A useful Kling 4.0 prompt describes a scene in motion: who appears, what happens, how the camera follows it, and how the sequence should sound. Start with a clear action, then add the visual and timing instructions needed to make that action readable.
This guide combines Kling's official text-to-video and image-to-video tutorials with its newer 4.0 creation guidance. The copyable prompts below are original teaching examples, not official samples or results from our own Kling 4.0 testing.
Availability checked October 3, 2026: Kling's official 4.0 guide describes an October launch with limited early access. Check which features your account exposes before planning a generation around them.
Cover created with imagegen as a concept illustration. It is not a Kling 4.0 output.
The Kling 4.0 prompt formula
Kling's 4.0 creation guide gives this structure:
Subject + Action + Setting + Camera + Lighting + Pacing + Sound
Use the formula to answer concrete production questions. You do not need a separate sentence for every component when the shot is simple.
| Component | Question to answer | Example direction |
|---|---|---|
| Subject | Who or what should the viewer follow? | An amber perfume bottle with a black cap |
| Action | What visibly changes? | A hand lifts the bottle and sets it down |
| Setting | Where does the action happen? | A stone counter beside a window |
| Camera | How do we see the action? | Medium close-up, slow push-in |
| Lighting | What shapes the image? | Warm side light and soft shadows |
| Pacing | In what order do events happen? | Lift, pause, then set down |
| Sound | What should we hear? | Glass touching stone, quiet room ambience |
For a short text-to-video shot, try this original example:
An amber perfume bottle with a black cap stands on a pale stone counter beside a window. A hand enters, lifts the bottle, pauses briefly, then sets it down in the same spot. Medium close-up with a slow push-in. Warm afternoon side light and soft shadows. One continuous shot. Sound: a gentle glass tap and quiet room ambience.Choose the clip duration and output settings in the generator controls. Words such as “4K” or “30 seconds” inside the prompt should not replace those settings.
Text-to-video: establish the scene before directing it
Without an input image, the prompt must supply appearance and surroundings as well as motion. Kling's foundational text-to-video guide describes subject, subject movement and scene, with camera language, lighting and atmosphere as optional additions. Its examples and parameter notes come from earlier models; use it for the writing method rather than current 4.0 specifications.
A practical starting point is one recognizable subject, one main action and one camera move. Add more detail only when it changes what should appear on screen.
Example 1: a cinematic street scene
A woman in a tan raincoat walks along a wet London street at dusk. She stops beneath a shop awning and looks back toward the crossing. Medium shot, camera tracking alongside her and settling when she stops. Warm shop-window light reflects on the pavement. The walking pace is unhurried. Sound: light rain and distant traffic.The coat identifies the subject. Stopping and looking back create a visible beat. The camera instruction follows that beat instead of introducing a competing movement.
Example 2: a product reveal
A matte white wireless earbud case sits closed on a dark desk. The lid opens slowly, revealing two earbuds, then holds open. Tight product close-up with a gentle push-in. Soft studio light outlines the case against the dark background. Keep the case in the same position throughout. Sound: a subtle hinge click; no dialogue.For a specific commercial product, upload a matching reference if that feature is available. A text description is useful for designing a fictional object, but it leaves important shape details open to interpretation.
Image-to-video: describe what moves
An input image already establishes appearance and composition. Kling's official image-to-video guide therefore emphasizes subject + movement, with background movement added when needed. Identify each moving object clearly, especially when several subjects share the image.
Example 3: animate a portrait
The person blinks, turns slightly toward the window, and gives a small relaxed smile. A light breeze moves a few strands of hair. The camera stays fixed. Keep the same clothing and background. Sound: quiet room ambience; no dialogue.Example 4: animate a product photograph
The camera makes a slow, shallow orbit around the bottle. The bottle stays upright and stationary on the plinth. Window reflections move gently across the glass as the viewpoint changes. Preserve the bottle shape and label from the input image. Sound: quiet studio ambience.Avoid spending most of the prompt redescribing every visible detail. Use those words for the change you want. If the photograph shows a seated person, a small turn is a simpler starting exercise than an abrupt sprint through a different location.
