Kling Camera Control and Motion Control: What's Documented (2026)

TL;DR: Kling camera control is four separate systems, not one. Prose camera direction in the prompt, a Camera Movement panel with six moves and four Master Shots, per-shot camera movement inside Custom Multi-Shot, and Motion Control, which is reference-video motion transfer and does not move the camera at all. Everything below is sourced to kling.ai.

What is Kling camera control?

Kling camera control means guiding the virtual camera through text, a control panel, or a shot list. Kling's own camera control guide puts it plainly: "In AI video, the motion is described through text, reference images, video inputs, or storyboard instructions" (kling.ai/blog/kling-ai-camera-control-video-guide, dated August 13, 2026, read August 27, 2026).

The confusion in every third-party guide comes from treating that as one feature. Kling ships four, each with its own documentation page and its own constraints. Getting them straight is most of the work.

Two of those pages carry a November 2025 date and describe the pre-3.0 interface. Kling has not republished them for VIDEO 3.0. That is a real gap in Kling's documentation, not something I can resolve for you, and it is why the dates are in the table.

Is Kling Motion Control the same as Motion Brush?

No. They are two live features that do different jobs, and the rename story circulating in AI-video newsletters is wrong.

Kling's Motion Control user guide opens with its own definition: "Motion Control enables precise control of a character's movements and facial expressions based on a reference image. When generating video from an image, you can assign motion to one character in the image. The motion can be extracted from an uploaded video or selected directly from the motion library." Kling's machine-readable summary classifies it the same way, as feature_type: video_to_video motion transfer.

Motion Brush is something else entirely. Its page describes designating "an area or element within the image using either 'automatic selection' or 'manual brushing'" and adding "a motion trajectory," and it supports "simultaneous settings for up to six elements and their trajectories," plus a Static Brush that fixes pixels in painted areas. That page was still live and still linked from Kling's quickstart index on August 27, 2026. If region painting is what you came for, our 12 tested Kling motion brush patterns covers that workflow specifically.

Motion Control also has a hard single-subject limit. Kling's edge-case note says the first frame "may contain multiple people, but only one element is supported; the system will select the person with the largest on-screen presence as the element, and if the elements occupy similar portions of the frame, no element will be selected." The feature page repeats it: Motion Control "currently supports one primary subject."

What camera moves does Kling actually name?

Three separate vocabularies, published on three different Kling pages. They do not fully overlap, and none of them is presented as an exhaustive list.

The Camera Movement control panel is the only one that is a UI control rather than prompt text. Kling's page states that it "supports six basic camera movements, including 'horizontal, vertical, zoom, pan, tilt, roll', as well as four Master Shots: 'move left and zoom in, move right and zoom in, move forward and zoom up, move down and zoom out'." You pick a move and set "the extent of the camera movement by adjusting the displacement parameters."

The prompt vocabulary comes from Kling's camera control guide, which lists nine words as "useful camera words": push in, pull back, pan left, pan right, tilt up, tilt down, track forward, orbit slowly, static camera. The same guide gives the rule that matters more than the list: "one main move per shot is usually better."

The framing vocabulary is separate, and Kling flags the separation itself. Its text-to-video prompt guide defines Camera Language as "ultra-wide angle shots, bokeh (background blur), close-ups, telephoto shots, low-angle shots, high-angle shots, aerial views, and depth of field," then adds a parenthetical: "(Note: This should be differentiated from camera motion control.)" Framing is where the camera is. Motion control is where it goes.

How does Multi-Shot change camera control?

Multi-Shot is the biggest change VIDEO 3.0 brought to camera work, and it has a gate most people miss. Kling's VIDEO 3.0 user guide states that multi-shot generation triggers "through two modes: 'Multi-Shot' and 'Custom Multi-Shot'," that when Multi-Shot is enabled "the model automatically plans the shot transitions," and that "this switch is a prerequisite for enabling 'Custom Multi-Shot'." Leave the toggle off and you get a single shot regardless of what your prompt says.

With the toggle on and Custom Multi-Shot open, Kling's Omni guide says you get "precise control at the shot level, specifying the duration, framing, angle, narrative content, camera movement, and more." Automatic mode plans the coverage for you; Custom mode follows your shot list.

That table is Kling's, from its Multi-Shot blog post dated July 28, 2026, read August 27, 2026. The six-shot cap is corroborated in Kling's own llms.txt, which lists multi_shot_storyboard: yes (up to 6 camera cuts). It does not appear in the VIDEO 3.0 user guide, so treat it as published-but-not-in-the-primary-spec.

