20 Kling 3.0 Prompts for AI Video: Ready-to-Use Examples
Kling 3.0 prompts are short descriptions of what should change in the frame you upload: which action happens, how the camera moves, and what light and sound fill the scene. In the IIshki studio you build a clip from a first frame, 3 to 15 seconds long, in 720 or 1080, with or without sound. Below you'll find the prompt formula, 20 ready-to-use examples grouped by task (portraits, products, landscapes, stylized animation, multi-shot, sound) and the mistakes that make Kling 3.0 ignore what you asked for.
If this is your first time with the model, skim the Kling 3.0 guide for the interface and settings. This article is a working prompt library: copy, swap in your subject, run.
The Kling prompt formula
Kling sees your whole frame, so there's no need to describe the picture. The prompt only answers one question: what happens in these seconds. A reliable order has four parts:
- Action — a verb and a subject: "the cat jumps onto the table", "steam rises from the cup".
- Camera — a single move: "slow push-in", "static camera", "arc around the subject".
- Light and setting — "golden backlight at sunset", "fog drifting low over the ground".
- Sound — if sound is on: "waves breaking", "a shutter click".
A template for any scene:
[Who/what] [does what]; [camera move]; [light, weather, mood]; [scene sound].
Precision beats length. "The woman smiles" is weak: the model decides how fast and where she looks. "The woman slowly raises her eyes to the camera and gives a faint smile, static camera" is a prompt Kling executes predictably.
Which prompt and which model for the task
| Task | Model | Why |
|---|---|---|
| Animate a portrait or product shot | Kling 3.0 | Holds the source frame and follows the described motion |
| A multi-scene clip up to 15 s | Kling 3.0, Multi-shot mode | Up to 5 scenes, each with its own prompt, in one file |
| No source frame yet | Nano Banana Pro → Kling 3.0 | Image from text first, motion second |
| Copy a dance or gesture from a video | Kling Motion Control | Transfers motion from a reference clip to your character |
| A character speaks a line | Veo 3.1 or Hailuo H3 | Speech and lip sync in the shot |
20 ready-to-use Kling 3.0 prompts
Every example assumes image-to-video: upload a matching image and paste the prompt. Swap the subject, place and light for your own — the formula still holds.
Portraits and people
- The woman in the photo slowly turns her head to the camera and gives a faint smile, a light breeze moves her hair; static camera; soft diffused window light.
- The elderly man in the old photograph blinks, glances to the side and nods; very slow push-in on his face; warm, muted light.
- A man in a jacket walks down a night street toward the camera; the camera smoothly pulls back, keeping him centered; neon signs reflect in puddles; footsteps and distant traffic.
- A runner ties her shoelace, stands up and takes the first running step; knee-height camera, gentle tracking; low morning sun from the side.
Products and ads
- A perfume bottle stands on wet black stone, a drop slowly slides down the glass; the camera arcs 90 degrees around it; backlight, highlights on the edges.
- A sneaker floats in the air and slowly spins on its axis; static camera; studio light, dark gradient background, fine dust in the beam.
- A barista pours milk into a latte and a pattern forms on the surface; slow top-down push-in; warm morning light; clinking cups and a hissing espresso machine.
- A soda can gathers condensation, a drop runs down the side, the tab pops open; macro shot, static camera; cold blue light; the sound of the can opening.
Nature, cities, interiors
- Fog slowly drifts through a mountain valley, clouds move along the slopes; the camera smoothly rises; soft dawn light.
- Waves roll onto an empty beach and pull back, leaving foam; static camera at sand level; the sun low on the horizon; surf and seagulls.
- An evening city from above: windows light up, headlights crawl along the streets; slow pan left to right; blue hour.
- A cozy room: a fire crackles in the fireplace, snow falls outside, a curtain sways; slow push-in on the fireplace; warm light; crackling logs.
Stylized animation
- A hand-drawn anime character turns and waves, hair and clothes flutter; static camera; keep the drawn style of the source.
- A watercolor landscape comes alive: the river flows, trees sway, clouds drift; static camera; brushstrokes stay visible.
- A clay fox figure takes a few steps forward and sits down; stop-motion look; static camera; soft studio light.
- An old postcard comes alive: snow falls on the roofs, windows flicker; very slow push-in; vintage grade, light grain.
Multi-shot: one clip, several scenes
In Multi-shot mode each scene gets its own short prompt and duration. Up to 5 scenes, each no longer than 12 seconds, 15 seconds for the whole clip.
- Scene 1 (4 s): a courier rings the doorbell, static camera. Scene 2 (4 s): the door opens, a woman sees the box and smiles, close-up. Scene 3 (4 s): she opens the box and steam rises from the pizza, slow push-in.
- Scene 1 (5 s): wide shot of a coffee shop in the morning, the camera glides along the counter. Scene 2 (5 s): close-up of the barista's hands steaming milk. Scene 3 (5 s): a customer takes a sip and closes their eyes, static camera.
If you'd rather not write scenes yourself, pick Auto and the model will split one prompt into shots. For a precise story, Custom shots are more reliable.
