorscope over your eyes, and keep skin tones along the vectorscope skin tone line.
• Log is a container for an image, not an image; convert with a technical transform before applying any creative look.
• A grade cannot recover clipped highlights, crushed blacks or focus, and free Resolve omits the strong noise reduction.
• Keep titles inside the title safe area, hold them long enough to read twice, and treat end credits as a contractual document.
• Make one high-quality master, derive H.264 viewing copies, SRT subtitles and a DCP from it, and archive two copies in two places.

Sources

1. Wikipedia contributors. (2024). Color grading. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (2024). Digital intermediate. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (2024). Rec. 709. Wikipedia. en.wikipedia.org
4. International Telecommunication Union. (2015). Recommendation ITU-R BT.709-6: Parameter values for the HDTV standards for production and international programme exchange. ITU Radiocommunication Sector.
5. Blackmagic Design. (n.d.). DaVinci Resolve. Blackmagic Design. blackmagicdesign.com

A wooden board full of coloured cardboard

For most of the twentieth century a film's schedule was a physical object: a wooden frame holding narrow cardboard strips, one strip per scene, each strip carrying the scene number, the set, whether it was interior or exterior, whether it was day or night, its length in eighths of a page, and the numbers of the cast members in it. The strips were colour coded by convention. White for a day interior, yellow for a day exterior, blue for a night interior, green for a night exterior. Rescheduling the film meant physically pulling strips out and sliding them back in somewhere else, and an assistant director could see the whole shoot at a glance because the colours clustered.

Software replaced the board, kept the colours, and kept the unit that makes the whole thing work: the eighth. A screenplay page is divided into eight strips of roughly one inch, and every scene is measured in them. A scene that runs a third of a page is three eighths. A scene of two and a half pages is 2 4/8. This sounds fussy until you try to answer the only question that matters in scheduling, which is whether a day's work fits in a day. Pages are too coarse and minutes are unknowable in advance. Eighths are countable, and a crew's daily capacity in eighths is something you learn about yourself after two shoots.

This lesson builds a schedule and a budget for one small film, end to end, and then changes a single assumption to show how fast both documents fall over.

Why this matters: A budget and a schedule are the same document expressed in two currencies. Anything that adds time adds money, and almost anything that saves money costs time.

The breakdown: reading the script as a shopping list

Before either document exists, somebody reads the script scene by scene and lists every element it requires. Traditionally this was done with coloured pencils on the script itself, one colour per category, and the categories are still the standard headings on a breakdown sheet: cast, background and extras, stunts, props, vehicles, wardrobe, makeup and hair, special effects, animals, music, sound effects, special equipment, and notes.

The discipline is literalism. If the script says she pours the last of the whisky, then the breakdown says: whisky bottle, contents to be cold tea, matching second bottle for resets, glass, and cloth for spills. If it says he checks his phone, then somebody has to decide what is on the screen, get it there, and check that the phone is charged, because a dead phone stops a unit of fourteen people in a way that is genuinely embarrassing.

Out of the breakdown falls the document that constrains the schedule most: the day out of days, a grid of cast members against shooting days that shows exactly who is needed when. Actors are the least flexible resource on a small film, because they have jobs, and the schedule you can afford is usually the one that gets each actor in and out in the fewest days.

The scheduling rules, in priority order

Shooting order has almost nothing to do with script order, and beginners find this genuinely disorienting until they price a company move. The rules, roughly in order of how much they cost to break:

1. Group by location. Everything that happens at a location is shot while you are there, regardless of where it falls in the story. Moving a unit is the single most expensive event in a day: it can cost two hours between the last shot at one place and the first shot at the next.
2. Within a location, group by time of day. All the day interiors, then all the night interiors, because relighting a room from day to night takes a lighting crew an hour or more.
3. Then group by cast. Cluster an actor's scenes so they work three consecutive days rather than three days spread over two weeks.
4. Put exteriors early in the schedule. Weather will eventually cost you a day, and you want somewhere in the plan to move it to. A scene with no cover set is a scene that can stop the film.
5. Shoot the hardest scene when you are strong, not on day one. Day one is when the crew is still learning each other. Day two or three is when the machine works and everyone is still fresh.
6. Respect turnaround. The gap between wrap and the next call is not a courtesy. Ten or eleven hours is the normal minimum, and driving home exhausted is how crews get hurt.

A last one, which is arithmetic rather than judgement: dawn and dusk are on a timetable you do not control. If a scene needs the light at the end of the day, you get roughly twenty to thirty minutes of it, once per day, and no amount of enthusiasm extends it. Schedule around the sun before you schedule around anything else.

The worked example: a twelve-minute short called Signal

Everything below describes one imagined production, and the figures are illustrative arithmetic rather than market rates. Price your own locally, because rates differ enormously by country and city. Figures are in pounds sterling.

The assumptions. The script runs twelve pages, so roughly twelve minutes. Four speaking parts. Three locations: a first-floor flat, the interior of a parked car at night, and a canal towpath at dawn. The flat belongs to the director's aunt and costs nothing. Nobody is being paid a fee, and everybody is being fed, insured and reimbursed for travel. Crew of nine. Three shooting days.

Day	Set	Scenes	Eighths	Cast
1	Flat, day interior	1, 4, 5, 9	34	1, 2
2	Flat, night interior, then car, night interior	10, 11, 12, 6, 7	30	1, 2, 3
3	Towpath, dawn exterior, then flat pickups	2, 3, 13	32	1, 4

Read what the schedule has quietly done. The story's opening scene, on the towpath, is shot last, because the towpath is the only location with no control over light and it sits at the end where a weather day can be absorbed. The flat is used on all three days, so there is only one real company move, on day two, and it happens after dark when the car scenes can be shot in a supermarket car park two streets away. Cast member 4 works one day. Cast member 3 works one evening. Twelve pages across three days is about thirty-two eighths a day, which is brisk but survivable for a small unit shooting a dialogue piece with a fixed set.