Camera prompts: choose a move with a purpose
These are ordinary filmmaking directions you can use in natural language, not special Kling commands. Specify the subject the camera should follow and whether the movement continues or ends.
| Intended effect | Direction to try | Clarification that helps |
|---|---|---|
| Build attention | Slow push-in toward the subject | End on a close-up of the face or product |
| Reveal surroundings | Pull back from the subject | Name the environment being revealed |
| Follow movement | Track alongside the walking subject | Keep a consistent distance and shot size |
| Show product depth | Shallow orbit around the product | State that the product remains stationary |
| Reveal height | Tilt upward from the base | Name the final framing |
| Hold on performance | Fixed camera, medium close-up | Describe the subject's action separately |
“Dynamic cinematic camera” leaves several choices unresolved. “Track alongside the cyclist at a steady distance, keeping the upper body in frame” supplies a shot the reader can visualize.
For a first draft, choose one camera move per shot. Once it works, add a transition or a second shot with a clear boundary.
Dialogue and audio: connect the line to the performer
Kling's VIDEO 3.0 guide explains pairing characters with their dialogue. Its VIDEO 2.6 audio guide also covers voice attributes, emotion, sound effects and music. These offer useful foundations; use the controls and voice-binding workflow actually available in your selected 4.0 mode.
Example 5: a short spoken product introduction
A creator sits at a kitchen table, holding a small travel mug at chest height. Fixed camera, medium close-up, soft daylight. She looks into the lens, smiles, and says in English with a relaxed conversational tone: "This is the mug I take on every early train." She finishes speaking before placing it on the table. Sound: clear foreground speech and quiet kitchen ambience, without background music.Example 6: two-character dialogue
Two friends stand beside a closed cafe on a quiet evening street. Hold a medium two-shot with both faces visible. The woman in the blue jacket looks at the door and says in English, disappointed: "We missed it." After a brief pause, the man in the grey sweater smiles and replies in English: "There's another place around the corner." Only one person speaks at a time. Sound: clear dialogue and faint street ambience.Use a short line that fits the selected duration. Label speakers consistently, write the intended words in quotation marks, and describe the performance. If speech feels rushed, shorten the line or allow more time before adding more voice adjectives.
Plan a 30-second prompt around story beats
Kling's Multiple Keyframes guide describes up to 10 keyframes and native generation up to 30 seconds. Keyframes establish important visual states; the prompt supplies motion and transitions between them. You can use fewer frames when the sequence has fewer meaningful changes.
The following is an original planning example for a fictional travel-mug ad. Match the duration and any keyframe positions in the available interface. Time labels express intended pacing; they are not a guarantee of frame-accurate timing from text alone.
| Time | Story beat | Possible keyframe |
|---|---|---|
| 0–6 seconds | Establish the commuter and kitchen | Character beside the counter |
| 6–13 seconds | Lift the mug and close its lid | Mug held clearly in both hands |
| 13–22 seconds | Put it in a bag and prepare to leave | Same character beside the doorway |
| 22–30 seconds | Finish on the product | Mug on the counter in close-up |
Example 7: a complete timeline prompt
Create a calm morning product sequence in the same sunlit kitchen. Keep the commuter's tan jacket and the navy travel mug consistent across shots.
0–6 seconds: Medium-wide shot. The commuter approaches the counter and reaches for the mug. Slow camera push-in. Soft morning light.
6–13 seconds: Cut to a close-up of the hands. The commuter lifts the mug and closes its lid with one deliberate movement. Pause briefly so the action is readable.
13–22 seconds: Cut to a medium shot beside the doorway. The commuter places the mug into an open bag, picks up the bag, and turns toward the door.
22–30 seconds: Cut to a separate product close-up of the same mug on the kitchen counter. A gentle camera orbit reveals its shape, ending with a steady front view.
Sound: quiet kitchen ambience, a lid click, soft fabric movement and restrained instrumental music. No dialogue. Use clean cuts between the stated shots.The final shot is an intentional product insert rather than a claim that the bag and mug suddenly return to the counter. Naming the cut makes that change easier to understand.
If your available mode lacks Multiple Keyframes, use the same plan as a storyboard for individual clips. Preserve the character and product references between generations, then assemble the clips in an editor.
Omni Reference: give each asset a clear job
For reference-led work, write a short brief before uploading assets: which image defines identity, which defines the product, and which clip guides motion? Mention those roles in plain language and use the actual attachment labels supplied by your interface.
