Two people sit across a table. One talks. Cut to the other listening. Cut back. Cut again. This is shot-reverse-shot, and it handles about 90% of dialogue scenes in most films. It works. Nobody's confused about who's speaking. The spatial relationship is clear.
But watch any scene you actually remember — the restaurant in Heat, the interrogation in The Dark Knight, the dinner table in Parasite — and you'll notice the editing is doing something more complicated than just bouncing between two faces. The cuts themselves carry information. They create rhythm. They withhold, delay, or stack reactions in ways that make the scene feel alive instead of merely functional.
There's a vocabulary for this, and most filmmakers never learn it because it isn't taught as a system.
The taxonomy
The taxonomy we're using here comes from the Shanyin Director Master, an open-source directing methodology by AI filmmaker Shanyin. It catalogs 13 distinct action-reaction structures that go well beyond standard shot-reverse-shot, and provides a ratio guideline for when to deploy each.
We won't pretend these 13 types are the only way to slice it. Different editors and scholars would draw the boundaries differently. But as a working vocabulary for storyboard design, this is the most practical taxonomy we've found.
Type 1: Standard action-reaction
A acts. Cut. B reacts. The building block of screen language.
This accounts for roughly 30% of all shots in a well-edited film. It's the default. The question isn't whether to use it — you will, constantly — but whether to use it exclusively. Most student films and first-time storyboards never leave this mode. Every conversation is two faces trading screen time.
When it works best: straightforward dialogue exchanges, cause-and-effect sequences, any moment where clarity is the priority.
When it gets lazy: when every scene in the film uses exactly this structure and nothing else.
Type 2: Reaction-first
Show the reaction before the action. The audience sees someone's face change — shock, horror, delight — and then the cut reveals what caused it.
Se7en's box scene is the textbook example. We see Brad Pitt's expression crumble before we understand why. The dread comes from reading his face and imagining what could cause that reaction. When the reveal comes, it lands harder because we've already been primed by the emotion.
Use it for: mystery, dread, any moment where the emotional weight matters more than the information.
Type 3: Action stacking
Multiple actions pile up before a single reaction. A insults B. Then C piles on. Then D adds the final blow. Only then does B react.
Whiplash lives on this structure. Fletcher's insults accumulate — "not quite my tempo" repeated with increasing venom — until Andrew finally snaps. The stacking creates pressure. Each action without a reaction tightens the spring.
Use it for: pressure accumulation, montage sequences, building toward an emotional or physical breaking point.
Type 4: Reaction omission
The action happens. The moment demands a reaction. And the editor cuts away before we see it.
The coin toss in No Country for Old Men. Anton Chigurh flips the coin. The gas station owner calls it. We never see his face when the result is revealed. The Coens cut to the exterior. The wind. The parking lot. The audience's imagination fills the gap with something worse than any reaction shot could provide.
Use it for: moments where the audience's imagination is more powerful than showing the reaction explicitly. Also works when you want to respect a character's privacy in a moment of grief.
Type 5: Hidden action
We hear the action but don't see it. The camera stays on someone's face while the sound tells us what's happening off-screen.
Horror films use this constantly. The sound of bones breaking while the camera holds on the victim's friend. A gunshot off-screen. A door creaking open behind a character who doesn't know anyone else is in the house.
It's also budget-friendly. If you can't afford to show something, you can sometimes make it more effective by not showing it.
Use it for: horror, tension, violence that's more impactful heard than seen.
Type 6: Delayed reaction
The action happens. The character behaves normally. Life continues. And then, minutes or scenes later, the reaction arrives.
Manchester by the Sea. Lee Chandler learns his brother has died. He walks out of the hospital. He goes through the bureaucratic motions. He drives. He does ordinary things. And then, much later, the grief surfaces. The delay is what makes it feel real. Humans don't always react in the moment. Sometimes the body keeps going while the mind is still processing.
Use it for: emotional realism, grief, trauma, any situation where immediate reaction would feel melodramatic.
Type 7: Same-frame
Action and reaction happen in the same continuous shot. No cuts. The camera captures both the cause and the response without breaking.
1917 is the extreme version — the entire film is designed to feel like one unbroken shot, so every action-reaction pair happens within the same frame. Birdman does the same thing. But you don't need to commit to a whole film. A single long take during a dinner argument, where the camera drifts between speakers without cutting, creates a theater-like tension. There's nowhere to hide. The audience watches everything in real time.
Use it for: real-time tension, immersion, moments where the continuity of time matters more than editorial emphasis.
Type 8: Reaction chain
One action triggers a cascade of reactions. Person A does something. B reacts. C reacts to B's reaction. D reacts to C. The chain grows.
