Prompt guide
Kling 3.0 and Kling 4.0 prompt guide
Kling follows motion well when the prompt is written as a sequence of events: who is in the shot, then what happens first, next and last, then the sound. Kling 3.0 Standard starts at 8 credits for 5 seconds at 1080p, and Kling 4.0 runs up to 30 seconds (59 credits at 720p) and takes up to 8 keyframes between a start and an end frame. In our data Kling 3.0 Pro finished 28 of 28 jobs with no policy refusals.
Updated November 7, 2026. Viralance is an independent platform that runs these models; it is not affiliated with their makers.
Versions this guide covers
| Version | Length | Resolution | Sound | Start + end frame | Credits |
|---|---|---|---|---|---|
| Kling 4.0 | 5 / 10 / 15 / 30s | 720p / 1080p | Yes | Yes | 10–89 |
| Kling 3.0 Pro | 5 / 10 / 15s | 1080p | Yes | Yes | 10–28 |
| Kling 3.0 Standard | 5 / 10 / 15s | 1080p | Yes | Yes | 8–22 |
The prompt structure
Roles → overall look → sequence of events (first, then, finally) → audio
- Roles
- Who or what is in the shot, each with a short, stable description you reuse word for word: "the courier (young man, yellow raincoat)". Kling's own 4.0 guide opens prompts with the asset roles.
- Overall settings
- Place, light, lens and style in one line: "rainy Tokyo side street at night, neon reflections, 35mm, cinematic".
- Sequence of events
- The actions in order with clear joins: "First… then… finally…". This is where Kling is strongest; give a 5-second clip two events and a 15-second clip four or five.
- Camera
- One move per event at most: "the camera tracks beside him", "slow push in on the parcel".
- Audio
- Dialogue with the speaker and tone in brackets, then sound effects. Kling 4.0 always generates sound; on Kling 3.0 the studio's Sound switch turns it off.
The courier (young man, yellow raincoat) and a parcel tied with string. Rainy Tokyo side street at night, neon reflections on wet asphalt, cinematic, 35mm. First, the courier runs into frame and skids to a stop under a shop awning. Then he checks the parcel label and looks up at the building. Finally, he smiles and rings the bell. The camera tracks beside him, then pushes in on his face. Rain on the awning, a bicycle bell in the distance.
Kling 4.0, Kling 3.0 Pro or Standard
Kling 3.0 Standard (8 credits for 5 seconds) and Pro (10) both render 1080p at 5, 10 or 15 seconds. Pro is the one to use for faces and fine detail, and it is the Kling model that takes reference images in Reference mode. Standard is the better first try for motion and product shots.
Kling 4.0 adds 30-second takes, a 720p tier (10 credits for 5 seconds) and keyframes. It always generates sound and has no negative prompt, so leave out "no subtitles" style lists except the two negatives Kling's own guide calls necessary: no subtitles and no background music.
Our data covers Kling 3.0 only so far: Pro completed 28 of 28 jobs, Standard 24 of 26, with median waits of 4.5 and 2.7 minutes. Samples are small.
Writing motion Kling can follow
Use verbs with a direction and a speed: "turns slowly to the left", "sprints toward the camera", "the cloth falls away in one motion". Vague verbs ("moves", "interacts") produce wandering motion.
Physical cause and effect reads well: "she drops the ball and it bounces twice" gives Kling a timeline. For a 15- or 30-second take, mark the beats with "then" so later events are not squeezed into the last second.
Our studio sends Kling 3.0 the provider's default negative prompt ("blur, distort, and low quality"); there is no field to change it, so describe the look you want positively.
Dialogue and sound on Kling
Write the speaker, the tone in brackets and the line: Mother (softly, surprised): "You came back." Kling's 3.0 guide uses this pattern. Keep one or two lines per 5 seconds.
Through fal, Kling 3.0 voices English and Chinese; other languages may be translated to English. fal's notes also say to write English speech in lowercase and acronyms or names in uppercase. Kling 4.0 lists more languages, including German, French, Spanish and Portuguese; test a short line before a long take.
If the clip needs your own voice or music, turn Sound off on Kling 3.0 and add a voiceover or a score in Audio Suite afterwards.
- Kling 3.0 Pro with a saved AI Influencer voice in Reference mode is priced separately (the studio shows it before you generate).
- The studio prompt limit is 2,000 characters. Kling 4.0 accepts far longer prompts in Kling's own app, but a tight sequence beats a long one.