Prompts with sound
- A campfire in a night forest, sparks fly upward, flames sway; the camera slowly lowers toward the fire; crackling wood, crickets, a distant owl.
- A motorcycle on an empty desert highway, the rider starts the engine and pulls away; the camera stays put as the bike rides into the distance; the engine roar fades, wind.
How to write Kling prompts: 5 rules
Describe motion, not appearance
Whatever is already in the frame — the dress color, the background, the props — the model can see. Extra appearance details may be read as a request to change something. Write in verbs: "turns", "pours", "slides down".
One camera move per scene
"Push-in and pan at the same time" almost always gives shake and a broken shot. Need two moves? Make two scenes in Multi-shot.
Lock first, then move
If something must stay recognizable (a face, a logo, a label), say so: "the face stays the same", "the label stays readable", and choose gentle moves. The sharper the motion, the more likely details will drift.
Sound in one closing phrase
With sound on, a single phrase like "rain and distant thunder" noticeably improves the mood. If the clip will get your own music or a voice-over from ElevenLabs, turn sound off: a clean track is easier to edit, and generating without sound costs less.
Match duration to the action
3–5 seconds for one simple move. 8–10 for an action that develops: stands up, walks over, turns around. Don't stretch a short action to 15 seconds — the model will start inventing things.
How to run a prompt in the studio: 4 steps
1. Upload the first frame
Open Kling 3.0 in the studio and upload an image. If you need a transition into a specific final pose, add a last frame too.
2. Paste the prompt
Copy an example from the list above and replace the subject with yours. Check there is one action and one camera move.
3. Pick format, quality and duration
16:9 for YouTube, 9:16 for Reels and Shorts, 1:1 for the feed. 720 for drafts, 1080 for the final cut. The token cost is shown on the model card before you run.
4. Run it and edit the prompt, not the frame
The result lands in My Generations and, if Telegram is linked, arrives as a file from the bot. Not happy? Simplify the prompt: drop a detail or soften the motion.
Common Kling prompt mistakes
| Mistake | Cause | Fix |
|---|---|---|
| Nothing moves | The prompt describes the picture | Rewrite with action verbs |
| Face or logo drifts | Sharp camera move | Soften the move, add "the face stays the same" |
| The image shakes | Two camera moves in one prompt | Keep one or split into scenes |
| Extra objects appear | The prompt mentions things not in the frame | Describe only what's visible and how it changes |
| The action gets cut off | Too few seconds for the story | Increase duration or use Multi-shot |
| Multi-shot scenes don't match | Different subjects and light per scene | Repeat the character and lighting in every scene |
What's next
Ready-made prompts are a starting point: after three or four runs you'll be writing your own with the same formula. Need a source frame? Make one in Nano Banana Pro with these photo editing prompts. Want to compare Kling with other video models? See Veo 3.1 vs Kling 3.0 and the best AI video generators. And to chain image → video → voice-over on a single screen, open Canvas.
Models in this article
Frequently asked questions
Can I write Kling 3.0 prompts in languages other than English?
Yes. In the IIshki studio Kling 3.0 accepts prompts in English or Russian, so there is no need to translate. What matters more than language is precision: what moves, how the camera behaves and what the light is like.
How long should a Kling prompt be?
One to three sentences is enough: the action, the camera move, the light and, optionally, the sound. The field accepts up to 2,500 characters, and up to 500 per scene in multi-shot mode, but a long prompt packed with details is followed less reliably than a short, precise one.
Do I need an image to use these prompts?
Yes. In the IIshki studio Kling 3.0 works from a first frame: you upload an image and the prompt describes what changes in it. If you have no frame, generate one in Nano Banana Pro first and pass it to Kling.
Why doesn't Kling follow my whole prompt?
Usually the prompt asks for several actions and several camera moves at once, or describes things that are not in the source frame. Keep one action and one camera move, and split a longer story into scenes with Multi-shot mode.
Can Kling 3.0 make a character talk?
Kling 3.0 generates sound together with the video and is at its best with ambience and scene effects. For spoken lines with lip sync, Veo 3.1 or Hailuo H3 are more reliable, and for a talking head driven by your own audio use Kling Avatars v2.
Every model in this roundup is available in the IIshki studio at reduced club prices. Club members get tokens for generations every month.
Join the clubMore articles
- How to Dub a Video With AI in 2026: Translate, Voice, Lip-SyncHow to dub a video with AI: translate the script, voice it in ElevenLabs or MiniMax, and lip-sync the speaker in Kling Avatars v2. Step-by-step, with prompts.
- Kling Avatar: How to Make a Talking Photo with AI in 2026How to make a talking photo with AI: portrait plus voice in Kling Avatars v2, text-to-speech in ElevenLabs and MiniMax, in-scene dialogue with Hailuo H3.
- Veo 3.1 reviews: what it actually doesVeo 3.1 reviews, checked against the real generation form: native audio, the 8-second ceiling, prompt rules, common errors, and how to run it without a VPN.