Category	Detail	Cost (GBP)
Cast	Four performers, no fee, travel and expenses reimbursed	320
Crew	Nine people, no fee, travel reimbursed	450
Camera	Camera body and three lenses, three days plus a prep day, borrowed body	360
Lighting and grip	Small LED kit, stands, flags, sandbags, three days	420
Sound	Recorder, shotgun microphone, boom pole, two radio microphones	240
Art department	Set dressing, props, whisky substitute, wardrobe purchases and returns	300
Locations	Flat free; car park permission free; towpath permit and marshalling	150
Catering	Thirteen people, three days, two hot meals and a running supply of snacks	390
Transport	Van hire two days, fuel, parking suspensions	310
Insurance	Short-period public liability and equipment cover	280
Media and backup	Two memory cards, two hard drives	220
Post-production	Free software, composer commissioned for a credit, one paid sound day	250
Delivery and festivals	DCP created free, subtitle files, twelve festival entry fees	420
Subtotal		4110
Contingency at 10 percent	The line you will actually spend	411
Total		4521

Three things in that table are worth arguing about, because they are where first films go wrong.

Catering is not a soft line. It is nearly ten percent of the budget and it is the line most often cut first. Do not cut it. Crews that are fed work faster, stay later without resentment, and come back for the next film. Crews that are handed a supermarket sandwich at four in the afternoon do none of those things, and the reputation travels.

Festival fees are a real budget category. A dozen entries at thirty pounds each is more than the camera hire, and this line is invisible to almost everybody planning a first short, because it arrives months after the shoot when the money is gone.

Contingency is not padding. It is the line item for the events you cannot name in advance but can name in category: a broken lens, a parking fine, a cast member with food poisoning, a hard drive that fails on the second night. Ten percent is a floor, not a cushion.

The point: The budget is not a prediction. It is a set of decisions written down early enough to be argued with while arguing is still cheap.

Now change one thing

Suppose the aunt sells the flat, and the only comparable interior available charges a location fee of 400 pounds a day. Watch what happens, because it does not stop at one line.

Three days at that location is 1200 pounds, which alone is more than a quarter of the original budget. So you compress to two days, which pushes the daily load from about thirty-two eighths to about forty-eight. That is a twelve-hour day rather than a ten-hour day, which means overtime on the van, an extra meal for thirteen people, and a real risk of losing the towpath dawn because a fourteen-hour day on Saturday makes a five o'clock call on Sunday a turnaround violation and an unsafe drive. To protect the dawn you drop scene 9, which is the scene where the audience learns why she stopped answering the phone. Now the edit has a hole in it that no amount of grading will fill.

The honest options at this point are three, and all three are legitimate: find another free flat and shoot the schedule you had, raise another eight hundred pounds, or rewrite scene 9 into scene 5 so twelve pages become ten. Notice that the third one is free and the first two are not, and that it is a writing problem being solved by a producer's arithmetic. That is the job.

What you owe people who work for nothing

Most first films are made by people who are not being paid, and there is a version of that arrangement which is fine and a version which is exploitation. The difference is specific and worth stating plainly.

Fine looks like this: everyone knows before they commit that there is no fee, the dates are fixed and short, travel is reimbursed the week it is spent, there is proper hot food, the days end when the call sheet said they would, the credit is agreed in writing, and everyone gets a copy of the finished film and a reference. Not fine looks like this: an implication that payment might appear, days that overrun by three hours because the plan was fantasy, and a director who disappears for eight months and never delivers the film that was the whole point of working free.

Deferred payment deserves one sentence of realism. On a short film, deferred means never, and a crew member who has heard the word before knows that. If you use it, say what you mean: that the film is unlikely to earn anything, and that a share of nothing is nothing.

Cash flow, which is a separate problem from budget

A production can be exactly on budget and still fail, because a budget says what the film costs and cash flow says when. The equipment hire is due on collection. The van deposit is due before you drive it. The insurance is due before the hire company will release anything. Reimbursements go out the week people spend, not at the end, because a runner who is owed sixty pounds for three weeks is being asked to fund your film out of a student loan. Draw the timeline of when money leaves, not just how much, and you will discover the week where it does not work.

Doing it without paid software

Dedicated scheduling and budgeting packages exist and are used industrially. You do not need one. A spreadsheet with one row per scene, columns for set, day or night, interior or exterior, eighths and cast numbers, sorted by location, is a stripboard. A second sheet with categories and line items is a budget. Write the script in Fountain, a plain text screenplay syntax, and any of the free tools that read it will produce a properly formatted page count for you. For pay rates, do not guess: national statistics agencies publish occupational pay data, and union and guild rate cards for your country are published and free to read.

Common misconceptions

• Films are shot in story order. They are shot in location order, then time of day, then cast, because company moves and relights are the expensive events.
• Contingency is padding you can trim. It is the line for the broken lens and the failed drive, both of which are certain in category if not in detail.
• If nobody is paid, the film is free. Food, travel, insurance, media, transport and festival fees are still real, and they add up to thousands.
• We can pick up whatever we miss later. Reassembling four actors, nine crew and a location for one scene is harder and costlier than the day it would have taken.
• A twelve-page script is an easy day. Around thirty eighths a day is a working pace for a small unit. Forty-eight is a twelve-hour day with consequences the next morning.
• Being on budget means the money works. Budget is how much; cash flow is when. Films die of the second while the first still balances.