One more honest note on syntax. Kling demonstrates Custom Multi-Shot in at least three different notations across its guide pages. Its API reference does publish a formal grammar, on the endpoint pages rather than in the guides: shot n, m, words; shot n, m, words; where n is the shot sequence number (1 to 6 shots), each shot runs at least one second, the shot durations must sum to the total video duration, and the words for each shot cap at 512 characters. The guides and the API reference simply do not cross-reference each other. The VIDEO 3.0 guide's examples use Shot 1, <description>. Shot 2, <description>. The Omni guide uses Shot 1 (3s): <description> with an explicit duration, and one example uses timecodes in the form [00:00 - 00:02] Medium shot:. The Multi-Shot blog uses [Shot 1: Wide shot] <description>. All four appear in Kling-authored examples. Pick one and stay consistent inside a prompt.

How do motion and camera control interact with duration?

Three different duration rules apply depending on which system you are using, and they are not the same number.

The model window is 3 to 15 seconds. Kling's VIDEO 3.0 guide says the model "generates up to 15 seconds of continuous video, with a flexible duration ranging from 3 to 15 seconds." That is a property of VIDEO 3.0 and 3.0 Omni, not of Multi-Shot. The Omni guide adds the before-and-after: VIDEO O1 allowed "between 3-10 seconds," and 3.0 raised the ceiling to 15.

Multi-Shot spends that window, it does not extend it. Six shots inside 15 seconds averages 2.5 seconds each. Kling's own Omni examples are correspondingly tight, running five shots of one to four seconds. A shot list that reads well on paper will get compressed if the arithmetic does not close.

Motion Control takes its duration from the reference clip. Kling's guide states the supported upload range is "3–30 seconds, and the generated video duration will match the length of the uploaded video," and warns that if the action is fast or complex the output "may be shorter than the original upload" because the model extracts only continuous valid motion, with a 3-second minimum. It then adds the line worth reading twice: "the consumed Credits are non-refundable."

Resolution is a fourth axis and it is genuinely independent. Kling launched native 4K for the VIDEO 3.0 series on April 23, 2026, and the announcement describes "true 4K videos in a single click." An integration partner's evaluation quoted in the same post puts it at "native 3840×2160 resolution content during the generation stage itself." Nothing on any Kling page ties 4K to Multi-Shot or to Motion Control. It is a mode you select, not a camera feature.

What do motion and camera control cost in credits?

Kling publishes per-second credit rates in two separate tables that use different axes, which is why cross-tool comparisons in third-party posts are usually wrong.

The first five rows are from the VIDEO 3.0 user guide, the four Motion Control rows from the Motion Control guide, and the 4K row from Kling's credit-cost post, all read August 27, 2026. Kling's credit post tells you to "confirm the live generation panel before production because product rules and access can change," which is good advice I am repeating rather than second-guessing.

Now the arithmetic, clearly labelled as mine rather than Kling's. Motion Control on 3.0 runs 12 credits per second in Professional and 9 in Standard. Plain VIDEO 3.0 without native audio runs 8 at 1080p and 6 at 720p. That is a 1.5x ratio on both rows, so a 50% premium is the right mental model for budgeting Motion Control. Kling does not state the ratio anywhere, and its Standard and Professional modes are documented as a speed-versus-detail tradeoff rather than a resolution tier, so the mapping is an inference from two tables that happen to line up.

Two structural traps in that table. Motion Control is the only camera or motion feature with its own line item, so Multi-Shot and prompt-level camera direction cost exactly what the underlying generation costs, and 4K at 30 credits per second is five times the 720p no-audio rate, which makes iterating on a camera move at 4K an expensive way to discover your shot list does not work.

What Kling does not publish

One spec matters here and is simply absent.

Also unpublished, as far as I could find: any per-move parameter range for the Camera Movement displacement sliders, any statement of whether the Camera Movement panel applies to VIDEO 3.0, since its page is dated November 24, 2025 and Kling has not republished it since the 3.0 launch, and any formal grammar for the Custom Multi-Shot shot syntax. Where Kling is silent, this post says so instead of guessing.

20 copy-paste Kling camera and motion prompts

These are written in the notations Kling demonstrates on its own pages, and the wording is ours, not Kling's. Replace the bracketed slots. Keep to one main camera move per shot, which is Kling's published advice and holds up in practice.

Single-move prompts, prose form

Kling's formula order: subject and description, then subject movement, then scene, then camera language, lighting and atmosphere.

1 — PUSH IN
[SUBJECT: a ceramicist in a linen apron], hands resting on a half-thrown bowl,
seated at a wheel in [SCENE: a cluttered daylight studio, shelves of glazed
work behind]. The camera slowly pushes in from a medium-wide to a medium
close-up on her hands. Soft north-facing window light, warm dust in the air,
calm and unhurried. Stable camera, no handheld shake.

