Choose Visual References That Clarify the Commercial’s Hardest Decision

Somewhere in the middle of a treatment is the thing the film actually has to solve: how a product shares space with the person using it, how a room reads as lived in rather than dressed, how a gesture lands when the camera stands close, what a street feels like at the hour the spot takes place. Those are the hard parts, and they are usually the reason a treatment gets rewritten.

They are also the reason a folder of beautiful images can fail. A reference does not earn its place by being impressive. It earns its place by making one of those decisions clearer than the prose alone can.

That is the whole method. The rest is about the conditions that make it work: naming what each image is for, comparing instead of collecting, and knowing when to stop.

Name the question before you collect its answer

Start with the film, not the file. Before you open a search tab, write down the decision the reference is supposed to help you make.

“Warm” is not a decision. “The lamp’s light should make the table feel like the only part of the room that exists” is closer, because it names a relationship between the light and the space. “The user and the lamp should read as occupants of the same small surface, not as a model and a prop” is a decision too.

Atmosphere is allowed to be a reason. A lot of treatments live or die on whether the air of a scene is right. But the atmosphere needs a job. If you want rainy dusk, the question might be how the light wraps the product when there is no sun to help. If you want kitchen-table intimacy, the question might be how close the camera has to stand before objects stop being objects and start being company. Write that question into your own notes, and into the treatment where it fits. It will sort every image you find.

The Gate Films, a UK production company, describes references in a treatment as a way of communicating the proposed film’s look and feel. That is a fair description of the ordinary use, and worth saying plainly, because it is also where the trouble starts. “Look and feel” is broad enough to cover an image that answers a question and an image that only adds polish. The company is describing its own practice, not handing down a rule, and it is not evidence that any particular set of references will get a director and a client to agree. What it supports is the basic idea: references are communication, and communication has a job.

An example with a small lamp

Here is a fictional commercial, used from here on because the decisions are easier to see when the film is small.

The spot is set at a cramped table in a small apartment. A person works late at it. The only light in the scene is a small lamp on the same table, close to their hand and their notebook. The idea is that the lamp and the user share the surface. Nothing on that table has room to be separate from anything else on it.

Call that the proposed film. Now suppose the working folder holds five references. They are described below, not shown, and they are invented for this example. Nothing here claims anything about a real image, its availability or its rights.

• Reference A shows a person and a small lamp at a table, framed so the lamp is close to the hand and the notebook. The shared scale is obvious: everything is crowded, and the crowding is the point.
• Reference B shows a small lamp alone, flanked by empty space, lit like a product. The isolation is the point.
• Reference C shows a lamp shade made of worn, slightly translucent material, glowing unevenly where it has thinned.
• Reference D is another version of A. Similar table, similar closeness, similar scale. Not identical, but answering the question A already answered.
• Reference E is warm and inviting, and it gets that warmth from a large room with a high window, linen curtains and a lot of open floor. The lamp in it is lovely. The room is the opposite of cramped.

Read against the proposed film, these five stop being a mood and become a short argument. That is what you want.

Give each candidate a bounded role

The first job with any image is to say what it demonstrates and what it does not propose. A reference can be evidence that a relationship can be depicted or realized some way. It is not a contract for every visible feature.

Reference A demonstrates a spatial relationship: this is roughly how close the lamp and the hand can sit and still compose. What it does not propose is the exact camera, the exact table, the exact person or the color. It answers one question — do the lamp and the user read as sharing the surface? — and it answers it clearly.

Reference B demonstrates the same question with the opposite answer. It shows the product alone, elevated, separate. That is useful precisely because it is a contrast, not because it is the direction.

Reference C demonstrates material: the shade is worn, translucent, unevenly lit. What it does not propose is composition. The photograph that shows the shade so well might be a close-up that never appears in the proposed film. Keep the material lesson; leave the frame behind.

This is where a lot of collections go soft. A striking image implies ten things at once — the casting, the palette, the lens, the location, the era, the budget — and it is easy to treat all of them as approved because the picture is good. They are not approved. They are visible. Those are different statuses, and a treatment that confuses them will promise a film nobody agreed to make.