Putting it together

• Scenes are measured in eighths of a page, and the coloured strip conventions are white day interior, yellow day exterior, blue night interior, green night exterior.
• A breakdown lists every element by category and produces the day out of days, which shows exactly when each actor is needed.
• Schedule by location, then time of day, then cast; put exteriors early; do not shoot the hardest scene on day one; protect turnaround.
• In the worked example, three days at roughly thirty-two eighths each cost about 4500 pounds with nobody paid a fee.
• Catering, festival fees and contingency are the three lines beginners cut and then regret; contingency of ten percent is a floor.
• Losing one free location cascaded into a compressed schedule, a turnaround risk, and a scene cut from the film.
• Unpaid crew are owed fixed short dates, prompt reimbursement, real food, an agreed credit, and a finished film.
• A spreadsheet is a sufficient stripboard and budget; Fountain will give you a real page count for nothing.

Sources

1. Wikipedia contributors. (2024). Film budgeting. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (2024). Pre-production. Wikipedia. en.wikipedia.org
3. U.S. Bureau of Labor Statistics. (n.d.). Occupational Outlook Handbook: Film and video editors and camera operators. U.S. Department of Labor. bls.gov
4. British Film Institute. (n.d.). BFI: funding, industry data and support for filmmakers. BFI. bfi.org.uk

A railway bridge in Wayne County, Georgia

On 20 February 2014, the first day of shooting on Midnight Rider, a film about Gregg Allman, a crew carried a hospital bed onto a CSX railway trestle over the Altamaha River in Wayne County, Georgia, to shoot a dream sequence. The railway had refused the production permission to film on the tracks. They went out anyway. A freight train came. Sarah Jones, a 27-year-old second camera assistant, was struck and killed, and several other crew members were injured. The film was never finished.

What followed was not treated as an accident. The director, Randall Miller, pleaded guilty to involuntary manslaughter and criminal trespass and was sentenced in 2015 to ten years, of which two were to be served in jail. Other members of the production were also convicted. The case changed the conversation on film sets in a way that decades of safety bulletins had not, because it established in front of an entire industry that a producer who takes a crew somewhere they have been told they cannot go is committing a crime, not making a bold creative choice.

Hold on to the specific shape of the failure. It was not equipment. It was not bad luck. It was a decision, made in an office in advance, to shoot on property the production had been refused access to, with a live railway line running through the shot and no way to stop a train. Nearly every serious injury on a film set traces back to a producing decision like that one, taken earlier, when it was still cheap to make differently.

The second name to know is Halyna Hutchins, the cinematographer killed on 21 October 2021 on the set of Rust in New Mexico, when a revolver being used in a rehearsal discharged a live round. The film's armourer was later convicted of involuntary manslaughter. Again the failure was procedural rather than mechanical: live ammunition should not have been within miles of that set.

Remember: Safety is not a department. It is a set of decisions taken in pre-production, when saying no still only costs a rewrite.

Who is responsible, and what a risk assessment is

The producer carries responsibility for the working conditions on a production, and that responsibility is legal rather than moral. Regulatory regimes differ, so learn the one where you are shooting, but the structure is remarkably consistent internationally. In the United States, the Occupational Safety and Health Act's general duty clause requires an employer to provide a workplace free from recognised hazards likely to cause death or serious harm. In the United Kingdom, the Health and Safety at Work etc. Act 1974 places duties on employers, the self-employed and anyone in control of premises. Nothing in either exempts a production because it is small, unpaid, or a student project.

The working instrument is a risk assessment, and it is much less intimidating than it sounds. It is a table. For each activity, you write the hazard, who could be harmed and how, the controls you are putting in place, and who is responsible for each control. A short film's risk assessment fits on two sides of paper. Writing it is also the moment you discover the problems: you cannot write down a control for filming on a live railway line, because there is not one, which is exactly the information you needed before you booked the day.

Two habits do more than any document. First, a safety briefing at the start of every shooting day, out loud, naming the specific hazards of that location: the cable run across the doorway, the drop at the edge of the towpath, the fact that the front door locks behind you. Second, a stated rule that anyone, of any rank, can stop the shoot for safety without needing to justify it first, and that nobody will be punished for a stop that turns out to be unnecessary. Say both on day one, in front of everybody, and mean them.

The hazards that actually turn up on small sets

Hazard	How it goes wrong	Control
Electrical load	Several lights and a kettle on one domestic circuit; daisy-chained extension leads	Know the circuit rating, spread the load, one extension per socket, and a fire extinguisher present
Cables and trip hazards	Cable across a doorway in a dark corridor	Route around, tape down, mat over, and name it in the briefing
Stands and overhead	A light on a stand falls onto a performer	Sandbag every stand, never rig above people, safety chains on anything overhead
Vehicles	Filming from or into a moving car; shooting on a live road	Permits, marshals, closed roads, and never operating a camera while driving
Water, night and cold	Riverbanks, docks, unlit exteriors, long waits in low temperatures	Lighting for the crew as well as the shot, a competent water safety person, warm shelter and hot drinks
Weapons	Real or realistic firearms; blades; police called by a passer-by	A qualified armourer, no live ammunition ever, weapons transported cased, and the local police notified in advance
Stunts and falls	An untrained performer does something that looks easy	If it needs a stunt, hire a coordinator; if you cannot afford one, rewrite the scene
Fatigue	A fourteen-hour day followed by an hour's drive home	Honest schedules, real turnaround, and paid transport or accommodation when a day overruns

The last row kills more film workers than any other line in that table, and it is the one nobody photographs. A schedule that was fantasy on paper becomes an exhausted driver at one in the morning. That is why the budget lesson treated turnaround as arithmetic rather than courtesy.