2 — PULL BACK TO REVEAL
[SUBJECT: a lone hiker in a red shell jacket] standing still, looking upward.
Start tight on the jacket's shoulder seam, then the camera pulls back steadily
to reveal [SCENE: a glacial valley opening below, cloud shadow moving across
the scree]. Overcast diffuse light, cold blue grade, sense of scale.
One continuous move, no cuts.

3 — PAN LEFT TO RIGHT
Smooth pan from left to right across [SCENE: a workshop bench], moving past
[OBJECT A: a row of chisels], [OBJECT B: a coffee cup with a chipped rim],
and settling on [OBJECT C: an unfinished dovetail joint]. Static subject,
camera moves only. Low warm tungsten light, shallow depth of field,
realistic wood grain detail.

4 — PAN TO FOLLOW ACTION
The camera pans right to follow [SUBJECT: a courier on a fixed-gear bike]
moving through [SCENE: a wet market street at dusk]. Subject and camera
travel the same direction at the same speed. Neon signage reflecting on the
road, natural motion, stable framing, no whip or overshoot.

5 — TILT UP
The camera tilts up from [LOWER DETAIL: worn leather boots on a steel grate]
to [UPPER SUBJECT: the welder's face behind a raised visor], in
[SCENE: a fabrication shop]. Slow, even tilt. Hard directional key light from
a work lamp, deep shadows, sparks settling. Vertical composition throughout.

6 — TILT DOWN
The camera tilts down from [UPPER DETAIL: a pendant lamp with a brass shade]
to [LOWER SUBJECT: a set table with two plates and a folded napkin], in
[SCENE: a small dining room at night]. Gentle continuous tilt. Warm interior
practicals, soft falloff, quiet and domestic.

7 — TRACK FORWARD
Track forward behind [SUBJECT: a marshal in a high-visibility vest] walking
away from camera down [SCENE: an aircraft hangar's centre line]. The camera
holds a constant distance. Wide angle, low camera height, hard overhead
sodium light, long floor reflections. Steady, mechanical movement.

8 — LATERAL TRACKING
Side-on eye-level tracking shot with slight lateral movement, holding
[SUBJECT: a violinist mid-phrase] in profile as she walks through
[SCENE: a colonnade]. Camera moves parallel to the subject at matched speed.
Columns pass through the foreground as natural wipes. Late afternoon sun,
strong rhythmic light and shade.

9 — ORBIT
The camera orbits slowly around [SUBJECT: a matte-black motorcycle on a
turntable-free concrete floor], one smooth 180-degree arc, subject stationary
and centred throughout. [SCENE: an empty industrial bay]. Single overhead
softbox, specular highlights travelling across the tank as the camera moves.

10 — STATIC CAMERA
Static camera, locked off. [SUBJECT: a barista] enters frame left, sets a cup
on [SCENE: the counter], and exits frame right. The camera does not move at
any point. Fixed framing, medium shot, morning window light, ambient cafe
sound. All motion is subject motion.

11 — ROLL
The camera performs a slow roll, rotating clockwise about ten degrees, while
holding [SUBJECT: a chess player mid-decision] in a centred medium close-up.
[SCENE: a tournament hall, other boards defocused behind]. Cold overhead
fluorescents, rising unease. Roll only — no pan, tilt or push.

12 — HANDHELD
Cinematic handheld, frontal medium shot of [SUBJECT: a paramedic checking a
kit bag] in [SCENE: the back of an ambulance, doors open to a night street].
Natural operator sway, no stabilisation, slight reframing as she moves.
Mixed light: cold interior LED, warm streetlight through the doors.

13 — CRANE / RISE
High-angle wide shot with a gentle downward tilt. The camera rises as
[SUBJECT: a combine harvester] cuts deeper into [SCENE: a wheat field],
leaving cut lanes behind it. Single continuous rise, no cuts. Golden hour,
long shadows, dust haze in the low sun.

14 — LOW-ANGLE REAR WIDE
Low-angle rear wide shot, camera near ground level, tracking behind
[SUBJECT: a rider on a gravel bike] as they move forward through
[SCENE: a forest fire road]. Foreground gravel passing close to lens.
Dappled canopy light, heavy contrast, sense of speed from proximity.

15 — FIRST-PERSON POV
First-person POV from [SUBJECT: the driver], with the steering wheel and
instrument panel visible ahead in [SCENE: a delivery van on a coastal road].
Camera fixed to the head position, small natural movements only.
Overcast flat light, rain beading on the windscreen, wipers in shot.

16 — OVER-THE-SHOULDER PUSH
Over-the-shoulder framing from behind [SUBJECT A: an editor at a desk],
looking past her toward [SUBJECT B: a wall of pinned photographs].
The camera pushes in slowly past her shoulder until the wall fills the frame.
[SCENE: a newsroom after hours]. Single desk lamp, deep surrounding shadow.

