FlashBoards LearnEditing · Film language
L-Cut
Also called: split edit, audio lag, sound overlap
An L-cut is a split edit in which the sound of one shot keeps playing after the picture has cut to the next. We still hear a line, a song or a room while already watching the listener's reaction or the new scene. It hides the cut, stretches a moment and lets a line land on a face.
Video script
Hey, I wasn't finished! Sir, they cut to your face. I'm still talking. That's my foot.
L-cut: picture changes first, old sound keeps playing. You hear A, you watch B.
Where sound runs past its picture, the timeline makes an L: a tall stroke and a foot. A J-cut is the guest who comes early. L is the guest who won't leave.
The Joker starts a line, the picture cuts to Batman, the voice runs on two more seconds, onto a silent face.
Comedy too: a teacher's monotone over glazed students. The joke's in the faces. Audio lag. Sound overlap. Anyone?
Worst crime: a mouth moving under someone else's line. Your Honor, I am a small pixel man and I love crossfades. Pick a closed mouth.
Cut straight, unlink the sound, roll the picture edit left. Ow. Fill the gap with room tone.
Six to twelve frames hides a cut. One or two seconds hands over a phrase. Two to four leaves a scene. Much more, and this.
The morning after an argument. He came here to be alone. Sir, one more thing about crossfades. Nobody's watching your sea now.
Gentlemen, ears won't help. I'm on the track.
Models can't hear, so direct the face. Lips pressed together, first flicker of hurt. Ma'am, all I said was crossfade.
Last cut lands on you. Lips pressed together, please. Scroll down. I'm still talking.
- What it does
- Keeps the outgoing sound running under the incoming picture, so a line or a room lingers after the cut.
- Use it when
- A line matters more on the listener's face, or a scene should fade out in the ear before it leaves the eye.
- Watch out
- Overlaps longer than the moment needs make scenes drag and blur whose line is whose.
- Try this prompt
Close-up of a silent listener, eyes on someone just off camera left, a line still hanging in the air, 85mm, soft window light
What is an L-cut in film?
A normal cut changes picture and sound on the same frame. An L-cut changes them at different moments: the picture cuts first, and the audio of the outgoing shot carries on underneath the incoming one for a few frames or a few seconds. The audience keeps hearing A while already looking at B.
The name comes from the shape on an editing timeline. Picture sits on the top track, sound below it. When the audio of shot A reaches further to the right than its picture, the outline of the clip looks like the letter L: a tall stroke where picture and sound run together, and a foot where only the sound continues into the next shot.
Editors also call it a split edit, but that is the umbrella term, not a synonym. There are two kinds of split edit, and they point in opposite directions:
- L-cut — the sound trails the picture. The old shot's audio lingers over the new image.
- J-cut — the sound leads the picture. The new shot's audio starts before we see it.
You will also hear an L-cut described as audio lag, a sound overlap or a "lagging" edit. In a dialogue scene it is the most common edit there is: the speaker finishes the sentence while we are already watching the person hearing it. Among the tools of editing, it is the one that makes conversations feel continuous rather than assembled.
Why do editors use L-cuts?
To put the line on the listener. The meaning of a line is often in how it is received. Holding the speaker's voice while cutting to the other face turns the listener's shot into a reaction shot with a cause attached. Interrogations and arguments live on this: the threat is spoken off screen, and the camera watches it hit. The interrogation room in The Dark Knight (2008) is largely built on one man talking and another taking it in.
To keep a conversation from feeling chopped. If every cut in a shot reverse shot scene falls exactly between sentences, the rhythm turns mechanical — speaker, cut, speaker, cut, like a tennis rally. Moving the picture cut a few words earlier, inside the sentence, makes the edit disappear. The opening of The Social Network (2010), a fast two-person argument in a bar, is a well-known study in cutting overlapping talk; the film won the editing Oscar for Kirk Baxter and Angus Wall.
To let a scene end in the ear. At the end of a scene the L-cut lets the last line, a piece of music or the room's ambience trail over the first image of the next location. The old scene fades out after the new one has started, which softens the jump in place and time. When that carry-over deliberately joins two spaces, it is a sound bridge.
