Camera & Shots

What Is a Long Take?

Also called oner, continuous take, sequence shot

A long take is a single uninterrupted shot that runs far longer than the average cut around it, often a minute or more, so a whole scene plays without an edit. Duration is the only requirement: the camera can be locked off on a tripod or travelling through a building.

What duration does

Cutting is a promise that time can be skipped. A long take withdraws that promise, and everything that follows comes from the withdrawal. The audience knows nothing has been removed, so what they are watching is happening at the speed of real life, and that alone makes them lean in.

It also removes the escape hatch. In a cut scene, an uncomfortable moment ends when the editor decides it ends. In an unbroken shot the discomfort continues at its own pace, which is why sustained duration is the standard tool for dread, for arguments that go too far, and for waiting.

The third effect is trust. A continuous take proves that the space is real and connected, that the actor really did that, that no cutaway was hiding a stunt double or a repositioned prop. That proof is exactly why action cinema keeps returning to the technique.

Static or moving

Two traditions, opposite temperaments. The static long take fixes the frame and lets the scene arrange itself inside it: the audience chooses where to look, the composition does the directing, and the effect is observational. The moving version threads the camera through space, and the pleasure is choreographic, closer to dance than to observation.

Neither is more advanced than the other. Locking off is harder to make interesting and easier to execute; moving is easier to make impressive and much harder to keep clean.

What it costs

Every element has to succeed simultaneously. Performance, focus, camera, light changes as the camera crosses the room, crew hiding behind furniture, extras hitting marks in the background. A take that fails 90 seconds in fails entirely, so the schedule buys rehearsal instead of coverage. That is the trade: no coverage means no way to fix pace, performance or a line in the edit.

Hidden joins are the pressure valve. The standard hiding places are a whip past a wall, a foreground object wiping the frame, a pass through a dark doorway, a body crossing the lens, and a digital morph between two matched frames. All of them work for the same reason: for a few frames there is nothing legible on screen, and continuity lives in motion rather than in detail.

Building one from AI clips

Models cap out at a few seconds, so in AI video every long take is a stitched take. The good news is that the stitching techniques from live action transfer almost exactly, and the seams are the whole craft.

  • Chain first and last frames. Generate a clip, export its final frame, and feed that frame in as the first frame of the next. This keeps the room, the wardrobe and the light continuous across the seam far better than repeating a text description.
  • Hand off on low information. Plan the seam to land where the frame is briefly unreadable: a pass behind a pillar, a swing past a blank wall, a body crossing the lens, a dark doorway. A seam in the middle of a clear wide shot will always show.
  • Match velocity across the seam. The most visible artefact is not a change in content but a change in speed. If clip one ends with the camera moving forward at a walking pace, clip two must say the same thing: camera continues moving forward at the same walking pace.
  • Keep the direction constant. A stitched sequence that keeps travelling one way reads as continuous. One that reverses direction at a seam reads as a cut, no matter how well the frames match.
  • Re-grade to one reference. Every generation drifts slightly in colour and contrast, and drift accumulates over five segments. Grade all of them to the first clip before you judge the join.
  • Use extend tools where you have them. A video-extension model that continues from your existing clip usually produces a better seam than a fresh generation from a still, because it inherits motion as well as content.
  • Accept a ceiling on identity. Faces drift across segments. If the shot has to follow one person for twenty seconds, favour a follow angle from behind, keep them at a distance, or plan the seams around moments when the face is turned away.

A practical target is 20 to 30 seconds from four or five segments. Past that, colour, identity and geometry drift faster than you can correct them, and the illusion that sells a long take, that nothing was removed, starts to break in exactly the places you cannot patch.

The prompt for this

A starting point that reliably produces the effect. Adjust the subject and setting; keep the technical clauses.

Continuous long take, camera follows a chef from behind through a narrow busy kitchen at dinner service, steady gimbal at chest height, constant distance, steam and pass-through windows in the near foreground, tungsten practicals, 24mm, no cuts

Try Long Take yourself

Open the generator with a starting point already filled in.

Frequently asked questions

How long does a shot have to run to count as a long take?
There is no fixed threshold, because it is relative to the film around it. In a modern action film with a two-second average cut, eight seconds already feels long. The working definition most crews use is a shot that plays a full scene or sequence without a cut, which usually means a minute or more.
What is a oner?
Crew slang for a scene shot as one take. It carries a slightly different emphasis: a oner is a production achievement, something planned, rehearsed and blocked so the whole scene can be captured in a single pass, often with the camera moving through multiple rooms.
Are famous long takes really one shot?
Often not. Many celebrated examples are stitched from several takes, joined behind a whip pan, a foreground object wiping the lens, a pass through darkness, or a digital morph. The intent is the unbroken feeling, and audiences read continuity from motion, not from file boundaries.
When is a long take the wrong choice?
When the scene needs emphasis the camera cannot give. Cutting is how you direct attention to a hand, a glance, a reaction at the exact frame it matters. Hold everything in one shot and you surrender that control, so a weak performance or a slack passage has nowhere to hide.
How do you make a long take with AI video?
By chaining short clips: generate three to eight seconds, take the final frame, use it as the first frame of the next clip, and hand off during a low-information moment such as a pass behind a pillar. Match the camera velocity in each prompt, and re-grade every segment to one reference so colour does not drift.

Related terms