Chain of title: why a good film can become unreleasable

A finished film is a bundle of other people's rights. The writer owns the script, the composer owns the music, the performers have rights in their performances, the owner of that painting on the wall owns the painting, and the person who shot the news footage you cut in owns that. Chain of title is the paper trail proving that the production has acquired or licensed every one of them. It is the first thing a distributor, a broadcaster or a festival with a serious legal department will ask to see, and it is the reason perfectly good films sit in drawers.

The documents, at minimum, are these.

• Writer agreement. An assignment or licence of the screenplay's copyright to the production, signed, even when the writer and the director are the same person and especially when they are two friends.
• Crew agreements. Short, plain, and containing an assignment of any copyright arising from their work, plus the agreed credit.
• Performer releases. One from every person who appears recognisably on screen.
• Location agreements. Written permission from whoever actually controls the property.
• Music licences. Synchronisation for the composition, master use for the recording, as the last lesson set out.
• Clearances. Artwork, photographs, posters, brand logos, third-party footage, and anything visible that somebody else made.

The rule that saves you is unglamorous: one folder, kept from the first day of pre-production, containing a signed copy of everything, named consistently. You will be asked for it three years later by someone who will not accept a promise that it exists somewhere.

In short: If you cannot prove you have the right to use something in the film, you do not have the right to use it, whatever anybody said on the day.

Releases, and the reason they are signed before the shoot

A performer release is a short document naming the person, the production and the date, granting the production the right to use their appearance, voice and performance in the film in all media, and confirming that there is no obligation to use it. It should be signed before the shooting day. This is not paranoia; it is leverage arithmetic. Before the day, an unsigned release means finding another performer. After the day, an unsigned release means a person who is in every shot of scene four holding a veto over your film, and people do fall out.

Some specifics that catch people out. Anyone under eighteen needs a parent or guardian to sign, and many jurisdictions add licensing, chaperoning and working-hours rules for child performers that are strict and enforced. Extras and background performers need releases too. A location agreement should name the dates, the hours, what you are permitted to do, who is liable for damage, and, critically, be signed by someone with the authority to sign it, which a helpful caretaker with keys usually is not. Distinctive private property, recognisable artworks and trademarks may need their own clearance depending on where you are and how prominently they feature.

Filming members of the public in a public place is the area where general advice is least reliable. In some countries it is broadly lawful; in others, privacy and data protection law makes it considerably more complicated, and rules for a documentary can differ from those for a drama. The workable practice everywhere is the same: post clear signs at every entrance to the filming area, announce what is happening, and take a release from anyone who is recognisable and significant in the frame. Then check the actual rules where you are shooting, because this paragraph is a starting point and not legal advice.

Permits: asking is the whole point

Shooting on a public street, in a park, or on any public land normally needs a permit from a local film office or council. New filmmakers resent this and misunderstand what it buys. The permit is not a tax on creativity. It is what gets you the suspended parking bays your van needs, the notification to the police so an armed response unit does not attend your gangster short, the notice to residents that stops a complaint becoming a shutdown, and the standing to say yes when a stranger asks who authorised this.

Private property needs written permission from the owner. Railways, ports, waterways, airports, motorways, hospitals and schools each have their own authority, their own process, and their own genuine ability to say no. Drone flying is regulated by the national aviation authority almost everywhere, typically requiring registration, a competence certificate and rules about distance from people and airports, and none of that is waived because the flight is short. When the answer comes back no, the answer is no. That is the sentence the Midnight Rider production did not accept.

Insurance, and what each policy is actually for

Cover	What it protects against	When you need it
Public liability	Injury to members of the public or damage to third-party property	Almost always; location owners and permit offices normally require proof
Employers liability	Injury to your own people, including unpaid crew in many regimes	Legally compulsory in some countries, including the UK; check locally
Equipment and hired-in cover	Loss, theft or damage to owned and rented gear	Rental houses will not release equipment without it
Cast insurance	The cost of reshooting if a principal performer cannot continue	When losing one actor would cost more than the premium
Errors and omissions	Claims about rights, clearance, defamation or privacy in the finished film	At distribution; broadcasters and platforms usually insist on it

Short-period production policies exist and are affordable at short-film scale, which is why the worked budget in the last lesson carried an insurance line of 280 pounds rather than nothing. Errors and omissions cover is the one that connects back to chain of title: an insurer will price it from your paperwork, and a production with a complete folder gets a policy while a production with a shrug does not.

The upshot: Insurance is not the thing that makes a set safe. It is the thing that stops one bad afternoon from ending everyone's finances, and it is priced on how carefully you documented everything else.

Common misconceptions

• Student and unpaid productions are exempt from safety law. They are not. Duties attach to whoever controls the work and the premises.
• Releases can be collected afterwards. After the day you have no leverage and the performer has a veto over every shot they are in.
• It is a public place, so anyone can be filmed. This varies enormously by country, and privacy and data protection law can apply. Sign, announce, and check locally.
• Permission from whoever has the keys is permission. The agreement must be signed by someone with authority over the property.
• A weapon that has been checked is safe. Rust turned on procedure, not on a single check. No live ammunition ever comes near a set.
• Insurance is for big productions. Short-period policies are cheap, and most locations and rental houses will not deal with you without one.