Compare alternatives instead of collecting agreement

The most useful move in the whole process is to put two plausible references next to the same question and describe how choosing one changes the commercial.

Take A and B. Both are about how the product occupies the frame. Choose A and the spot commits to the lamp as part of a crowded life — the hand is in shot, the notebook is in shot, the scale is intimate, and the camera has to work in a tight space. Choose B and the spot commits to the lamp as a subject in its own right, isolated and elevated, which means the table and the human would have to give it room the proposed film does not have. In this commercial, the crowded version is the idea. A survives; B becomes the road not taken.

Notice what the comparison did. It made the choice visible. A set that contained both A and B without comment would suggest the film could do both, or had not decided. A set that contains A with a caption explaining the crowding is making a choice on the page.

Now look at D. D repeats A. It is not harmful on its own, but it contributes nothing A did not already contribute, and it has a cost: it makes the folder look more settled than it is. Five images that all agree feel like confidence. Sometimes they are. Often they are the same decision photographed five ways, and the treatment’s real remaining questions have quietly gone unanswered because nothing in the folder was busy disagreeing.

Remove D. If the folder feels emptier afterward, that is closer to the truth of what the set actually knows.

Separate a useful cue from a misleading promise

Reference E is the interesting one, because it is genuinely appealing and genuinely wrong in its dominant signals. The warmth is real and it is the reason the image is in the folder. But the room carrying that warmth is large, open and soft. The proposed film is cramped. If E goes into the treatment unqualified, it promises a spacious apartment the spot cannot afford or wants to avoid. The prose and the picture are arguing.

There are two honest options. The first is to reframe: if the only thing you need from E is the quality of light on a shade, then a tighter crop — the lit part of the lamp against a dark background — extracts the warmth and drops the room. Reframing is a repair, and it works when the thing you want survives the crop. It does not work when the thing you want is actually the room.

The second option is to replace E with a reference whose warmth comes from a compatible space. Same color temperature, same glow, achieved in a tight room. This is usually the better long-term move, because it asks the question the film actually has — how does a warm lamp read in a cramped space at night? — instead of borrowing warmth from a place the film is not set in.

Annotation is available as a third move, but it is worth being honest about how far it goes. A caption that says “here for the light only, not the room” can save a useful image. It cannot do that for every feature a viewer will read. If the dominant signals of a picture keep fighting the caption, the picture is winning, and you should probably replace it.

Arrange the set around the reader’s decision

Once the set is down to references that each answer something, order and grouping do real work.

StudioBinder’s mood board guidance recommends arranging reference images by relationships and notes that a palette or atmosphere can make a partial reference useful. That is provider workflow advice rather than independent evidence that boards produce alignment, but it points at the right habit. Group images that answer the same question. If two references only make sense one after the other — context first, then the detail that pays it off — sequence them. Otherwise, do not imply an order you do not mean.

This is the part that keeps a set honest about what it is. None of these images is the proposed shot. Each contributes one property — a spatial relationship, a material, a quality of light — and the treatment’s job is to say which property came from where rather than presenting one borrowed image as the film. A treatment that arranges six unrelated pictures in a tidy grid is quietly implying a shot list. It is not one. A treatment that puts the spatial reference, then the material reference, then the light reference, each with a caption naming its job, is telling the reader how to use the page. It is also showing which questions this film has decided and which ones are still open.

Captions should do the bounded-role work in miniature, and provenance matters too. If the reader cannot tell where an image comes from, they cannot tell what it is evidence of — a frame from a finished film, a still photograph, an artwork, a test. Those carry different weights. Naming the source is part of using the reference honestly.

One limit is worth stating on the page rather than discovering later: a still cannot demonstrate a time-based mechanism. If the question is how the lamp’s glow changes across the shot as the room darkens, or how the hand reaches for the switch and what the light does in the second after, no collection of stills answers it. Say so. Note that the moving relationship needs a moving reference, and let the still set do the job it can.

Stop when another image will not change the choice

There is no correct number of references, and looking for one is a way to avoid deciding.