To stretch a moment. An L-cut buys screen time without new dialogue. The words are over, but we stay on a face, a hand, an empty room, and the silence after the line becomes part of the scene. Comedy uses the same trick: a teacher's monotone drones on over a row of glazed students, and the joke is in the faces, not the voice.
To rescue a performance. The L-cut frees picture from sound. If an actor's delivery is perfect but the image drifts soft or the eyeline breaks halfway through, keep the audio and cut the picture away to the listener or a cutaway. The audience hears the best take and sees the best picture.
To control pacing. Because the sound holds the scene together, you can cut picture faster or slower than the dialogue without the scene falling apart. That separation is one of the main levers of pacing.
How to edit an L-cut
Every editing application does it the same way at heart: move the picture edit point without moving the sound edit point.
- Cut it straight first. Lay the two shots end to end with sound and picture cutting together. Watch the scene through and decide where the line should land — on the speaker's face or on the listener's.
- Unlink or select picture only. Temporarily break the link between video and audio, or use the modifier that selects only the video side of the edit.
- Roll the picture edit earlier. With a rolling edit on the video track, drag the cut to the left. Shot B's picture now starts sooner while shot A's audio keeps playing to its original out-point. The timeline shows the L.
- Fill the audio underneath. If shot B has sync sound, its track now starts later than its picture. Fill that gap with B's room tone or let A's track cover it, so the listener's shot is not dead silent.
- Smooth the join. Add a short crossfade of 2–4 frames on each audio edit and match the background levels, or the change in ambience will reveal the cut you just hid.
Useful numbers for dialogue: an overlap of 6–12 frames (a quarter to half a second at 24 fps) is enough to move the cut inside a sentence. Overlaps of 1–2 seconds hand a whole phrase to the listener. At a scene transition, letting music or dialogue trail 2–4 seconds into the new scene is common; much more and the audience starts listening to the old scene instead of watching the new one.
Two checks before you move on. First, play the scene without looking at the screen: the dialogue should still sound like one continuous conversation. Second, watch it without sound: the picture cuts should still follow the eyelines and the thought of the scene.
How to shoot for L-cuts
L-cuts are made in the edit, but they depend on what the set gives the editor. Good coverage for split edits looks like this:
- Shoot the listener through the whole scene. The off-camera actor should play every line, full strength, while the camera is on the other face. Reactions recorded to a script supervisor reading flat lines look flat.
- Frame the listener as a real shot. A clean single or an over-the-shoulder at 50–85mm on a full-frame camera, eyeline just off the lens on the side of the speaker, eye height or slightly below. Match the lens and size to the speaker's coverage so the two sides cut together.
- Keep off-camera dialogue clean. Mic the off-camera actor too, and avoid overlapping lines on set unless the scene demands it. Clean lines can be slid across the cut; overlapping ones lock the editor to one take.
- Let takes run. Ask the camera operator to hold for three or four seconds after the last line. Those tails are what an L-cut lands on at the end of a scene.
- Record room tone. Thirty to sixty seconds of the set's silence, with the crew still, fills the gaps that split edits leave in the audio.
- Light for silence. A listener close-up often plays longer than the line it sits under. Give it a catchlight and some shape — soft side light, a darker background — so a still face holds the screen.
L-cut vs. J-cut: what is the difference?
Both are split edits; the difference is which side of the cut the sound belongs to.
| | L-cut | J-cut | |---|---|---| | Sound | outgoing shot's audio lingers over the new picture | incoming shot's audio starts before its picture | | Viewer hears | the past — what was just said or where we just were | the future — who speaks next, where we are going | | Feeling | aftermath, weight, reflection | anticipation, pull, curiosity | | Typical use | reactions, scene endings, a line landing on a face | scene openings, a new speaker, a phone about to ring | | Timeline shape | audio extends right of the picture: L | audio extends left of the picture: J |
In a dialogue scene they alternate constantly: you hear B start to answer before you see B (J-cut), then you stay on B while A reacts (L-cut). In editing a dialogue scene, a mix of both is what makes coverage feel like one unbroken conversation.
The sound bridge overlaps with both. Any L-cut or J-cut that carries audio across a change of scene is a sound bridge; an L-cut inside one conversation, between two faces in the same room, is not. When the lingering sound deliberately contradicts the new image — a wedding song trailing over a funeral — it moves toward asynchronous sound.
