No Reference Shows the Film You Mean. How Do You Explain It?
No Reference Shows the Film You Mean. How Do You Explain It?
There's a moment in a treatment when you have six good references and no film. Each one earns its place: one has the material, one has the camera, one has the way a body reads against a light. Put them on the page and something goes wrong. The page shows a range; what's in your head is a sequence. The viewer either decides the combination has been proved — it hasn't — or decides the combination is incoherent, which it may not be either.
A seventh reference won't fix this. What fixes it is naming the relationship the six fail to demonstrate, then choosing the cheapest form that can answer the question that relationship leaves open: a written bridge, a drawn study, or a small bounded test. The choice follows the missing information, not the polish of the deck.
Every reference is a noun. The film is a sentence.
A useful reference contributes one thing, and you should be able to say it in a clause. The way light comes through a paper shade. How far the camera stands from a group of people. A body read as a dark edge against a bright field. A material behavior, a camera relationship, a performance distance, a quality of light.
None of that is a film, because a film is mostly verbs. This turns into that. She crosses, and the frame tightens, and the room stops being the subject. A reference set gives you vocabulary. The thing you're missing is usually the sentence.
So the first move is mechanical. Write one line per reference, stating its contribution. Then write the line that isn't there yet: what should happen when those contributions meet. If you find that you can't write it, that's the diagnosis. The gap isn't more references. It's a missing relation — a specific interaction between two things that have only ever appeared separately.
There's a useful test for whether you've actually found the gap. If the sentence you just wrote could be illustrated by one of your existing references, you never had a gap; you had a reference you hadn't finished reading. The gap is only real when the sentence describes something none of the sources shows.
A reference shows a look, not a cause
Here's where careful decks start quietly inventing things. The image in front of you is evidence of an appearance. It is not evidence of what produced it. You can see the edge of a shadow; you cannot see the size of the light that cast it. You can see a warm surface; you cannot see the exposure it was graded at, the material it's made of, or the distance between it and the lens.
Any claim you make about how a reference was made is your inference, not a reading of the file. That matters less as an ethics point than as a practical one: inferred causes get repeated downstream. Someone builds the set to your inferred source, and the inference turns out to be wrong, and the look you wanted needs a different light entirely. Now the problem isn't the reference. It's the sentence you wrote about the reference.
So attribute the look and leave the cause open. "This is the edge quality I want" is defensible. "This was lit with a small source behind the figure" is a guess unless the maker told you. If the cause is what matters, the maker is the source, or the experiment is.
Find the condition both cues want at the same time
References don't conflict because they look different. They conflict when they ask the same condition to hold two different values.
A condition is anything the shot has to decide once: camera distance, the size of the light relative to its subject, exposure, the state of a surface, the duration of a beat, the speed of a move. Take the cues you're trying to combine and ask, for each, what it depends on. Then look for overlap.
If a cue lives in a close view and another lives in a wide one, and both have to be true in the same frame, you have a candidate problem. If one cue depends on a broad glowing surface and another depends on a hard-edged shadow, and both have to be produced by the same lamp at the same moment, you have another. Note that these are arguments, not verdicts — you should be able to point at the shared condition, not just at a difference in mood.
But notice how often the overlap isn't there. Two cues that look worlds apart may be asking for different conditions entirely, which means they were never in conflict. They were two beats. A great many "incompatible" reference sets dissolve the moment you stop placing every image in the same frame and start placing them in the same running order. The wide is at the top of the shot; the detail is at the end. No problem exists.
Which leaves the interesting cases: the ones where two cues genuinely need the same condition — the same moment, the same surface, the same instant of light — at different values. That's where you owe an authored decision, and where the next section lives.
And don't claim incompatibility from surface difference. Two images at opposite ends of a color range may be exactly the contrast the film is about. The mechanism has to be named. If you can't name it, you don't have a conflict; you have a taste preference wearing a conflict's coat.
The lamp
Here is an invented commercial with three invented rationales, to show the whole movement. Nothing here has been shot or tested.
