Free tool · No signup · Works with every major model
AI video prompt generator— for Veo, Seedance, Kling & Sora
Every video model speaks a different dialect. Pick your target, build the prompt with real cinematography controls — camera, lighting, native audio — or type one line and let AI fill draft the rest. Copy-paste ready.
Structured JSON with native-audio fields — copy as JSON or flattened prose. The prose output works on any model.
JSON prompt
// fill fields or use AI fill to start
Veo accepts natural language — structured JSON is a community technique that keeps complex prompts organized and repeatable. "Copy as prose" gives the flattened sentence version.
The model dialect cheat sheet
Veo 3 / 3.1: loves real cinematography terms and is the only one where audio direction truly pays off — quoted dialogue with a speaker note produces lip-synced speech. JSON or prose both work; the dedicated Veo builder has field-level presets.
Kling: responds well to motion-first prompts — lead with the action and camera move, keep style notes short. Strong physical dynamics, no native audio: skip dialogue, describe movement instead.
Sora:handles long descriptive paragraphs and scene logic well — write it like a short film treatment. Be explicit about shot boundaries if you don't want it inventing cuts.
Comparing output costs before you burn credits? The cost calculatorhas verified per-second prices for all of these. Planning a multi-shot sequence? That's the Shot Planner.
Copy-paste examples to start from
Three originals from the prompt library (44+ prompts, updated weekly) — a viral vlog, an ASMR macro, and a surrealist product ad.
Bigfoot's morning vlogViral & memes
{
"subject": "Bigfoot holding a selfie stick, matted auburn fur, kind tired eyes",
"action": "whispers to the camera while walking, glances over his shoulder, ducks under a branch that thwacks the lens",
"dialogue": "\"Day 400 of hiding from hikers. Found the good berry spot. Do NOT tell the raccoons.\" — gravelly whisper, out of breath",
"scene": "misty Pacific Northwest forest at sunrise, ferns and cedar trunks",
"shot": {
"composition": "selfie ultra-wide, face slightly too close",
"camera_motion": "handheld shake with walking bob",
"lens": "action-cam fisheye"
},
"lighting": "cold dawn fog with warm sun shafts breaking through",
"style": "hyper-real 4K vlog footage, influencer framing",
"audio": {
"sfx": "twigs snapping, birds scattering, fur brushing the mic",
"music": ""
},
"color_palette": "forest green, fog gray, auburn",
"negative_prompt": "text, watermark, cartoonish fur, extra arms"
}
{
"subject": "a translucent amber glass mango on a dark walnut cutting board",
"action": "a chef's knife slices clean through; crystalline slices fall in sequence and ring softly as they settle",
"scene": "dark minimalist kitchen studio, single spotlight on the board",
"shot": {
"composition": "extreme close-up at board level",
"camera_motion": "static",
"lens": "100mm macro"
},
"lighting": "single hard key through the glass creating amber caustics on the board",
"style": "hyper-real ASMR macro, 120fps slow-motion feel",
"audio": {
"sfx": "high crystalline shear, glass slices chiming like bells, knife tap on wood",
"music": ""
},
"color_palette": "amber, walnut brown, black",
"negative_prompt": "text, hands in frame, shattering, real fruit flesh"
}
{
"subject": "a faceted glass perfume bottle standing on wet black sand at the tide line",
"action": "an incoming wave rushes toward it and, at the moment of impact, freezes and transforms into billowing white silk that settles around the bottle",
"scene": "volcanic beach at dawn, low mist",
"shot": {
"composition": "low wide shot at sand level",
"camera_motion": "slow push-in through the silk",
"lens": "35mm"
},
"lighting": "pale dawn backlight through the fabric, bottle catching a single sunray",
"style": "luxury surrealist commercial, Jonathan Glazer restraint",
"audio": {
"sfx": "wave roar cutting to fabric whisper",
"music": "one sustained cello note resolving"
},
"color_palette": "black sand, silk white, dawn rose",
"negative_prompt": "text, logo, people, water droplets on lens"
}
Which AI video models does this prompt generator work with?
All of the current generation. The Veo 3 tab outputs the community JSON format plus a flattened prose version; the prose transfers cleanly to Kling, Sora, Hailuo and Wan. The Seedance tab outputs the flowing prose-with-references format ByteDance's models prefer. Every model speaks a slightly different dialect — the cheat sheet below covers the differences.
Is it really free with no signup?
Yes. The builders run in your browser and the AI fill (one line in, all fields drafted) runs on our infrastructure, free, limited to 10 uses per day. No account, no watermark on anything — we don't touch your video, only the text prompt.
What makes a good AI video prompt?
Specificity in four layers: a concrete subject doing one clear action, real cinematography language (composition, camera motion, lens), a lighting description, and — for models with native audio like Veo 3 — explicit dialogue, sound-effect and music direction. Vague prompts average everything out; specific prompts commit. Steal structure from the examples below or the full prompt library.
Should I write prompts as JSON or natural language?
Models accept both. JSON isn't officially required by any of them — it's a working technique: fields keep complex prompts organized, let you change one variable per re-roll, and stay reusable across shots. For Seedance specifically, flowing prose with Image/Video references outperforms rigid structure, which is why it gets its own tab.