Common L-cut mistakes
- The overlap outstays the thought. If the voice keeps talking long after the listener has reacted, the reaction shot goes stale. Cut back or shorten the audio.
- Mouths moving under someone else's line. If the listener starts speaking while the outgoing line is still playing, the image says one person talks and the track says another. Choose a take where the listener's mouth is still.
- Ambience jumps. Two setups rarely share the same background noise. Without room tone and short crossfades, the audience hears the edit even though they cannot see it.
- Every cut split. Scenes where every edit is an L-cut feel soft and unfocused. Save straight cuts for moments that should land hard — a slap, a reveal, a smash cut.
L-Cut examples in film
14 framesA still frame can't play sound, so these frames show where an L-cut usually lands: the listener's face, the room reacting to a voice, the space a scene leaves behind as its sound fades.
How to create a l-cut with AI
4 promptsAn image model can't hear, so with AI you direct the shots an L-cut needs: the listener close-up the line lands on and the quiet frame a scene fades into. Describe who is speaking off camera and what the listener just heard, and keep the face alive but silent. Nano Banana 2 Lite builds the listener's start frame, GPT Image 2.5 a scene-exit wide for a trailing sound, and MiniMax H3 Max Turbo animates the reaction so you have picture to lay the dialogue under. Keep both sides of the conversation on one FlashBoards board and mark where each line should spill over. The fourth prompt does the whole L-cut in one MiniMax H3 generation: a structured JSON prompt writes the cut and splits the Joker's line across it, re-shooting the interrogation from The Dark Knight with the original playing next to the result.
Over-the-shoulder close-up of a listener, 85mm lens, eye level, cinematic film still. Blurred dark foreground on the left: the shoulder and ear of a man in a gray wool coat, speaking. Sharp on the right third: a woman in her forties with short copper hair, a small mole on her right cheekbone and a green corduroy jacket, sitting at a kitchen table, listening. Lips pressed together, eyes fixed on the speaker, the first flicker of hurt in her brow. Soft window light from the right, dim warm kitchen behind her, shallow depth of field, film grain.
Close-up of a listener and lips pressed together keep the frame a reaction, not a second speaker, so a line can play over it. The blurred shoulder speaking in the foreground places the voice in the room, and the first flicker of hurt gives the face somewhere to go while the line finishes.
Open FlashBoardsWide establishing shot at dawn, 35mm lens, eye level, anamorphic widescreen film still. An empty two-lane road along a gray sea, wet asphalt, a single parked red hatchback on the shoulder. A young man with a shaved head and a long navy parka stands beside the car, small in the right third of the frame, looking out at the water, hands in his pockets. Low fog, flat blue light, the sun a pale disc near the horizon. Quiet, still, the morning after an argument.
This is the other half of an L-cut: the first frame of the next scene, where the last scene's sound can trail off. Empty, small in the right third and quiet, still leave space for a lingering voice; the morning after an argument keeps the mood linked to what came before.
Open FlashBoardsCamera: locked-off over-the-shoulder close-up at eye level, a very slow push-in of a few centimeters across the whole shot. Action: she is already listening, eyes on the man in the foreground, her brow tightening as his words land. In the middle she swallows, blinks slowly and looks down at the table. At the end she looks back up at him with wet eyes, lips still pressed together, and breathes out.Already listening and the tightening brow make the shot alive from the first frame. Lips still pressed together keeps her silent, so the speaker's line can run over the whole clip in the edit. The beginning, middle and end beats give you a reaction to cut out of at any of three points.