The spot is thirty seconds for a table lamp. The intended film: a workroom at dusk, three people at a long table, one of them stands and crosses in front of the lamp, and the shade keeps glowing after they've passed.
Three references, each doing real work.
The first is a close still of a translucent paper shade lit from within — close enough that the fiber structure is visible and light appears to bleed through the material. Its contribution is a material behavior: light diffusing through a fibrous paper surface.
The second is a wide photograph of a room with several people at a worktable, small figures, deep space. Its contribution is a camera relationship: distance and depth that hold a group in one frame with the room still readable around them.
The third is a photograph of a person crossing a lit panel, framed closely enough that the body is a crisp dark shape against the light. Its contribution is a figure-to-light relation: a body read as a clean dark edge.
Now the missing sentence. No reference shows a body crossing a glowing paper shade while the camera holds the group wide. That's the interaction, and it splits into two questions that are easy to confuse.
The first is compositional. In the invented wide, the shade occupies about a tenth of the frame's width — a small bright shape in a large room. A person crossing it changes a strip inside that tenth, for about as long as it takes to take a step. The eye will register motion. It will not register a crisp silhouette, because at that scale there is no edge to be crisp; there's a small interruption in a small part of the frame. Reference three's clean dark shape doesn't survive the wide. It's a tight-framing phenomenon, and pretending otherwise is how a deck promises a shot the lens can't deliver.
The second question is physical, and it's the one people skip. Reference three shows an edge. It does not say what made the edge. A broad diffusing surface sitting behind a body tends to wrap light around the outline; a smaller, more distant source tends to cut harder. So the same shade that gives you reference one's glow — a large, soft, luminous surface — may be exactly the wrong thing to produce reference three's crispness. That isn't a proven conflict either. Reference three's edge may have come from something else entirely. But the reference set contains no evidence either way, and that's the honest statement.
So write the bridge. It's a director's proposal, not an observation, and it should read like a fragment of shooting script:
Hold the wide while the first person stands. Start the move in as the chair comes back, so the crossing happens mid-travel — the shade is between a surface and a shape, the figure between a presence and an edge. Settle on the shade for the last beat.
Notice what that sentence does. It answers the compositional question by declining to make the crisp silhouette at all. The crossing becomes a passing shape during a move rather than the graphic beat reference three was chosen for. That's a real choice with a visible cost: you lose the punch that reference earned its place with. Naming the cost is part of what makes the bridge credible. A proposal with no downside is usually a proposal that hasn't been thought through.
It also leaves the physical question exactly where it was. Whether the shade can hold reference one's glow and still give the crossing any kind of defined edge at the exposure the wide needs is not settled by better adjectives.
Three ways to explain, and what each one buys
A written bridge is the right answer when you can already specify the relation. It's cheap, it's fast, and it's precise. It has one hard requirement: it must name an action, an order and a duration. What moves first, what changes as a result, how long the change takes, and what the camera is doing while it happens. A bridge that can't answer those four isn't a bridge; it's a mood with a verb bolted on.
If the reader still can't picture the spatial relationship — where things sit relative to each other, what covers what — more adjectives are the wrong next step. That's a drawing question.
A frame study answers placement, proportion and sequence for a chosen moment: a rough plan of the wide, the table, the three people, where the crossing figure lands at each beat of the move, how much of the frame the shade occupies and for how long. It's rough on purpose, and it settles composition without saying a word about brightness, diffusion or edge quality. A labeled study — composition study, placement and proportion only — is honest about its own limits in a way a rendered frame isn't.
That distinction matters more than it sounds. The finish level of an image is read as a claim. A sketch reads as a question. A beautifully rendered frame reads as an answer, and if the question it answers isn't the one you needed settled, you've spent money to create a misunderstanding. If someone else is asking for the study — a client, an EP, a producer — agree first on which question it's for and what it is not, before anyone starts work. Otherwise you'll be answering someone else's question at your own cost. The most expensive way to answer the wrong question is to answer it beautifully.