Camera words Kling 3.0 and 4.0 follows
| Write | Use it for |
|---|---|
| the camera tracks beside | Walking, running and driving shots; Kling keeps pace well. |
| slow push in on | Ending a sequence on a face or a product detail. |
| orbit around … 180 degrees | Product turns and hero reveals; say how far it goes. |
| crane up to reveal | Ending a scene on the city or the crowd. |
| low angle, looking up | Sports, sneakers, heroic characters. |
| first-person view | Action and POV ads; pair with what the hands do. |
| locked-off camera | Dialogue and transformations where only the subject should change. |
| slow motion | Splashes, hair, fabric, impacts; costs nothing extra but uses up the clip's time. |
| cut to a close-up | A change of shot inside a Kling 4.0 long take; results vary, so test it at 10 seconds first. |
Start and end frames
All Kling models on Viralance except Kling 2.5 Turbo Pro take an end frame in image-to-video mode, at the image-to-video price: 8 credits for 5 seconds on Kling 3.0 Standard, 30 for 10 seconds of Kling 4.0 at 1080p.
Kling 4.0 keyframes. After you add an end frame, Kling 4.0 shows a keyframe strip: up to 8 images between the start and the end, in order. Each keyframe is a moment the clip has to pass through, which is how you get a controlled multi-stage change (sketch → clay → painted model → final product) in one take.
- Write the prompt as the path between frames: "first the sketch fills with color, then it gains depth, then the camera pulls back to the finished object".
- Space keyframes evenly in your head: 3 keyframes in a 10-second clip means a change every 2 to 3 seconds. Crowding 8 keyframes into 5 seconds produces flicker.
- Keep the camera position the same across all frames unless the move is the point.
- Kling 4.0 image-to-video takes the shape of the start photo; crop every frame to the same aspect ratio before uploading.
Refused or misread prompts, and the fix
A named actor, athlete or creator
Blocked before the job starts. Describe a fictional person instead. See why AI video prompts get rejected.
Motion goes wrong halfway through a long take
Too many events for the length, or events without order. Number them with "first, then, finally", or split the idea into two 5- or 10-second clips and join them in an editor.
Hands or limbs warp during fast action
Slow the action down ("slowly reaches"), keep hands out of the foreground, or use Kling 3.0 Pro for close shots of people.
The subject changes between shots
Repeat the exact same role description in every prompt, or use Reference mode on Kling 3.0 Pro with images of the subject.
Unwanted subtitles or music on Kling 4.0
Add "No subtitles. No background music." at the end; Kling names these two as the negatives worth writing.
10 Kling 3.0 and 4.0 prompts to copy
Each one opens in the studio with the model, mode and prompt filled in. Photo prompts need your own image uploaded first.
- Kling 3.0 Standard · text · 10s · 14 credits
1. Courier in the rain, three beats
The courier (young man, yellow raincoat) carrying a parcel. Rainy city side street at night, neon reflections, cinematic, 35mm. First, he runs into frame and skids to a stop under an awning. Then he checks the label. Finally, he rings the bell and smiles. The camera tracks beside him, then pushes in. Rain on the awning.
Three events in 10 seconds leaves each one about three seconds.
Use this prompt - Kling 3.0 Standard · text · 5s · 8 credits
2. Sneaker drop, low angle
A white high-top sneaker falls onto wet black asphalt in slow motion, water splashes outward, then the camera orbits 180 degrees around it at ground level. Studio rim light, dark background, commercial look. A deep thud on impact, then silence.
One impact and one camera move; no brand names.
Use this prompt - Kling 3.0 Pro · text · 10s · 18 credits
3. Kitchen dialogue
Grandmother (70s, flour on her apron) and her grandson (teenager) in a sunny kitchen. Grandmother (laughing): "you put salt in it again." Grandson (embarrassed): "it was in a sugar jar." The camera holds on both of them at the table. Clock ticking, a kettle starting to whistle.
Lowercase English speech per fal's note; tone in brackets.
Use this prompt - Kling 3.0 Standard · photo · 5s · 8 credits
4. Animate a product photo
The perfume bottle stays centered and unchanged. A slow orbit of 90 degrees to the right, soft light moving across the glass, a few petals drifting down behind it. Locked focus on the label.
Upload your own product photo. Kling 3.0 Standard finished 17 of 19 photo jobs in our data.
