We Made a 30-Second AI Short From One Prompt. Here's the Prompt
This deep-sea short came from one prompt. Watch the untouched result, read the exact input, and see which details gave the script and scene planner something useful to work with.

Long AI video prompts often create more confusion than control. A list of camera moves, moods, subjects, and references can pull the result in five directions at once. For this short, we used two focused sentences: one for the idea and one for the visual language. The finished video below is the untouched result.
The result first
The render took about seven minutes. We supplied the prompt, chose the length, resolution, and music mood, then downloaded the result. The script, narration, scenes, soundtrack, and caption timing were generated as part of that run.
The exact prompt
“Why the deepest place on Earth is scarier than space — the Mariana Trench hides crushing pressure, alien creatures, and sounds science still can't explain. Visual style: cinematic deep-ocean documentary look — bioluminescent creatures in pitch-black water, submersible lights cutting through darkness, immense scale, photoreal, moody teal-and-black palette.”
The 3-part prompt formula
Break that prompt apart and you get a formula you can reuse for any topic in any niche:
- 1A claim that needs an answer -- "Why the deepest place on Earth is scarier than space." The comparison creates a question the rest of the video has to resolve.
- 2Concrete evidence -- "crushing pressure, alien creatures, and sounds science still can't explain." Those nouns give the writer facts to organize and the scene planner objects and environments to show.
- 3One visual direction -- "cinematic deep-ocean documentary look, bioluminescent creatures, submersible lights cutting through darkness, moody teal-and-black palette." The look, light, and palette keep separate scenes in the same visual world.
The settings we picked
| Setting | What we chose | Why |
|---|---|---|
| Format | AI Clips (vertical short) | Every scene is a moving cinematic clip, not a still slideshow |
| Length | 30 seconds | Long enough for a payoff, short enough to be rewatched |
| Resolution | 1080p | Crisp on every phone screen |
| Music mood | Suspense | Supports the tension in the opening claim |
| Everything else | Defaults | The default voice, image, and motion models are already tuned for this |
Why this video holds attention
- The first two seconds state the claim. There is no intro, logo, or welcome message to wait through.
- One idea, argued three ways. A 30-second short has room for exactly one claim and its evidence.
- Dark scenes with one strong light source remain legible on a small phone screen.
- Word-timed captions let the story work when the phone is muted.
- The last line returns to the unresolved mystery instead of adding a generic call to action.
5 prompts you can steal
These examples keep the same structure but change the subject matter. Use them as starting points, then replace the evidence and visual references with details that belong to your topic.
- 1Why Rome's concrete outlasted ours -- seawater that makes it stronger, a lost recipe, and bridges still standing after 2,000 years. Visual style: sun-bleached ancient stone, golden-hour ruins, slow reveals, photoreal.
- 2The vending machine business nobody talks about -- weekly revenue per machine, placement deals, and why location beats product. Visual style: neon-lit city corners at night, clean product close-ups, moody teal and amber.
- 3How deep does the ocean actually go -- the sunlight zone ends fast, pressure that folds steel, and creatures photographed only a handful of times. Visual style: descending depth-scale journey, bioluminescence, pitch-black water, photoreal.
- 4Why airplane food tastes different -- your tongue at altitude, dry cabin air, and why recipes need stronger seasoning. Visual style: clean editorial food shots, cabin-window light, muted blues.
- 5The design choices casinos use to hold attention -- no clocks, indirect paths, and near-miss slot machines. Visual style: saturated neon interiors, shallow depth of field, cinematic slow push-ins.
Make your own in one prompt
Choose the AI Clips short format, paste the prompt, set the length and music mood, and generate. Watch the finished result once for the story and once with the sound off for the visuals and captions. If one scene is weak, regenerate that scene instead of replacing the full video.
Frequently asked questions
No. It is the untouched file produced from the prompt shown above. We chose the topic, length, resolution, and suspense music mood. The script, narration, scenes, music, and captions came from that generation.
Ship your first faceless video today.
Pick your niche. Pick your models. We render. From idea to finished short in under 7 minutes — no camera, no editor.
Keep reading

Faceless Reels: The Complete 2026 Instagram Growth Playbook
Trial-testing hooks before you publish takes a faceless Reels account's hit rate from 18% to 64% — here's the exact niche, character, and autopublish setup behind that number.

How to Go Viral on TikTok in 2026: The Hook-Math Playbook
94% of the TikToks that break 5M views open with a number, a question, or a contradiction in the first second — none of it is luck.

Best Free AI Video Generator in 2026, Ranked (8 Tools Tested)
One of the eight free AI video generators we tested made us wait 47 minutes for a 45-second export; another charges $3.10 for a video a rival delivers at $0.30. Here's exactly where each tool wins, where it quietly costs a weekend, and which one fits your workflow.