Recap

• Sarah Jones was killed on 20 February 2014 on a Georgia railway trestle the production had been refused permission to use; the director was later jailed.
• Serious set incidents almost always trace to a producing decision made earlier, not to an accident on the day.
• The producer holds legal responsibility; a written risk assessment listing hazards, harms, controls and owners is the working instrument.
• Brief the specific hazards every morning, and establish out loud that anyone may stop the shoot for safety without penalty.
• Fatigue and the drive home injure more film workers than any single piece of equipment, which is why turnaround is non-negotiable.
• Chain of title means signed writer, crew, performer, location, music and clearance paperwork, kept in one folder from day one.
• Releases are signed before the shooting day; minors need a guardian and often a licence, and location agreements need a signatory with authority.
• Permits buy parking, police notification and standing; aviation, railway and similar authorities can and do refuse, and no means no.
• Carry public liability, employers liability where required, equipment cover, and errors and omissions at distribution.

Sources

1. Occupational Safety and Health Administration. (n.d.). Worker safety and health. U.S. Department of Labor. osha.gov
2. Health and Safety Executive. (n.d.). Health and safety guidance for employers. HSE. hse.gov.uk
3. U.S. Copyright Office. (n.d.). Copyright basics, Circular 1. Library of Congress. copyright.gov
4. FilmLA. (n.d.). Film permits and permitted production activity. FilmLA. filmla.com
5. Wikipedia contributors. (2024). Model release. Wikipedia. en.wikipedia.org

Wisconsin, April 1960, walking backwards up a staircase

Robert Drew's team followed John F. Kennedy and Hubert Humphrey through the Wisconsin Democratic primary with equipment that had not existed a few years earlier. Richard Leacock, D. A. Pennebaker and their collaborators had helped rebuild a 16mm camera so one person could carry it on a shoulder while a separate recorder captured synchronised sound without a cable joining the two. In the resulting film, Primary, Albert Maysles walks backwards up a staircase in a single unbroken shot, ahead of Kennedy, through a crowd pressing in on both sides, into a hall.

Before that rig, sync sound meant a blimped camera on a tripod, a cable, and a scene that waited for you. After it, a documentary could go where people actually were and record what they actually said while they were doing something else. The technology did not merely make filming easier. It changed what documentary could be about, from prepared statements to behaviour.

It also started an argument that has never been settled, and the argument is more useful to you than the history.

Two positions on what a camera does to a room

The American filmmakers around Drew took the view that the film should observe and not intervene. No interviews, no narration, no questions from behind the camera, no staging. Shoot enough, stay long enough, and people forget you are there; what you capture is behaviour rather than performance. The evidence they offered was the footage itself: the Maysles brothers filming a Bible salesman's exhaustion in a motel room, Pennebaker filming Bob Dylan being unpleasant to a journalist. Nobody performs that.

In Paris in 1961, Jean Rouch and Edgar Morin made Chronique d'un ete on the opposite principle. They put the camera in front of people and asked them directly whether they were happy. They appear in their own film discussing what they are doing. At the end they screened the assembled footage to the participants, filmed them arguing about whether it was truthful, and put that argument in the finished film. Their claim was that a camera always changes a room, and that a film pretending otherwise is not more honest but less, because it hides its largest intervention.

What would settle it? One part is empirical and has an answer: people do behave differently when filmed, and the effect diminishes with time and familiarity but does not vanish. Both camps accept this. The disagreement is about what follows from it. Direct cinema says that since the effect fades, patience gets you close enough to behaviour to be worth having. Cinema verite says that since the effect never fully fades, the responsible move is to make the camera visible and work with the reaction rather than against it.

Contemporary documentary mostly declines to choose. A single film will hold observational sequences shot over months, sit-down interviews that no one pretends are spontaneous, archive, and sometimes a director's voice admitting what they wanted and did not get. Knowing which tradition a technique comes from is what stops you using it out of habit.

What matters here: Decide, in advance and explicitly, what your camera is claiming to be in the room. Films get into trouble when they mix an observational promise with a staged reality and hope nobody notices.

The interview, which is a craft and not a conversation

Most documentary interviews fail for reasons that have nothing to do with the questions. Deal with those first. Put the subject in a soft room, not a corner with hard walls; the sound lesson explains why. Seat them so the light falls across their face rather than flatly into it. Place yourself just beside the lens so their eyeline falls slightly off camera, which reads as thoughtful, or directly down the lens using a reflecting device if you want the confrontational version. Frame with the shoulders in, and get a second angle if you have a second camera, because it lets you cut without a jump.

Then the questions. Three techniques do most of the work.

Ask for the scene, not the summary. A question like "was it difficult?" earns a yes. A question like "walk me through the morning you found out" produces a person in a kitchen with the radio on, which is a scene, and scenes are what films are built from.

Ask for the answer to contain the question. Say plainly at the start that you will not be in the film, so instead of "yes, in 1994" you need "I moved to Leeds in 1994." Otherwise you will discover in the edit that half your best answers are unusable without a narrator you did not want.

Wait. The three seconds after a subject stops speaking are the most productive in the interview, and the instinct to fill them with the next question is almost irresistible. Sit still. People finish their thought, then, in the silence, say the truer version of it. Close with the question every documentary maker learns eventually: what have I not asked you about?

Observational shooting: the material you will actually need

Beginners shoot the event and nothing else, and then discover in the edit that they cannot get into or out of it. Shoot before people arrive and after they leave. Shoot the room without anybody in it. Shoot hands doing the work, the sign on the door, the view from the window, the drive there. These are not filler; they are the connective tissue that lets you compress two hours into ninety seconds without a cut that lies.

