КАДР

Kling 3.0 prompts: rules and 15 ready prompts for portraits, products, action and speech

Alen founder of KADR Updated 2026-10-22 6 min read

A good Kling 3.0 prompt starts with the shot size and the camera move, describes one action of the character, the light and the sound, and has only one main camera move per shot: if you mix a push-in, a pan and a zoom, the picture starts to «float». A Kling 3.0 clip lasts up to 15 seconds. The model is strong with people, facial expressions and camera movement, and it is more relaxed than others about photos of real people. On KADR 6 seconds on Kling v3.0 at 1080p with sound cost 106 tokens, about 168 rubles, and on Kling O3 94 tokens, about 149 rubles (October 2026). Below are the rules and 15 ready prompts.

Rules

1. Order of description.

Shot size and camera move → who is in the frame → one action → surroundings → light → sound

2. One main camera move per shot. A push-in, or a pan, or an orbit, but not all at once. If you need several moves, split them into shots.

3. Camera terms Kling understands: static camera, slow push in, pull back, pan left, tilt up, tracking shot, orbit slowly, handheld.

4. Duration up to 15 seconds. The shorter the clip, the lower the chance that something breaks.

5. Do not overload a close-up with movement. Fast movement in a close-up of a face and several characters with complex actions are what most often leads to distortion.

Animate a photo

1. Portrait: subtle movement

Medium close-up. The woman slowly turns her head toward the camera and gives a soft, natural smile, a light breeze moves her hair. Static camera, shallow depth of field, warm golden hour light. Subtle, natural movement, the face stays the same.

2. Old photo

Medium close-up. The man slowly blinks, slightly turns his head toward the camera and breathes calmly. Static camera, soft natural light. Subtle, natural movement, the face stays exactly the same. No speech.

3. Family photo

The people in the photo slightly turn their heads to each other and smile, natural subtle movement, static camera. Sound: quiet warm melody.

More on animation: how to animate a photo with AI, and on restoring a photo before animating it: restore an old photo.

Products

4. Orbit around a product

The camera orbits slowly 90 degrees around the perfume bottle standing on a wet black stone. Water droplets on the glass catch the light. Dark studio background with a single rim light. The bottle stays unchanged and in sharp focus.

5. Product in hands

Close-up. A woman's hands slowly open a small gift box and lift a silver ring toward the camera, soft window light, static camera. The ring stays unchanged.

6. Push-in on a product

Slow push in on the thermo mug on a wooden table, steam rises from the open lid, soft morning light from the window. The mug stays unchanged.

Action

7. Running behind the character

Tracking shot from behind at shoulder height: a young man in a yellow raincoat runs up wet stone stairs in an old city at night. Puddles splash under his feet, street lamps reflect on the wet stones. Sound: rain and footsteps.

8. Sport

Low angle tracking shot: a skateboarder rolls fast along an empty concrete embankment at sunset, the camera follows at the side. Sound: wheels on concrete, wind.

9. Dance

Wide shot, static camera: a girl in a white dress dances slowly in an empty sunlit studio, her dress flows with the movement, dust in the light beams.

Nature and atmosphere

10. City

Slow pan right across a rooftop view of the city at blue hour, lights turn on in the windows, light clouds move in the sky. Sound: distant city ambience.

11. Nature

Slow push in through tall grass toward a calm lake at sunrise, mist over the water, birds fly across the frame. Sound: birdsong.

Speech

12. Speaks to the camera in Russian

The man looks into the camera and says in Russian: «Попробуйте сами, это проще, чем кажется». Natural lip movement, slight head nods, static camera, soft office light.

(«Попробуйте сами, это проще, чем кажется» means «Try it yourself, it is easier than it seems».)

13. A short greeting

The woman smiles, raises a glass and says in Russian: «С Новым годом!». Warm Christmas lights in the background, slow push in.

(«С Новым годом!» means «Happy New Year!».)

Keep the lines short: in Russian the lip sync is worse than in English, and Kling speaks with an accent. If Russian speech is the main point of the clip, try Veo 3.1. More: AI talking avatar.

Multi-shot scenes

Kling v3.0 understands several scenes in one prompt, up to 15 seconds in total: label the shots and repeat the description of the character in each one.

14. A walk in the snow

Shot 1 (0-4s): wide shot, a girl in a red coat walks into an empty snowy park, static camera. Shot 2 (4-8s): medium shot, the girl in the red coat stops and looks up at the falling snow, slow push in. Shot 3 (8-12s): close-up of her face, snowflakes on her eyelashes, she smiles. Same girl, same red coat and same lighting in all shots.

15. Morning at a café

Shot 1 (0-4s): wide shot of a small cozy café in the morning, a barista with curly hair opens the door, static camera. Shot 2 (4-8s): medium shot, the same barista makes a latte, slow push in. Shot 3 (8-12s): close-up of the latte art in the cup. Same barista and same warm light in all shots.

If there are more scenes or the clip is longer than 15 seconds, generate each shot separately and join them in editing.

If it did not work

  • The face drifted: slow the movement down, move the camera back.
  • The camera jerks: keep one move.
  • The character changes between shots: repeat their description in every shot, use a reference (Kling O3 takes up to 7 reference photos).
  • The model refused: remove celebrities and brands. More: AI refuses to generate.

KADR has four Kling models: Kling v3.0 (up to 15 seconds, 720p or 1080p, with or without sound), Kling O3 (first and last frame, up to 7 reference photos, editing your own video), Video O1 (720p without sound, exactly 5 or 10 seconds) and Motion Control (the person from a photo repeats the movements from your video). Model review: Kling v3. If you do not want to write it yourself, the «Improve» button on KADR will rewrite the prompt for Kling.

In short

A Kling 3.0 prompt: shot size and camera move, one action, surroundings, light, sound. One main camera move per shot, a clip of up to 15 seconds, short lines in quotation marks. The 15 prompts above cover photo animation, products, action, nature, speech and multi-shot scenes.

Frequently asked questions

How do I write a prompt for Kling 3.0?
In order: shot size and camera move, who is in the frame, one action, surroundings, light, sound. One main camera move per shot. Be specific: not «beautiful», but «warm sunset light from the side».
How many seconds does Kling 3.0 make?
Up to 15 seconds per generation: on KADR Kling v3.0 makes 4 to 15 seconds, Kling O3 3 to 15. For longer clips, several generations are joined in editing.
Can I write a Kling prompt in Russian?
The model understands Russian, but English is more precise, especially for camera terms. The character's lines can be written in Russian in quotation marks if you need Russian speech.
What is a multi-shot prompt in Kling?
It is a prompt made of several labelled shots (Shot 1, Shot 2 and so on), each with its own shot size and camera move. The character is described the same way in every shot so that they do not change.
Why does the face get distorted in Kling?
Most often because of fast movement in a close-up or several characters doing complex actions at the same time. Slow the movement down, move the camera back, keep one action.
How much does a Kling 3.0 video cost?
On KADR 6 seconds at 1080p with sound: Kling v3.0 106 tokens, about 168 rubles, Kling O3 94 tokens, about 149 rubles. At 720p without sound both cost 53 tokens, about 84 rubles (October 2026).

Kling 3.0 without a VPN

Kling v3.0, Kling O3, Video O1 and Motion Control on KADR, paid in rubles. Paste a prompt from the article or press «Improve», and the AI will adapt your prompt for Kling. Tokens never expire.

Open Kling
Alen
Founder of KADR. Kling is one of the most popular models among our users, and prompts for it are what people ask about most often.