The climax of Parasite. One discovery in the basement triggers a chain reaction that touches every character in the house, each reacting not just to the original event but to the reactions of others. The scene escalates not because the initial action is so dramatic, but because the reactions multiply.
Use it for: escalation, domino effects, ensemble scenes, any moment where the social dynamics matter as much as the triggering event.
Type 9: Repeated mutation
The same action repeats, but each iteration changes slightly. The repetition itself becomes the meaning.
Requiem for a Dream's "hip-hop montage" — the pupil dilating, the arm injecting, the money exchanging — repeats throughout the film. Each time, the sequence gets shorter, faster, more degraded. The repetition with mutation shows the characters' spiral into addiction without a single line of dialogue about it.
Use it for: addiction, obsession, ritual, any process that degrades or intensifies over time. The audience registers the change between iterations subconsciously.
Type 10: Counterpoint
The visual track says one thing. The audio track says the opposite. Or a character's words contradict what we're seeing.
American Beauty. The family portrait is perfect. Everyone smiles. The voiceover tells us the marriage is dead. The gap between what we see and what we hear creates irony — and it makes the audience an active participant, because they have to reconcile the contradiction themselves.
Kubrick loved this. Cheerful music over violent imagery in A Clockwork Orange. A soldier singing "Mickey Mouse Club" while marching in Full Metal Jacket. The dissonance is the point.
Use it for: irony, unreliable narrators, internal contradiction, dark comedy, any moment where surface and truth diverge.
Type 11: Omitted action
The action never appears on screen. We only see people reacting to something we can't see.
The Blair Witch Project. The characters hear sounds in the forest. They react with terror. We never see what's making the noise. The entire film's horror depends on showing reactions to an action that's permanently off-screen.
Use it for: horror, mystery, unseen threats, cosmic dread. Also useful when the "action" is abstract — grief, loss, the passage of time — and can't be shown directly.
Type 12: Linked chain
Each shot serves double duty: it's the reaction to the previous shot and the action that causes the next one. The sequence flows like dominoes.
Baby Driver's chase sequences. Each shot connects to the next through motion, rhythm, or cause-and-effect, creating an unbroken editorial flow that feels like music. Edgar Wright builds entire sequences this way, matching cuts to the beat of the soundtrack.
Use it for: action sequences, music-driven editing, any scene where momentum and rhythm matter more than individual moments.
Type 13: Combined
Most memorable scenes don't use one pure type. They mix two or three.
The interrogation scene in The Dark Knight starts with standard action-reaction (Batman and Joker trading dialogue). Then it shifts to counterpoint (the Joker laughing while being beaten). Then delayed reaction (Batman's slow realization about the address swap). The scene's power comes from the transitions between structures, not from any single one.
Don't aim for purity. Aim for the right tool at the right moment.
The 70/30 ratio
After cataloging these 13 types, the obvious question: how often should you use the non-standard ones?
The answer from the Shanyin framework is roughly 70/30. Use conventional structures (standard, action stacking, linked chain) for about 70% of your shots. These keep the narrative clear and the audience oriented. Reserve the creative types (reaction-first, omission, counterpoint, delayed) for 30%, deployed at narrative peaks.
If you make every shot creative, the audience has no baseline. The unusual becomes the norm, and nothing stands out. But if 70% of your shots are clean, functional, unremarkable coverage, the 30% that does something different will land with real force.
The pro move is being willing to be boring most of the time.
Quick reference
| # | Type | Best for | Usage |
|---|---|---|---|
| 1 | Standard | Dialogue, clarity | ~30% |
| 2 | Reaction-first | Mystery, dread | 5-8% |
| 3 | Action stacking | Pressure, montage | 8-10% |
| 4 | Reaction omission | Imagination, respect | 3-5% |
| 5 | Hidden action | Horror, tension | 3-5% |
| 6 | Delayed reaction | Grief, realism | 2-5% |
| 7 | Same-frame | Immersion, theater | 5-10% |
| 8 | Reaction chain | Escalation, ensemble | 3-5% |
| 9 | Repeated mutation | Obsession, decay | 2-3% |
| 10 | Counterpoint | Irony, contradiction | 3-5% |
| 11 | Omitted action | Horror, mystery | 2-4% |
| 12 | Linked chain | Rhythm, action | 8-10% |
| 13 | Combined | Complex scenes | varies |
The percentages are rough. They'll shift depending on genre, tone, and the specific demands of your story. A horror film will lean heavier on hidden action and omitted action. A music video might be 60% linked chain. The point isn't to hit exact numbers but to have a vocabulary for what you're doing and a sense of proportion for how much of each type to deploy.
The taxonomy in this article comes from the Shanyin Director Master, an open-source directing methodology by AI filmmaker Shanyin. All illustrations generated using Genkee AI.