Example 8: reference-led product direction
Use the attached portrait for the creator's appearance and the attached product photo for the navy mug's shape and color. Use the attached motion clip only for its slow camera push-in. Place the creator at a bright kitchen counter. The creator lifts the mug, turns it slightly toward the camera, and sets it down. Keep the mug design and the creator's clothing consistent. Sound: quiet kitchen ambience, no speech.Here, “attached portrait” is a readable placeholder for this teaching example. Replace it with the matching asset label in your workspace. Do not assume a label such as @Image1 is valid everywhere or that mentioning a file attaches it.
Keep reference images, keyframes and written directions in agreement. If the mug is navy in the product reference but red in a keyframe, fix the conflicting input before trying to solve the problem with a longer prompt.
How to fix a prompt that misses your intent
Treat these as practical editing suggestions rather than guaranteed model behavior. Change one instruction at a time so you can tell what helped.
| Problem | Revision to try |
|---|---|
| The main action is unclear | Name the actor, verb and object: “The woman lifts the mug” |
| The camera moves unnecessarily | Start with “Fixed camera” and remove competing camera directions |
| A product changes shape | Check the product reference and simplify handling or rotation |
| Events occur in the wrong order | Split the sequence into explicit beats; add matching keyframes if available |
| A person speaks too quickly | Shorten the line or give the speaking beat more time |
| The image barely animates | Specify visible movement for a named subject |
| The scene changes unexpectedly | Check whether the requested action or reference contradicts the starting image |
| The prompt feels overloaded | Remove adjectives that do not alter framing, action, light or sound |
For example, replace “a beautiful cinematic luxury product ad with amazing motion” with “a perfume bottle on a stone plinth; slow camera orbit; warm side light; the bottle stays still.” The second version tells the model what should change over time.
Use the Kling 4.0 prompt skill in your AI assistant
To apply this guide while working with an AI assistant, install our open-source Kling 4.0 Prompt Skill. It turns a scene brief into a copyable prompt, adapts instructions for text-to-video or image-to-video, and helps revise camera movement, references, dialogue and multi-shot plans.
Install it with the Skills CLI:
npx skills add kingvid-ai/kling-4-prompt-skill --skill kling-4-prompt
Then ask your assistant:
Use $kling-4-prompt to write a 10-second perfume ad with a slow camera push-in, warm window light and quiet room ambience.The GitHub README also includes manual installation instructions for Codex and Claude Code. The skill writes prompts; it does not generate videos or require a KingVid account. Its examples are teaching prompts, and the features available in your selected Kling mode still determine which settings you can use.
Frequently asked questions
How long should a Kling 4.0 prompt be?
There is no single length that suits every scene. A short shot may need only a few sentences. A multi-shot sequence needs room for separate beats. Write enough to resolve the important choices, then remove repetition. The larger prompt allowance described in the official 4.0 guide is capacity, not a target you must fill.
Should I use the same prompt for text-to-video and image-to-video?
Adapt it to the input. Text-to-video needs appearance and setting details. Image-to-video can focus on the motion of subjects already shown. Keep the camera, timing and sound instructions relevant to the chosen mode.
Can timestamps control the exact timing?
Use them to communicate the intended order and pace. For important visual milestones, pair the text plan with the timeline controls and Multiple Keyframes where available. Review the generated result rather than assuming text timestamps enforce exact timing.
Do these examples work in Kling 4.0 Flash?
They are natural-language starting points, not a verified Flash compatibility test. Check the features available to your account and selected mode. For rollout context, read Kling 4.0 Flash and what Kling 4.0 is.
Where can I find more prompts?
Browse the Kling prompt collection for more ideas, then adapt the subject, motion and camera to your own references. Reusing a structure is useful; copying details that contradict your input image is less useful.
Official sources and further reading
This article is an independent guide based on these official materials, checked October 3, 2026:
- Kling 4.0 creation guide
- Kling 4.0 Multiple Keyframes guide
- Text-to-video prompt guide
- Image-to-video guide
- VIDEO 3.0 model guide
- VIDEO 2.6 audio guide
The general writing principles transfer across versions. Feature availability, attachment syntax and generation settings should always follow the selected model's current interface.