Open FlashBoards{
"output": "Full-screen live-action scene from The Dark Knight (2008), from its very first frame: two shots joined by one hard cut, an L-cut: the speaker's sentence starts on his own face and his voice carries on over the next shot, the silent listener's close-up. 2008 35mm film look with fine grain, 2.39:1 scope frame, a bare police interrogation room at night: pale grey-green painted walls, one hard top light over the table, deep shadows, low-key; palette of sickly green-grey, bruised purple, clown white, smeared red and matte black.",
"characters": {
"S1": {
"name": "The Joker, played by Heath Ledger",
"look": "man in his late twenties, greasy shoulder-length dirty-blond curls with a green tinge, cracked and smeared white face paint, black-smudged sunken eye sockets, a smeared red painted grin over scars at the corners of his mouth, purple-blue patterned shirt under a dark green waistcoat",
"voice": "quiet, nasal, drawling and sing-song, with wet lip smacks between phrases"
},
"S2": {
"name": "Batman, played by Christian Bale",
"look": "matte black armoured cowl with short pointed ears covering his head and face down to the nose, only his mouth and square jaw bare; black armoured suit, a black cape bunched over his shoulders",
"voice": "silent: he never speaks in this video"
}
},
"shots": [
{
"shot": 1,
"camera": "tight close-up of S1 across the interrogation table from S2's side, seated eye level, long lens, shallow depth of field: S1's face fills the left-centre of the frame, the top of his hair touching the top edge, his purple shoulders and green waistcoat along the bottom edge; behind him the pale grey-green wall, soft, with a dark vertical door edge at right; at the start a dark blurred mass of S2's shoulder sits in the lower-right foreground; near-static",
"action": [
"BEGINNING: alive from the first frame, S1 leans forward over the table, head lowered, face turned three-quarter to screen right, glaring up from under his brows at S2 just right of the lens; lips closed, the red-painted mouth drooping; silent.",
"THEN: the dark blurred shoulder in the lower-right foreground sinks out of frame and S1's face settles a little further left, more of the pale wall opening up on the right; S1 stays silent, lips closed, eyes fixed on S2.",
"THEN: he rolls his head loosely: dips it down toward screen left, eyes sliding sideways and half closing, lips parting with a small wet smack, then swings it back up.",
"THEN: head back, tilted slightly, eyes locked on S2 again, he starts his sentence quietly, lips and tongue moving on every word.",
"END: he is mid-sentence, mouth open on the word freak, eyes on S2, when the picture cuts away; the sentence is not finished yet."
],
"dialogue": [
{
"speaker": "S1",
"on_screen": true,
"when": "after he rolls his head and brings his eyes back to S2, running right up to the cut: he reaches the word freak just as the picture cuts",
"delivery": "quiet, confiding, drawn-out, nasal",
"line": "<d>[English] To them, you're just a freak <scenetrans></d>"
}
]
},
{
"shot": 2,
"cut_at": "00:04.460",
"camera": "reverse angle, over-the-shoulder close-up of S2 from behind S1's head, same seated eye level, long lens: S2's masked face sits left of centre, turned three-quarter to screen right toward S1, the cowl's ears cut by the top edge, his bare mouth and jaw lit from above, the black cape bunched at his shoulders along the bottom; the back of S1's head, a mass of soft out-of-focus greasy curls, fills the right half of the frame in the foreground; pale grey-green wall at the left edge; near-static, a very slow slight drift to screen right",
"action": [
"BEGINNING: alive right after the cut, S2 already stares at S1 just right of the lens, lips pressed together, jaw set; S1 is only a dark blur of curls in the right foreground, and his voice is still talking.",
"THEN: S2 only listens: his mouth stays closed through the whole shot, his lips never part and never move, no nod, head still, while S1's voice finishes the sentence from behind the blurred curls; the same voice carries on over the cut and the listener's mouth stays closed.",
"THEN: S1's blurred curls in the right foreground shift left and catch the overhead light as S1 leans in closer; S2 does not move.",
"END: S2 holds the stare, mouth still shut, while S1's voice starts his next sentence off screen."
],
"dialogue": [
{
"speaker": "S1",
"on_screen": false,
"when": "after a short beat right after the cut: the same voice carries on over the cut; the listener's mouth stays closed",
"delivery": "quiet, pointed, a small smile in the voice",
"line": "<d>[English] <scenetrans> like me.</d>"
},
{
"speaker": "S1",
"on_screen": false,
"when": "after a pause, near the end of the shot, still S1's voice from behind his blurred head; the video ends as these words finish",
"delivery": "low, confiding, a little quicker",
"line": "<d>[English] They need you right now</d>"
}
]
}
],
"sound": {
"ambience": "quiet room tone of a bare interrogation room, a faint electrical hum from the top light",
"speech": "exactly these spoken phrases, each said once, in this order; every word is S1's voice; after the cut S1 is seen only as blurred hair from behind and S2's lips never move",
"never_spoken": "no narration and no voice-over: the descriptions, performance notes and step words of this prompt are never spoken aloud",
"music": "a film score under the scene, always below the voices: one low sustained cello note, tense and unresolved, slowly swelling",
"subtitles": "none"
}
}Remakes the interrogation-room L-cut from The Dark Knight (2008), dir. Christopher Nolan, DP Wally Pfister: the Joker begins "To them, you're just a freak" in close-up, the picture cuts to Batman on freak, and "like me" carries on in the Joker's voice over Batman's closed mouth, with the film's own words and cut point.