When the missing relation is physical, neither prose nor drawing will do. That's a bounded experiment: one question, one setup, small in scale. In the lamp example, the question would be something like — with a piece of the intended paper and a lamp of comparable output at a similar distance, at the exposure the wide requires, does the shade still read as a glowing surface, and does a crossing figure hold any definition at all. Bounded is doing real work in that sentence. One question. One afternoon. Results reported with their conditions attached: this paper, this lamp, this distance, this exposure, this frame size.
What a test like that cannot establish is just as important to write down. It can't speak for the final lamp, the final lens, the final set scale or the final grade. A rough success at one scale is a reason to proceed, not evidence about the finished film. Report it that way and no one downstream is surprised.
You don't need the finished image to make these decisions. Which moment the crossing happens in is a decision. Whether the paper glows at all at the relevant exposure is a decision. Whether the whole film is good is not decidable at this stage, and it shouldn't be the bar you're clearing.
Keep the dependency where the reader can see it
The deck should end the way an honest proposal ends: here is the interaction we're proposing, here is the representation we used to explain it, and here is the one condition that remains unproven.
For the lamp, that reads something like — we're proposing that the crossing happens during the move, so the film never attempts the crisp silhouette; we've explained that relation in writing and, if it's commissioned, in a composition study; and what remains unproven is whether the shade's own glow will leave the crossing enough definition at the exposure the wide requires. That last clause is not an admission of failure. It's the actual state of the work, and it's the thing most likely to change the film.
What you gain by saying it out loud is that nobody has to guess. The viewer understands one film — a wide room, a move, a body crossing a glowing shape, the glow continuing after the body is gone — without being asked to believe that three separate pictures prove it. The ingredients stay ingredients. The proposal stays a proposal.
The temptation to let beautiful source material imply that everything has been solved is strong, and it's understandable: the references are persuasive, and persuasion is what you came for. But a set of references will argue on your behalf whether you ask it to or not, and left unauthored it will argue for the wrong thing — that a combination has been demonstrated, when what you actually have is a sentence you wrote, an intention you're offering, and one condition still waiting for an answer.
Frequently asked questions
What do I do when six good references still don't show the film I mean?
Name the relationship the references fail to demonstrate. Write one line per reference stating its contribution—material behavior, camera relationship, figure-to-light relation, quality of light—then write the missing sentence: what should happen when those contributions meet. If the sentence could be illustrated by an existing reference, you had an unfinished reading, not a gap. A real gap describes something none of the sources shows.
How can I tell whether two references actually conflict?
Do not judge by surface difference. Identify the condition each cue depends on—camera distance, size of light relative to subject, exposure, surface state, duration of a beat, speed of a move—and see whether both cues need the same condition to hold two different values at the same moment. Often they ask for different conditions, which means they are two beats, not a conflict. If you cannot name the shared condition, you have a taste preference, not a demonstrated conflict.
When is a written bridge enough, and what must it include?
A written bridge works when you can already specify the relation. It must name an action, an order, and a duration: what moves first, what changes as a result, how long the change takes, and what the camera is doing meanwhile. If it cannot answer those, it is a mood with a verb attached. If the reader still cannot picture the spatial relationship, a drawing is the next step.
What can a composition study settle, and what can't it?
A frame study can settle placement, proportion, and sequence for a chosen moment—where things sit, what covers what, where the crossing figure lands at each beat, and how much of the frame the shade occupies. It does not settle brightness, diffusion, or edge quality. Label it as a composition study only. Finish level is read as a claim: a sketch reads as a question, while a beautifully rendered frame reads as an answer, so answering the wrong question beautifully is expensive.
How should I present what remains unproven?
End with the proposed interaction, the representation used to explain it, and the one condition still unproven. In the lamp example, that means proposing the crossing happens during the move so the film never attempts the crisp silhouette, explaining that in writing and possibly a composition study, and stating that whether the shade's glow leaves the crossing enough definition at the wide's exposure remains unanswered. A bounded test can be one question, one setup, small scale, but it cannot speak for the final lamp, lens, set scale, or grade. A rough success is a reason to proceed, not evidence about the finished film.