How to prompt Kling v3 for typography and steady motion
By the FXI Studio team · Updated
Kling v3 is a text-to-video model with unusually strong on-screen text rendering and steady, native multi-shot framing. Prompt it with the exact words you want on screen in quotes, one camera grammar, and a clear subject — it holds legible typography where most models smear it.
Try this prompt
“Cinematic 8k, static wide. A rain-streaked window at dusk overlooking a blurred amber cityscape, bokeh headlights below. Centered title card in clean serif reading "NIGHT SHIFT" fades up, holding steady and razor-sharp for three seconds. Soft teal-and-amber grade, gentle film grain.”
Opens the studio with this prompt filled in · 65 free credits
Every strong prompt is built, not written.
| Subject | One subject, clearly staged. Kling holds framing well — give it a stable focal point. |
|---|---|
| On-screen text | Put the literal words in "double quotes" and name where they sit — "lower third", "centered title card". |
| Camera | Static, slow dolly, or a named multi-shot cut. Kling’s multi-shot handles two beats cleanly. |
| Lighting | Name source + direction. Kling renders clean gradients — high-key or moody both read well. |
| Duration | Kling v3 runs 3–15s. Hold titles on screen for at least two seconds to stay legible. |
Reproducible prompts, ready to run.
Copy any recipe, or run it free in the studio — the prompt prefills and the model is preselected for you.
Title card
A legible opening title that actually holds shape.
Cinematic 8k, static wide. A rain-streaked window at dusk overlooking a blurred amber cityscape, bokeh headlights below. Centered title card in clean serif reading "NIGHT SHIFT" fades up, holding steady and razor-sharp for three seconds. Soft teal-and-amber grade, gentle film grain.
Sound: Rain on glass, a low synth pad, a single distant siren fading out.
- Put the exact title in double quotes so Kling renders it verbatim
- Name where the text sits and how long it holds
- A static frame keeps typography crisp — motion smears letterforms
Product lower-third
A spec callout that stays readable over motion.
Cinematic 8k, slow dolly-in. A matte-black espresso machine on a warm walnut counter, steam curling from the group head, morning light from frame-left. A crisp lower-third caption reads "18-bar pressure" in clean sans, holding steady in the bottom-left third as the camera pushes in.
Sound: The hiss of steam, a low café hum, one ceramic clink on the beat.
- Lower-thirds survive slow camera moves; centered titles want a locked frame
- Keep captions to a few words — Kling holds short strings best
- Warm practical light makes product surfaces read as premium
Two-beat multi-shot
A cut inside one generation for momentum.
Cinematic 8k, native multi-shot. Beat one: a low-angle of running shoes hitting a wet forest trail, mud spraying, dawn mist. Beat two: cut to a wide of the runner cresting a ridge as the sun breaks the treeline. Steady handheld energy, cool blue-green grade warming into gold.
Sound: Breath and footfalls in beat one, a swelling string rise as the sun breaks in beat two.
- Label "beat one / beat two" so Kling stages a clean internal cut
- Let the grade shift across the cut to sell the passage of time
- Two beats is the sweet spot — three starts to compress each shot
Common failures, and how to unlock them.
On-screen text is garbled.
Quote the exact words, keep them short, and hold on a static or slow frame — fast motion smears letters.
The multi-shot cut feels random.
Explicitly label "beat one" and "beat two" and describe each as its own composed shot.
Motion looks stiff.
Add a human energy cue — "steady handheld", "gentle drift" — rather than leaving the camera fully locked.
Colors look flat.
Commit to two dominant hues and name a grade — "cool blue-green warming into gold".
A single prompt is a clip. The workflow is the film.
The real leverage is in the pipeline between generations — keeping one character across every shot, and frame-chaining each clip’s last frame into the next so a storyboard becomes one continuous move. That is a built-in FXI Studio capability.
Kling v3, answered.
What is Kling v3?
Kling v3 is a text-to-video AI model known for steady framing, native multi-shot generation, and unusually legible on-screen text. In FXI Studio it renders 3–15-second clips and is the go-to model when a shot needs a readable title or caption.
How do you get readable text in Kling v3?
Put the exact words in double quotes, keep them short, name where they sit on screen, and hold on a static or slow-moving frame. Fast camera motion smears letterforms, so lock the frame while a title is on screen.
Can Kling v3 do multiple shots in one clip?
Yes. Kling v3 supports native multi-shot generation. Label the shots "beat one" and "beat two" and describe each as its own composed frame; two beats per clip is the reliable sweet spot.
What is Kling v3 best for?
Kling v3 is best for title cards, product captions, lower-thirds, and any shot where on-screen typography must stay crisp — plus steady, well-framed motion that other models render too loosely.
How do I try Kling v3?
You can try Kling v3 in FXI Studio. New accounts start with 65 free credits, and a one-click button prefills each recipe’s prompt and preselects Kling v3 for you.
Related guides
- GuideWhat is a first-frame / last-frame pipeline in AI video?
- GuideWhat is the WAN AI video model and what is it good for?
- GuideWhat is the Veo 3.1 AI video model and how do you use it?
Looking for a specific tool? Browse the documentation, the guides, or all tutorials.
Ready to direct? Jump into the studio, browse more field guides, or see pricing.
Read a guide. Then direct one.
Every prompt on this page is reproducible inside FXI Studio. Start with 65 free credits — plans run $15, $35, and $135 a month when you are ready to scale.