Keep rolling through the moments that feel like nothing. The camera going up before the subject is ready, the walk from the car, the joke afterwards when the shoulders drop: these are frequently where the film is. And record five minutes of clean sound at every location, because a documentary edit rearranges time constantly and you will need that ambience to hide every join.

Expect a large shooting ratio. Fifty hours of material for a thirty-minute film is unremarkable in documentary, and that ratio is the reason the next section exists.

Finding the structure: the paper edit

In fiction, the script is written, then shot, then cut. In documentary, the writing happens in the edit, and the tool for it is a paper edit.

Transcribe everything. This used to be the most expensive step in a documentary and it has become nearly free, because automatic speech recognition is now good enough for a working transcript and open models such as Whisper run on an ordinary computer. Then read the transcripts, not the footage. Mark the passages that are alive, tag each one with a theme, and physically arrange them, on paper or in a document, into an order that tells a story. You will do this five or six times, and it costs minutes rather than the hours the same experiment costs on a timeline.

What you are looking for is the same thing fiction looks for: a person who wants something, an obstacle, and a change. Documentary subjects rarely announce it. Occasionally the film turns out to be about something other than what you set out to make, and the discipline is to notice rather than to force the original plan through.

One warning about narration. When a documentary structure is not working, the reflex is to write a voice-over that explains the connective logic. Sometimes narration is the right form and used deliberately it is excellent. Used as a patch, it announces that the film could not find its shape, and the audience hears an argument being asserted rather than watching one being demonstrated.

Worth holding on to: The paper edit is the cheapest place to be wrong. Every restructure you try on paper is a restructure you did not spend a week discovering on the timeline.

The obligations you take on when someone lets you film them

A documentary subject signs a release before the film exists, which means they consent to something neither of you can describe yet. That asymmetry is the ethical centre of the form, and there is no procedure that dissolves it. What experienced makers do instead is treat consent as something maintained rather than obtained.

Be specific at the outset about what the film is and where it might be seen, and go back to the subject when the film changes substantially. Distinguish clearly between editorial control, which you do not give away, and a courtesy screening before release, which many documentarians offer and which costs little except nerve. Think about what happens to your subject after the film exists: someone who tells you something about their employer, their family or their government is exposed in a way you are not, and in some contexts that exposure is dangerous. Anonymity, blurring and a decision not to use material are all legitimate outcomes.

Paying subjects is generally avoided, because payment gives a person a reason to say what they think you want. Covering costs, travel and lost wages is a different matter and is normal. And whatever you promise on the day, write it down, because you will remember it differently in a year.

Clearance, which is heavier than a drama's

A documentary tends to include material somebody else made: news footage, family photographs, a television playing in the corner, a song on a car radio, a painting on a wall. Each of these is a separate right, and the last lesson's chain of title folder applies with more force here. Archive holders licence footage at rates that vary by term and territory, and archive costs have sunk more documentaries than equipment ever has. Some jurisdictions have fair use or fair dealing provisions that permit limited quotation for criticism and comment, but they are narrower and more fact-specific than filmmakers hope, and a distributor's insurer will want an opinion rather than an optimism. National film and sound archives, including the National Film and Sound Archive of Australia and equivalents elsewhere, publish their access and licensing terms, and some material is genuinely free to use.

Short form is not a small feature

A short film that fails is usually a feature idea compressed. A short that works does one thing. It has a single situation, one turn, and an ending that lands rather than a story that concludes. The economical way to think about it: arrive as late into the situation as you can and leave as early as the meaning allows.

Two practical constraints shape the form more than any aesthetic argument. First, a short film has roughly ninety seconds to earn its audience's attention, whether that audience is a festival programmer with two hundred submissions or someone on a sofa with a remote. Second, programming is arithmetic: a fifteen-minute film fits into a ninety-minute block in a way a twenty-eight-minute film does not, so shorter films are materially easier to programme, and many festivals cap the category outright. Check whether credits count towards the stated maximum, because in many festival rules they do.

Vertical and social short form deserves an honest paragraph rather than a sneer. Its grammar is genuinely different: no establishing shots, the subject arriving in the first frame, faces held large because the screen is small, and legibility with the sound off, which makes burned-in captions structural rather than decorative. Much of what this course teaches transfers less well there than people claim. But the discipline of the first three seconds transfers completely, and it is the discipline most short films lack.

Common misconceptions

• Documentary cannot be planned because it is unscripted. You plan access, questions, coverage and structure hypotheses. What you do not plan is what people will say.
• Observational filming is objective. Where the camera is, when it rolls and what survives the edit are all authorship. Direct cinema claims restraint, not neutrality.
• A signed release settles the ethics. It settles the law. Consent to a film that does not exist yet is why documentarians treat consent as ongoing.
• Narration will fix a structural problem. As a form it is excellent; as a patch it tells the audience the film could not find its shape.
• Archive footage is background material. It is licensed material with its own costs, and archive has sunk more documentaries than equipment.
• A short is a calling card for a feature. Some are. The ones that travel are the ones that work completely as short films.

