All tools
format

Text-to-video prompts your model will actually understand

Each generated ad concept comes with a 30–60 word visual prompt — concrete subjects, lighting, camera move, mood — ready for Sora, Runway, or Kling.

Prompts are concrete: subject + setting + lighting + camera + mood. No vague 'cinematic' filler.

Always written in English regardless of UI language — for maximum compatibility with text-to-video models.

Each of the 10 ad concepts gets its own prompt, matched to that concept's storyboard.

Sample outputs

#1

Macro shot of a vitamin C serum bottle on white marble. Soft morning light from camera-left, 50mm lens. Slow push-in. Calm, clinical mood.

#2

Wide shot of woman in linen dress walking down sun-dappled street. Golden hour, handheld, slight shake. 35mm. Optimistic, alive.

#3

Top-down kitchen counter. Espresso machine pulls a shot. Steam, rich crema. Warm tungsten light. 60fps slow-motion.

Frequently asked

Which video models do these prompts work with?

Tested mental model is Sora and Runway Gen-3 — they reward concrete subject + camera + lighting language. Kling and Pika behave similarly.

Why are prompts always in English?

All major text-to-video models are trained primarily on English captions. Mixed-language prompts produce worse results. The Chinese-language scripts pair with English visual prompts on purpose.

Can I get just the prompts without the full ad concept?

Not yet — every generation includes hook + script + storyboard + prompt. Most teams use the full bundle, but you can extract only the prompts.

Try it. 20 free credits.

Sign up and get 20 credits — 4 free generations of 10 creatives each.

Generate 10 video prompts