Free AI Video Prompt Generator
Type one line. Get the prompt a director would write: camera, lens, action by the second, one named light, audio and exclusions, ready for Seedance, Kling, Veo or any text-to-video model.
- Free, no account
- Shows every choice it made
- Scored by the free prompt checker
Your prompt
Example: one line in, a production prompt out
This is the generator's real output for the idea “a neon night market in the rain, a woman buying noodles” (5 seconds, 9:16). The free checker scores it 100/100.
SCENE: A 5-second vertical (9:16) live-action shot — a neon night market in the rain, a woman buying noodles. One unbroken single take in real time, no cuts. SUBJECT: A woman in her mid-twenties, dark hair in one long braid over her shoulder, wearing a black bomber jacket over a white tee, with a thin silver chain at the collar. Her face, hair, wardrobe and accessories stay exactly the same for the whole take. LOCATION: A crowded night market: rows of steaming food stalls under strung bulbs, hand-painted signs, wet pavement between the stalls; light rain falling, wet surfaces reflecting the light. CAMERA: The camera pushes in slowly on a dolly, about 30 cm across the take — one movement only. Framed as a medium shot, waist-up, at chest height, horizon level, no zoom. OPTICS: 35 mm lens at f/2.8: focus held on her eyes, background softly out of focus, natural 180-degree-shutter motion blur. ACTION: 0–2 s: She is buying noodles. 2–5 s: She comes to rest and holds the final position. PHYSICS: Her hair and clothing move with every head turn and settle naturally; fabric creases at the elbows. Hands keep five fingers and touch objects with visible contact — fingers wrap around, never pass through. Nothing floats, morphs or changes size; everything moves in real time. LIGHTING: Key light: a magenta-and-cyan neon sign behind and camera-right as the only key, about 6000K mixed, reflecting on nearby surfaces; it is night, everything outside the light falling off into darkness. It stays constant for the whole take; no second sun, no lens flares. AUDIO: No dialogue and no mouthed words; ambience only: sizzling woks, chatter and clattering bowls; no music. STYLE: Photoreal cinematic footage: a natural, slightly warm grade, fine film grain, natural skin texture. No on-screen text, no captions, no logos, no watermarks.
The choices it made for you
- Look: dark hair in one long braid over her shoulder; a black bomber jacket over a white tee; a thin silver chain at the collar — invented to lock the cast; replace with the real person's look.
- Camera: push-in (one movement; none was named).
- Audio: native audio on — no dialogue, ambience only, no music.
Change any of them in your idea (“a man in his forties in a yellow raincoat”, “handheld”, “neon pink sign as the only light”) and run it again.
A product ad, same idea length
Idea: “a pair of white running shoes on a wet rooftop at sunrise, product ad”
SCENE: A 5-second vertical (9:16) product shot — a pair of white running shoes on a wet rooftop at sunrise, product ad. One unbroken locked-off take in real time, no cuts. SUBJECT: The white running shoes — its shape, colours, materials and printed label stay exactly the same in every frame; nothing is added to it or removed from it. LOCATION: A city rooftop: a low concrete parapet, string lights overhead, the skyline behind. CAMERA: The camera sits on a tripod at the product's height, about 60 cm away, and moves in one slow 20-degree arc to the right at constant speed. No zoom. OPTICS: About 85 mm-equivalent at f/5.6: the whole product is sharp and the background falls off softly. ACTION: 0–2 s: The white running shoes stands centred on the plinth, front facing the lens, then the person is running shoes. 2–5 s: The white running shoes sits perfectly still, front sharp, for the final second. PHYSICS: The white running shoes has real weight: it dips slightly when lifted and settles when held still; its shape never warps. Nothing floats, morphs or changes size; everything moves in real time. LIGHTING: Key light: early-morning sun low from camera-left, about 4000K, pale and soft. It stays constant for the whole take; no second sun, no lens flares. AUDIO: No voices; ambience only: a distant city hum; no music. STYLE: Photoreal commercial product footage: crisp edges, accurate colours and materials, clean controlled reflections. No added on-screen text, captions, logos or watermarks; the product's own printed label stays exactly as it is.
An image prompt
Idea: “a ceramic coffee cup on a sunny windowsill” (output: image, 1:1)
IMAGE: A single square (1:1) still — a ceramic coffee cup on a sunny windowsill. SUBJECT: A ceramic coffee cup on a sunny windowsill. SETTING: A quiet residential street: parked cars along the curb, trees overhead, houses set back behind lawns. COMPOSITION: Subject centred, eye-level, generous negative space around it. OPTICS: 85 mm at f/8: everything on the subject sharp. LIGHTING: Key light: late-afternoon sun from camera-right, about 4500K; one source, no competing highlights. STYLE: Photoreal, natural texture, true-to-life colour, no beauty filter. No on-screen text, no captions, no logos, no watermarks.
The ten blocks every prompt gets
Text-to-video models do best with the same things a film crew needs. Leave one out and the model guesses.
- SceneLength, aspect ratio and one unbroken take, so the model knows the frame and the clock.
- SubjectAn age, hair, wardrobe and one distinguishing detail, locked for the whole shot so the face never drifts.
- LocationWhere it happens, with props the camera can actually see.
- CameraOne physical move (dolly push, handheld, locked-off), the framing and the height.
- OpticsLens, aperture, where focus sits and natural motion blur.
- ActionWhat happens, second by second, so the length is filled without rushing.
- PhysicsHair, fabric, hands and contact, the details models get wrong first.
- LightingOne named key light and its colour temperature, held constant.
- AudioDialogue, ambience or silence, stated so the model does not invent speech.
- StyleThe look and grade, ending with "no on-screen text, no captions, no logos, no watermarks".
Prompts that worked, with the video
The AI video examples show real renders next to the exact prompt, model, length and cost, plus one honest limitation each. Use them as starting points, or open the AI video generator, the AI image generator or image to video to make your own. Already have a prompt? Score it with the free AI video prompt checker.
Questions
Is the AI video prompt generator free?
Yes. It needs no account and spends no credits. It runs ClipSpeed's own prompt engine, so there is no daily cap beyond a short per-minute limit that stops abuse.
Which AI video models do the prompts work with?
The prompts are written in plain, physical language, so they work in Seedance, Kling, Veo, MiniMax, Sora, Runway and most text-to-video models. The tool suggests a model and settings, which you can ignore.
Why is the generated prompt so long?
Every gap you leave, the model fills with a guess. The generator closes the usual gaps (camera, lens, light, timing, hands and fabric, audio, exclusions) and lists each choice it made so you can change it.
Can it write AI image prompts too?
Yes. Choose Image and it writes a still-image prompt with subject, setting, composition, light and style.
Can it write image-to-video prompts?
Yes. Choose Animate a photo and it writes a motion-only prompt: what moves, how the camera moves and what must stay exactly as in your photo. You add the photo as the start frame in the generator.
Is my idea saved?
No. The idea is sent to ClipSpeed only to write the prompt and is not stored or tied to an account.
Can I use the prompt in ClipSpeed?
Yes. "Make this video" carries the prompt into the ClipSpeed generator; you see the credit price before anything runs.