What to remember

• Portable synchronised sound around 1960 made Primary possible and changed documentary from statements to behaviour.
• Direct cinema observes without intervening; cinema verite, in Chronique d'un ete, provokes and shows its own process. Both accept that cameras change rooms.
• Decide what your camera is claiming to be, and do not mix an observational promise with staged reality.
• Interview technique: a soft room, an eyeline just off lens, questions that ask for scenes, answers that contain the question, and silence.
• Shoot before and after, shoot hands and rooms and doors, keep rolling through the moments that feel like nothing, and record ambience everywhere.
• Structure is found through a paper edit built on full transcripts, which automatic speech recognition has made nearly free.
• Consent is maintained rather than obtained; keep editorial control, consider a courtesy screening, and think about the subject's life after release.
• Documentaries carry a heavier clearance burden than dramas, and archive costs are a real budget category.
• A short does one thing: enter late, leave early, earn attention in ninety seconds, and check whether credits count towards the festival maximum.

Sources

1. Encyclopaedia Britannica. (n.d.). Documentary film. Encyclopaedia Britannica. britannica.com
2. Wikipedia contributors. (2024). Direct cinema. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (2024). Cinema verite. Wikipedia. en.wikipedia.org
4. National Film and Sound Archive of Australia. (n.d.). Collection access and licensing. NFSA. nfsa.gov.au

The short that existed because the feature could not be financed

Damien Chazelle had written a feature about a drummer and a teacher who bullies him, and could not get anybody to pay for it. So in 2013 he took one sequence out of the script, shot it as an eighteen-minute short with J. K. Simmons in the teacher's role, and sent it to the Sundance Film Festival, where it won the Short Film Jury Prize. Money for the feature followed. Whiplash was shot in under a month, premiered at Sundance in January 2014 taking both the Grand Jury Prize and the Audience Award, and won three Academy Awards the following year.

That is a real path, and it is worth being precise about why it worked, because the wrong lesson is easy to draw. The short was not a general demonstration of talent. It answered the specific question the financiers were asking, which was whether an audience would sit through a man screaming at a drummer for two hours. Chazelle did not make a calling card. He made evidence.

Other first films did other jobs. Martin McDonagh, a playwright with no film experience, made a short called Six Shooter that won the Academy Award for Live Action Short Film in 2006 and started a directing career. Neill Blomkamp made a six-minute science fiction short, Alive in Joburg, in 2005, and it became District 9 four years later with Peter Jackson producing. These are not templates. They are proof that the door is not locked, which is different from the claim that walking through it is common.

Key idea: Decide what you want your first film to do before you decide what it is about. Learning the craft, reaching an audience, and proving one specific thing to one specific person produce three different films.

The economics, stated honestly

A short film is not a product with a market. There is a small licensing trade, in which broadcasters, airlines and a handful of curated platforms pay modest fees for short work, and there is prize money at some festivals, and neither adds up to a business. The overwhelming majority of short films lose every pound spent on them and were always going to. This is not a failure of the form; it is what the form is for. A short film is a piece of work, an education, and an argument for the next thing.

Two consequences follow, and both are practical. First, do not finance a short in a way that requires it to earn, because it will not. Second, since the film cannot pay you back in money, be deliberate about what it does pay you back in: a skill you did not have, a collaborator you now trust, an audience that found you, or a piece of evidence you can hand to someone who is deciding whether to fund you.

Festivals: how the circuit actually works

The circuit is layered, and treating it as one undifferentiated pile is the most expensive mistake a first-time submitter makes.

Tier	Examples	What it gives you
Major international	Cannes, Berlin, Venice, Toronto, Sundance, Locarno	Industry attention, press, and a line that opens doors elsewhere
Major short film specialists	Clermont-Ferrand, among the largest short film festivals in the world	An audience that has chosen to watch shorts, and a genuine market
Strong regional and specialist	National, city and genre festivals with real programming teams	Screenings, local press, and the filmmakers you will work with next
The long tail	Everything else, including online-only events with no screening	Sometimes a real audience; sometimes only a laurel and an invoice

The rule that governs strategy is premiere status. Many of the festivals at the top of that table require a world, international or national premiere, meaning your film must not have screened publicly, or not in that territory, before theirs. Screening at a small local festival in March can therefore make you ineligible for a much larger one in September. So the order of submission matters more than the number of submissions: start at the top of your realistic list and work down, and read each festival's premiere requirement before you send anything anywhere.

Submissions mostly run through platforms, of which FilmFreeway is the most widely used. Using the platform is free; each festival sets its own entry fee, and fees rise as deadlines approach, so an early submission is materially cheaper. Fee waivers exist and are given out more often than people think, particularly to student films and to filmmakers from countries where the fee is disproportionate. Ask politely, once, by email.

What a programmer is actually looking at

A programmer at a mid-sized festival may watch several hundred short films for a few dozen slots. They are watching on a laptop, often with headphones, often after a full day of work. Several practical things follow.

Your screener link must work, in every browser, on the first attempt, with the password you supplied, and it must be downloadable if the festival asks for that. A link that fails is a rejection, and nobody will email you to say so. Your first ninety seconds carry disproportionate weight, and while it is uncomfortable to hear, it is simply how a viewing pile gets processed. Your submission package should include a fifty-word synopsis, a longer one of about one hundred and fifty words, one high-resolution still that is actually a good photograph rather than a frame grab of somebody blinking, a poster, a short director statement, and correctly spelled credits. Subtitles in English, or in the festival's language, widen your eligible list enormously and cost you an afternoon.

Then the part that keeps people sane: rejection mostly is not a verdict on quality. Programming is a jigsaw. A festival that has already selected two films about grief in a kitchen will not take a third, however good, and it will not tell you that was the reason. Submit to somewhere between ten and twenty carefully chosen festivals, budget for it as the worked budget did, and treat each rejection as one piece of information about fit rather than fifty pieces about your worth.

