Find Performance References That Reveal an Action, Not a Type
Find Performance References That Reveal an Action, Not a Type
The request arrives shaped like a personality. Something awkward, please, but not cringe. Confident, but not smug. Natural. And the researcher comes back with three familiar faces and two timestamps that don't quite land in the meeting.
The faces aren't the whole problem. A face is a summary. It arrives carrying a filmography, a press tour, and the way strangers talk about the person online. When someone says "she's got that energy," the room nods and nothing has been decided about the scene.
A reference that earns its place does something narrower. It points at a moment where a person is trying to do something to another person, and it describes that attempt closely enough that a director can disagree with it. Not "she's charming," but: she asks a question she doesn't want answered, and keeps her eyes on him while he answers it, because getting him talking is the way out of the room.
That's a less glamorous deliverable than a name. It's the one that gives a director something to say yes or no to.
Turn the adjective into a question the scene can answer
Awkward is an outcome. It's what the audience ends up feeling while watching. If you hand it to a performer, nothing happens on the day, because there's no decision to make. The useful move is to ask what the person is doing that produces that feeling.
Someone who is awkward might be filling a silence without adding anything true. Or absorbing a remark they've decided not to answer. Or asking permission in the wrong tone and then talking over the reply. Those are different scenes, and they will read differently on camera.
Charismatic usually means the person is doing something to hold the room. Keeping someone else talking. Spending attention on the one person nobody else is looking at. Withholding a beat so the others lean in.
Natural is the least useful of the three, because it's usually a note about the writing and the camera rather than the performer. The playable version is closer to: this person appears to be deciding things in real time, and sometimes decides late.
Here's a working test. Could two performers do this action differently and both be right? If yes, you have an action. If there's only one way to do it and that way is a look, you have an image.
Watch the exchange around the moment
Once you've found something, you usually need more of it than you thought. Record what happens immediately before the behavior, what the other person does during it, and whether the first person changes tactics. Most of what we call a performance is a response to something the scene partner just did, and inventing the response is how actors work.
This is where excerpted moments betray you. A face that reads as contempt in a still frame may be someone holding in a laugh. A flash of alarm may be a reaction to a sound you cut out. If you show the isolated expression in a meeting, you're showing your memory of the scene rather than the scene, and anyone who hasn't seen it won't get it.
A simple three-column method holds up:
- What I can see and hear: who does what, in order, including where the other person is and what they do back.
- What I'm reading into it: the motive, the mood, the relationship. Mark these as readings, not observations.
- What the shot is doing: cutting, holding, framing, sound. This column saves you later.
Say what the camera and the edit did
Before the note goes out, separate the performance from its construction. Shot size makes small movements large and large movements small; the same choice played wide can look like nothing and played close can look like a decision. The length of a hold after a line is often where the feeling is created. Facial expressions can be assembled by an editor out of two takes that never existed together.
Where you can attribute an intention — an actor or director describing what they were doing — that supports a reading of the moment. It doesn't prove that every viewer takes it that way, and it doesn't travel with the clip. If you can't source an account, say the reading is yours.
An invented example: the host and the plate
The scene below is written for this piece. It isn't from a film or a commercial, and nothing in it has been shot, watched, or tested. It's here to make the method checkable.
Two people at a table in a home. The first course is finished. The second isn't ready. The guest is in the middle of a story, mid-sentence. The host's own plate is already cleared; the guest's is the last one on the table, and it's empty.
The host reaches across, lifts the guest's plate, and moves it off the table edge, out of the guest's line of sight, onto a sideboard. While the hand works, the host asks, "And then what did she say?" The guest keeps going.
The weak brief: "Charmingly chaotic host energy."
Nothing in that sentence tells a performer where to put a hand, when to speak, or what to do if the plan fails. You could cast from it. You can't shoot from it.
The action note: the host is buying time, and the specific tactic is to keep the guest talking so the empty plate stops being a clock. What's observable: the host's eyes stay on the guest's face while the hand works; the hand doesn't hesitate; the question lands before the plate clears the table edge, and the guest keeps talking without glancing at the plate; the host's other forearm stays flat on the table, holding a relaxed lean. The upper body performs ease while one hand does the errand.
What's inference: that the host is worried about the food. That the host doesn't want help, because help would mean admitting the delay. Those are readings the action supports. They aren't facts about what the performer was thinking.
When the guest's answer changes, the host's action changes
Now imagine the same beat with a different answer. Invented again: the guest stops mid-sentence and looks at the host's hand, then at the space where the plate was.
The concealment is over. The plate is gone from the table, but the guest now knows the table is being managed. The host cannot simply continue, because continuing as though nothing happened is no longer a concealment — it's a bluff, and a bluff is a different action with a different body. The hand has to keep moving evenly. The eyes have to arrive at the guest's face a beat late. The performer is now doing two things at once, holding the easy lean while knowing they've been seen.
There are at least three honest ways out. Convert: name it, stop pretending, put a real time on the delay. Ask for a task instead of a story — "Would you open the wine?" — which occupies the guest's hands and is a different tactic than a question, because a question gives someone something to notice while a task gives them something to do. Or spend something: tell a story of your own, which costs the host attention they were using to watch the room.