The L-cut is the split line: the Joker's sentence closes shot 1 on <scenetrans> and resumes in shot 2 with on_screen: false, so the same voice carries on over the cut. Shot 2's action has Batman only listen, his lips never part, and speech gives every word to the Joker. The film's cut_at puts the picture change on freak, mid-sentence.
Common mistakes
- The listener starts talking. Models open mouths by default. Write lips pressed together or listening in silence, and describe what the face does instead.
- The speaker turns into a second face. For an over-the-shoulder, say blurred shoulder in the foreground and put the speaker on the frame edge, or the model centers both people.
- A posed emotion. "Sad woman" gives a stock face. Name what she just heard and one or two physical beats: brow, swallow, eyes dropping.
- Dead video. If the prompt only says "she listens", H3 slowly pushes in on a still. State the change in the face from the first second.
Keep the reference frames, prompts and every generated take side by side — images and video in one canvas.
FAQ
6 questionsWhat is an L-cut in film?
An L-cut is an edit where the picture cuts to the next shot but the sound of the previous shot keeps playing. We might still hear a character's line while seeing the other person react, or hear a scene's music while the next location appears. The name comes from the L shape the audio makes on an editing timeline.
What is the difference between an L-cut and a J-cut?
They are opposites. In an L-cut the sound of the outgoing shot lingers over the new picture; in a J-cut the sound of the incoming shot starts before its picture appears. The L-cut looks back — it lets something just said sink in. The J-cut looks forward — it pulls you into what comes next. Both are split edits.
Why is it called an L-cut?
On an editing timeline, picture sits on a track above the sound. When a clip's audio extends past the end of its picture into the next shot, the outline of the clip forms a letter L: the tall part where image and audio run together, the foot where only the audio continues. A J-cut makes the mirror shape.
Is an L-cut the same as a split edit?
An L-cut is one kind of split edit. Split edit is the umbrella term for any cut where picture and sound change at different moments. If the sound trails the picture, it's an L-cut; if the sound leads the picture, it's a J-cut. Editors sometimes also call the L-cut audio lag or a sound overlap.
How long should an L-cut overlap be?
It depends on the job. Inside a dialogue scene, 6–12 frames is often enough to move the picture cut into a sentence and hide it. Handing a whole phrase to the listener takes one or two seconds. At the end of a scene, letting sound trail two to four seconds into the next is common. Longer overlaps draw attention to themselves.
When should you use an L-cut?
Use an L-cut when a line means more on the listener's face than on the speaker's, when a conversation feels chopped by cuts between sentences, when a scene should fade out in the ear after the picture has moved on, or when you need to keep the best audio take while cutting away from a weaker image.
Related terms
6- J-CutEditing
In a J-cut the sound of the next shot starts before its picture, so we hear the new scene or speaker a moment before we see it.
- Sound BridgeSound & Music
A sound bridge carries sound across a cut — the next scene's audio starts early or the last scene's lingers — tying two shots together and pulling the viewer forward.
- Editing a Dialogue SceneEditing
Editing a dialogue scene means choosing when to show the speaker, the listener or both, and where to cut on lines, pauses and reactions so the conversation keeps its rhythm and subtext.
- Reaction ShotCamera Shots
A reaction shot cuts away from an action or line to show how another character responds — usually a close-up of their face — so the emotional meaning of the moment lands through the listener.
- Asynchronous SoundSound & Music
Asynchronous sound doesn't match what we see at that moment — a sound from another place or time, or deliberately out of step with the image — used for contrast, irony or subjectivity; accidental drift out of sync is simply an error.
- CutEditing
A cut is the instant change from one shot to the next with no transition effect — the basic unit of editing, from invisible continuity cuts to jarring jump and smash cuts.
Further reading
Walter Murch, In the Blink of an Eye (2nd ed., 2001). Karel Reisz and Gavin Millar, The Technique of Film Editing (1968).