One specific mechanism is worth knowing because it is public and rule-based. The Academy of Motion Picture Arts and Sciences maintains a list of qualifying festivals, and winning particular awards at one of them makes a short eligible for Academy Award consideration without a theatrical run. If that matters to you, read the current rules on the Academy's own site rather than a summary, because the list and the conditions change.

So what? Sequence your submissions by premiere requirement, spend the fees on a short considered list, and make the screener and the stills as reliable as the film.

Online release, and the sequencing that protects it

Putting a film on the internet is a public screening, and most festivals treat it as one. A film freely available online is ineligible for a large part of the circuit, and uploading it during your festival run can disqualify you from the festival you are waiting to hear from. The conventional sequence is therefore: festivals for twelve to eighteen months, then online.

The exception is deliberate and legitimate. If your goal is an audience rather than a circuit, going straight online is a coherent strategy, and it is the right one for a film whose subject has a community already looking for it. What is not coherent is doing both by accident.

When you do release online, remember that uploading is not distributing. A film on a platform with no context reaches nobody. Give it a page of its own with a synopsis, credits, stills and a way to contact you. Add subtitles, because a large share of viewers watch with them on. Tell the people who are already interested, which means the crew's networks, the communities connected to the subject, the location, and any organisation that appears in it. And accept that the numbers will be smaller than you hope, because everyone's are.

Getting the second film made

The strongest predictor of somebody making a second film is having finished the first one. This sounds like a joke and it is not. Enormous numbers of projects are shot and never completed, because finishing is unglamorous and slow and happens after the excitement has gone. A finished five-minute film is worth more than a spectacular forty percent of a feature, in every way that matters, including to funders.

Three things carry forward from a completed film. The work itself, which is now evidence. The team, which is the real asset: the people who shot it with you know how you behave at hour eleven, and a crew that would work with you again is the thing that makes the next production possible. And the specific list of what went wrong, which is the only reliable curriculum in this field.

Public money exists and is worth understanding without romanticising. National and regional screen agencies, arts councils, broadcasters' new talent schemes and film institutes run funds for short films and first features. Almost all of them ask for a track record, which is the circularity that first films exist to break. So the sequence for most people is: make something small with your own resources, finish it, use it to apply.

What a course made of text cannot do

This is the last lesson, so here is the limit stated plainly. Everything in these fifteen lessons is a description of practice, and a description of practice is not practice. You can now say what an eighth is, why the boom sits above the frame line, what a split edit does to a dialogue scene, and why a location agreement signed by a caretaker is worthless. None of that will stop you, on your first day, from forgetting to record room tone, framing an actor against a blown window, and losing the light.

The only thing that closes the gap is a shoot. So here is the prescription, and it is deliberately small. In the next month, make a film of three minutes or less. Use the camera you already own. One location you control, two performers, no dialogue you cannot record cleanly, no scene you cannot light with a window and a bounce card. Write it, board it as this course described, shoot it in a day, cut it, mix it with attention to room tone and split edits, grade it with the scopes open, put titles on it, and export a master and a viewing copy. Then, and this is the part people skip, write down the three things that went worst.

Then make a second one, in the next month, designed specifically to fix the first item on that list. Two finished three-minute films and a list of your own errors will teach you more than another course, another book, or another camera. Almost every filmmaker you admire has an early film they would rather you did not see, and they made it anyway, which is why they are one.

Bottom line: Finish small things repeatedly. Finishing is a rarer skill than shooting, and it is the one everything else is built on.

Common misconceptions

• Short films can make money. Licensing fees and prizes exist and do not constitute a market. Plan for the film to earn you something other than cash.
• Submit to as many festivals as possible. Premiere rules mean a badly ordered submission list can disqualify you from the festivals you most wanted.
• Put it online while you submit, for momentum. Most festivals treat online availability as a public screening, and it ends eligibility.
• A rejection is a verdict on the film. Programming is a jigsaw, and the third good film about the same subject loses to arithmetic.
• You need better equipment before you start. The camera you own will make a film that teaches you more than the camera you are saving for.
• Someone will discover a great script. Funders and collaborators respond to finished work. The first film is the one you make without permission.

The short version

• Chazelle's Whiplash short was evidence answering one financing question, not a general demonstration of talent; McDonagh and Blomkamp took different routes.
• Shorts almost never earn. Decide in advance whether the film buys you skill, an audience, a collaborator, or a piece of proof.
• The circuit has tiers, and premiere requirements mean submission order matters more than submission volume.
• Submit early for lower fees, ask for waivers, and send ten to twenty considered entries rather than fifty scattered ones.
• Programmers watch on laptops: a working screener link, a real still, a poster, two synopses and subtitles are part of the film's chances.
• Rejection is mostly about fit and programme balance, not quality.
• Run festivals for twelve to eighteen months before going online, unless you have deliberately chosen an audience-first release.
• Uploading is not distributing: give the film a page, subtitles, and the specific communities connected to it.
• Finishing predicts everything. Make a three-minute film this month, list what went worst, and make a second one that fixes the first item.

Sources

1. Sundance Institute. (n.d.). Sundance Film Festival: submissions and programmes. Sundance Institute. sundance.org
2. FilmFreeway. (n.d.). FilmFreeway: festival submission platform. FilmFreeway. filmfreeway.com
3. Academy of Motion Picture Arts and Sciences. (n.d.). Rules and eligibility for the Academy Awards. AMPAS. oscars.org
4. British Film Institute. (n.d.). Funding and support for filmmakers. BFI. bfi.org.uk
5. Wikipedia contributors. (2024). Clermont-Ferrand International Short Film Festival. Wikipedia. en.wikipedia.org