Which one is right depends on the guest. That's the point. The moment isn't a pose with a face attached. It's a loop: the host acts, the guest answers, the host revises. A note that describes only the first move is a note about a photograph.
A pause the performer chose and a pause the cut made
In the invented scene, the host's hand might seem to hang for a beat before the reach. That stillness can come from three different places, and which of them you get depends on what you shot.
The performer chose it. A beat of decision before the hand moves.
The editor made it. Two cameras, two takes, and the cut lands between the last frame of one and the first frame of the other. The stillness is a join.
The viewer made it. The host isn't on screen at all during the moment; we watch the guest's face and infer stillness from theirs. Then it isn't in the footage.
This matters when you write the brief. "Hold a beat before the reach" only exists if someone can cut, or if the take can run long. In a single continuous take, the playable version is tempo inside the action: the hand starts, stops short, continues. Whether that's better is a creative call. Whether it's available is not.
What transfers, and what you leave in the source
What travels is the tactic, the loop, and the relationship between them. The host never looks at what they're handling. The question arrives before the object clears the table. The upper body stays easy while one hand works. Those are playable rules, and a performer can find their own version of each.
What doesn't travel: the source performer's identity, persona, voice, and mannerisms. The source's story — the particular friendship, the particular money trouble, the fact that it's dinner at all. Where your scene's equivalent of the plate is, that's a question for your scene, not a thing to copy.
This is also where the recognizable face stops being an innocent shortcut. "She's warm, like someone who's had a hard year" is not something a performer can do. "She keeps the other person talking" is. Identity can be a fact about who gets cast; it isn't an instruction for how to act, and two people who share it can do opposite things with the same scene. If the real question in the room is casting, say so and go do casting. If the question is what to play, the note has to describe behavior.
One practical limit. Describing a moment in words and showing the clip are different jobs with different permissions. If a real scene is going to be pointed to, or shown, or quoted in a deck, that reuse gets sorted out separately from your judgment about whether it's the right moment. Public availability isn't permission, and a description in your own words carries far less baggage than a file.
The note
Four parts, in this order:
The exchange: who wants what from whom, in this moment. What's observable: the action and the response it's made of, including the other person's part. Why it's useful here: the problem it solves in our scene. What to leave behind: identity, likeness, mannerisms, the source's story, the source's lines — and any beat that belongs to a cut rather than a performer.
Filled in, using the invented example:
The exchange: A host is stalling a guest so the late course doesn't get noticed. What's observable: The host's eyes stay on the guest's face while one hand moves the empty plate off the table edge; the question that keeps the guest talking lands before the plate is gone; the upper body holds a relaxed lean throughout. Why it's useful here: Our scene needs a person managing a problem without naming it. The tactic gives us something to play — redirect with a question, move the evidence, keep the body easy — instead of a mood. What to leave behind: The source performer's identity, likeness, manner, and voice. The source's dinner, its history, and its lines. The pause before the reach, unless our edit owns it. Our equivalent: In our scene the thing being managed is ______, and the thing that would give it away is ______.
A director can argue with that note. Nobody can argue with "charmingly chaotic."
Frequently asked questions
Why is a familiar performer or a persona not a usable performance reference by itself?
A face is a summary that arrives carrying a filmography, a press tour, and the way strangers talk about the person online. It can make the room nod while nothing has been decided about the scene. A reference earns its place by pointing to a moment where a person tries to do something to another person and describing that attempt closely enough that a director can disagree with it.
How do you turn an adjective like awkward, charismatic, or natural into a playable action?
Treat the adjective as an outcome and ask what the person is doing that produces that feeling. Awkward might be filling a silence without adding anything true, absorbing a remark they decided not to answer, or asking permission in the wrong tone and talking over the reply. Charismatic might mean keeping someone talking, spending attention on the person nobody else is looking at, or withholding a beat. Natural is often a note about writing and camera; the playable version is closer to deciding in real time and sometimes deciding late. A working test: could two performers do the action differently and both be right? If there is only one way and that way is a look, it is an image.
Why can an excerpted moment or isolated expression mislead?
A face that reads as contempt in a still frame may be someone holding in a laugh, and a flash of alarm may be a reaction to a sound you cut out. Showing the isolated expression shows your memory of the scene rather than the scene, and anyone who has not seen it will not get it. Record what happens immediately before the behavior, what the other person does during it, and whether the first person changes tactics.
What should be separated from the performance itself before judging it?
Separate the performance from its construction. Shot size makes small movements large and large movements small; the length of a hold after a line is often where the feeling is created; facial expressions can be assembled by an editor from two takes that never existed together. Where you can attribute an intention to an actor or director, that supports a reading of the moment but does not prove every viewer takes it that way. If you cannot source an account, say the reading is yours.
What four parts should a performance note contain?
The exchange: who wants what from whom in this moment. What is observable: the action and the response it is made of, including the other person's part. Why it is useful here: the problem it solves in our scene. What to leave behind: identity, likeness, mannerisms, the source's story, the source's lines, and any beat that belongs to a cut rather than a performer.