17 — MOVE-AND-FREEZE FOLLOW
One continuous take, no cuts. The camera follows [SUBJECT: a nurse] in a
medium shot, moving in sync with her: it tracks as she walks and freezes
instantly when she pauses, then resumes when she does. [SCENE: a hospital
corridor at shift change]. Fluid, natural camera work, no lag on the stops.

Multi-shot prompts, Kling's shot-list notations

18 — CUSTOM MULTI-SHOT, DURATION-TAGGED (Omni notation)
Shot 1 (2s): Wide shot. [SUBJECT: a chef] enters [SCENE: a cold prep kitchen
before service]. Static camera, high angle.
Shot 2 (3s): Medium shot. She sets a crate on the pass and pulls a knife from
the block. Slow push-in.
Shot 3 (4s): Macro shot of the blade meeting a shallot, hands only.
Static camera.
Shot 4 (3s): Close-up on her face, checking the line. Gentle tilt up from the
board to her eyes.
Shot 5 (3s): Wide shot, the pass now full. Slow pull back to the doorway.
Consistent cold morning light and stainless-steel grade across all shots.

19 — CUSTOM MULTI-SHOT, BRACKET NOTATION
[Shot 1: Establishing wide] [SUBJECT: a surveyor] on [SCENE: a windswept
ridge], tripod set, static camera.
[Shot 2: Medium shot] She bends to the eyepiece; the camera pans right to
follow her sightline.
[Shot 3: Close-up] Her hand writing a figure in a weatherproof notebook;
camera locked off.
[Shot 4: Low-angle wide] The camera tilts up from the tripod legs to the
ridge line and sky.
High consistency, overcast diffuse light, cold grade, realistic textures.

20 — TIMECODED STORYBOARD (Omni notation)
[00:00 - 00:03] Wide shot: [SUBJECT: a luthier] at [SCENE: a bench under a
skylight], planing a spruce top. Static camera.
[00:03 - 00:07] Close-up: shavings curling off the blade. The camera pushes
in slowly. Audio: the dry rasp of the plane, room tone low.
[00:07 - 00:11] Medium shot: he lifts the top to the light and taps it.
The camera orbits ninety degrees around him.
[00:11 - 00:15] Close-up on face: he nods once. Camera fixed.
Warm north light throughout, fine dust in the beam, no cuts in lighting.

Getting the reference material right for Motion Control

Kling publishes unusually specific upload guidance here, and it is the difference between a clean result and a wasted, non-refundable generation.

The element library "only uses facial information for reference. It does not include clothing, hairstyle, makeup, or props," so clean facial close-ups are what to upload. For accurate head turns, Kling asks for "a front-facing view, side views (left and/or right)." For expressions, "a neutral front-facing image, a smiling front-facing image." For a full 360-degree rotation while smiling, it wants five: front, left profile, right profile, upward-facing and downward-facing. For emotional transitions combined with head movement, front plus smiling plus sad plus a side view. Where you need complex expressions with high identity accuracy, Kling's guide says "we strongly recommend uploading a video," which "provides richer and more continuous facial information."

The reference clip itself has constraints too, stated under the 2.6 section: the short edge "must be at least 340px, and the long edge must not exceed 3850px," a single continuous shot with the character always visible, and the guide is explicit that you should "avoid cuts, shot changes, or camera movements; otherwise, the video may be truncated." That last clause is the one that catches people who feed in an edited clip.

Where to keep the shot grammar

Twenty prompts in a browser tab is a shot library for about a week. The predictable failure is that you adapt prompt 9 for a client, the adaptation is better, and it dies in that chat window.

A prompt template only compounds when it is stored with variables rather than pasted and hand-edited each time. That is the layer Prompt Architects builds. You save the shot grammar once, mark [SUBJECT], [SCENE] and [LIGHTING] as variables, and fill them per project instead of retyping the structure. Video prompt generation sits on the Advanced and Team plans rather than Pro, and the free plan includes 5 prompt enhancements per day, forever, per our FAQ.

If you are choosing between engines before you invest in a shot library, Veo 3 vs Sora vs Kling covers the tradeoffs, and 30 cinematic camera prompts for Veo 3 and Kling gives you a broader cross-model set. For Kling specifically, the free Kling 3.0 prompt generator builds the subject and scene half so you can spend your attention on the camera half.

One thing to be clear about. We build the prompt layer. We do not generate video, we do not host Kling, and we are not affiliated with Kuaishou. Generating any of the above needs a Kling account and credits. Every spec on this page was read from kling.ai on August 27, 2026, and Kling ships fast enough that the panel is always the final authority.