The better stopping point is this: the set now makes the chosen direction understandable, and it leaves the unresolved questions visible. In the lamp example, the set might end at three references — the spatial relationship from A, the material limit from C, and a resolved version of E. The direction is legible. What is not answered is written down: how the light changes over time, how the user’s hand actually meets the lamp, whether the cramped table photographs the way the reference suggests once real furniture is involved.

Those open questions are not failures of the set. Some of them are not answerable with stills at all. If an image you need does not seem to exist, that may justify a narrower search through a different pool. If the missing thing is a new interaction nobody has photographed — this lamp, this hand, this table — then no amount of searching will supply it, and the honest next step is a small, authorized experiment rather than another afternoon in the folder.

That decision is worth its own conversation, and it is not something every treatment has to drag behind it. Most sets are done when they have made the choice clear. A few are done when they have also named a question that only a test can settle.

The set you keep should do one thing well: make the film’s hardest decision legible, and stay honest about what it has not solved. A folder that looks impressive and settles nothing has done the opposite — and everyone reading the treatment can usually tell.

Frequently asked questions

What makes a visual reference earn its place in a treatment?

It clarifies a specific decision the film must solve, such as how a product shares space with its user, how a room reads as lived in, how a gesture lands close, or what a street feels like at the hour of the spot. It earns its place by making that decision clearer than prose alone can, not by being impressive. Atmosphere is allowed, but it needs a job.

How does the lamp example show comparing alternatives instead of collecting agreement?

References A and B both address how the product occupies the frame. Choosing A commits to the lamp as part of a crowded life, with the hand and notebook in shot, intimate scale, and a camera working in a tight space. Choosing B commits to the lamp as an isolated, elevated subject that would need room the proposed film does not have. Comparing them makes the choice visible; a set containing both without comment suggests the film could do both or has not decided. Reference D repeats A and should be removed.

Why is Reference E risky, and what are the honest options?

E is warm and appealing, but its warmth comes from a large room, high window, linen curtains, and open floor, so unqualified it promises a spacious apartment the film cannot afford or wants to avoid. One option is to reframe by cropping to the lit part of the lamp against a dark background, dropping the room, if the quality you want survives the crop. A better long-term option is usually to replace E with a reference whose warmth comes from a compatible tight space. A caption saying 'here for the light only, not the room' can save a useful image, but it cannot overrule every dominant signal; if the picture keeps fighting the caption, replace it.

What is a bounded role for an image, and how should the set be arranged?

Say what the image demonstrates and what it does not propose. A demonstrates a spatial relationship but not the exact camera, table, person, or color; B demonstrates the opposite answer as contrast; C demonstrates worn, translucent material but not composition. A striking image may imply casting, palette, lens, location, era, or budget, but those are visible, not approved. Arrange images by relationships: group those answering the same question, sequence only when context then payoff is intended, and otherwise do not imply an order you do not mean. Captions should name each image's job, and provenance matters. A still cannot demonstrate a time-based mechanism.

When should you stop collecting references, and what limitation remains?

There is no correct number, and looking for one is a way to avoid deciding. Stop when the set makes the chosen direction understandable and leaves unresolved questions visible. In the lamp example, that might be three references: the spatial relationship from A, the material limit from C, and a resolved version of E. Open questions might include how the light changes over time, how the user's hand meets the lamp, and whether the cramped table photographs as the reference suggests once real furniture is involved. Some questions are not answerable with stills; if the missing thing is a new interaction nobody has photographed, searching will not supply it, and the honest next step may be a small authorized experiment. Provider guidance on mood boards is workflow advice, not independent evidence that boards produce alignment.

Further reading

• Find the Directorial Idea That Holds a Commercial Treatment Together — Read this before sorting your folder: it shows how a governing directorial idea decides which attractive references promise the wrong film and must be refused.
• No Reference Shows the Film You Mean. How Do You Explain It? — When your set shows a range but no sequence, this names the relation the references fail to demonstrate and chooses a written bridge, drawing, or bounded test.
• Design Reference Pages That Preserve the Relationship You Are Citing — Once the set is chosen, this shows how to lay each reference on the page without cropping away the relationship it was selected to prove.