Use this prompt - Kling 4.0 · text · 30s · 720p · 59 credits
5. 30-second single take, Kling 4.0
The hiker (woman, red jacket, grey backpack). Alpine ridge at sunrise, wide lens, natural colors. First, she climbs the last rocky steps toward the camera. Then she stops and catches her breath. Then she turns and looks over the valley as the sun clears the peaks. Finally, the camera cranes up and back to reveal the whole mountain range. Wind, boots on gravel, her breathing. No background music.
Four events for 30 seconds. At 720p this is 59 credits; test the first 10 seconds as its own prompt before paying for the long take.
Use this prompt - Kling 4.0 · photo · 10s · 1080p · 30 credits
6. Sketch to finished product, keyframes
First the pencil sketch of the chair fills with color, then it gains wood texture and shadows, then the camera slowly pulls back as the finished chair stands in a bright room. Locked-off camera until the last second.
Start frame: the sketch. Keyframes: a colored version and a textured render. End frame: a photo of the real chair.
Use this prompt - Kling 4.0 · text · 10s · 720p · 20 credits
7. Street dance, first-person crowd
A dancer (woman, oversized denim jacket) in a city plaza at dusk. First, she starts with small shoulder moves. Then she spins twice and drops into a freeze. The crowd around her claps. The camera circles slowly at chest height. Sneakers squeaking, claps, distant traffic. No background music.
Dance is motion-heavy; Kling keeps limbs more coherent when the moves are named in order.
Use this prompt - Kling 3.0 Standard · text · 5s · 8 credits
8. Food, cause and effect
A spoon cracks the caramel top of a creme brulee in close-up, the crust breaks into shards, then the spoon lifts out a creamy bite. Warm cafe light, shallow depth of field. A crisp crack, then a soft scrape.
A visible cause (spoon) and effect (shards) gives Kling a clear timeline.
Use this prompt - Kling 3.0 Pro · text · 5s · 10 credits
9. Fantasy character turn
The ranger (elf woman, green hooded cloak, silver bow). Misty pine forest at dawn, painterly fantasy style. She lowers her hood, turns her head toward the camera and draws an arrow in one smooth motion. Slow push in on her eyes. Birds, wind in the trees.
Pro holds faces and costume detail better than Standard in close shots.
Use this prompt - Kling 3.0 Standard · photo · 10s · 14 credits
10. Before and after renovation, start and end frame
The camera stays still while the room changes: the old wallpaper peels away, fresh white paint spreads across the walls, the floor turns to light oak and the furniture slides into place. Daylight stays the same throughout.
Start frame: the room before. End frame: the same room after, shot from the same spot.
Use this prompt
Running a Kling 3.0 and 4.0 prompt on Viralance
- 1
List the roles
Write each person or object once with a short, fixed description you can reuse word for word.
- 2
Write the events in order
Use first, then and finally. Two events for 5 seconds, three or four for 10 to 15, four or five for Kling 4.0 at 30.
- 3
Add one camera move and the sound
Attach one move to the event it belongs to, then write dialogue with the speaker and tone, then sound effects.
- 4
Test short on Kling 3.0 Standard
Open the studio with Kling 3.0 Standard at 5 seconds; the cost shows before you generate.
- 5
Scale up
Move to Kling 3.0 Pro for faces, or to Kling 4.0 for long takes and keyframes once the sequence works.
Kling 3.0 and 4.0 prompt questions
What is the best prompt structure for Kling?
Roles first, then the overall look, then the events in order with first, then and finally, then one camera move, then audio. Kling follows ordered events better than a list of adjectives.
Does Kling support negative prompts?
Not in the Viralance studio. Kling 3.0 runs with the provider's default negative prompt (blur, distortion, low quality) and Kling 4.0 has none. Kling's own guide says the only negatives worth writing in the prompt are no subtitles and no background music.
How do Kling 4.0 keyframes work?
In image-to-video mode, add a start frame and an end frame; Kling 4.0 then accepts up to 8 more images in order between them. The clip passes through each one. The price is the same as plain image-to-video.
How long can a Kling video be?
Kling 3.0 Pro and Standard make 5, 10 or 15 seconds. Kling 4.0 makes 5, 10, 15 or 30 seconds; a 30-second clip costs 59 credits at 720p. For anything longer, join clips in an editor.
Can Kling make characters talk?
Yes. Kling 3.0 and 4.0 generate speech and lip movement. Write the speaker, the tone in brackets and the line. Through fal, Kling 3.0 voices English and Chinese and may translate other languages to English. Kling 3.0 Pro at 15 seconds costs 28 credits ($8.40 on the entry pack).