24 AI video prompts to copy and paste, sorted by goal
By AI-0 editorial team

On this page
The best AI video prompts describe the camera before the subject: the shot size, the lens, the rig and how it moves, the speed, and where the shot ends, then what the subject does and what we hear. Below are 24 copy-paste prompts sorted by goal, each with a note on why it works and which AI-0 video model suits it.
Most video prompts shared online read like a mood board: "epic, cinematic, golden hour, 4K". The model has nothing to film, so you get an aimless drift and a person who just blinks. Every clip prompt we write for AI-0's own videos names the camera first, and so does every prompt below.
What goes into a good AI video prompt?
We use seven parts, roughly in this order:
| Part | What to write | Example |
|---|---|---|
| Shot size | How much is in frame | Wide, medium, close-up, extreme close-up |
| Lens | The focal length | 24mm for big spaces, 85mm for faces and products |
| Rig and move | What holds the camera, what it does | Drone, gimbal, slider; push in, orbit, track alongside |
| Speed | How fast things move | Slow, steady, fast, slow motion |
| Subject action | One clear thing that happens | "She slides the cup toward the camera" |
| End framing | Where the camera lands | "Ends with the cup filling the lower half of the frame" |
| Sound | What we hear | "Sound: the cup touching wood, café chatter" |
End framing is the part almost everyone leaves out. Without it the camera drifts until the clip runs out.
Keep to one action per clip, and drop "cinematic", "epic" and "4K", which are overrated and change very little. For writing a scene from scratch, see our text-to-video guide; for stills, the AI image prompts guide.
Prompts with a "Sound:" line need an audio model: in AI-0, MiniMax H3 MAX, or Grok Imagine, which adds sound itself. Elsewhere, delete the line and add sound in the edit.
How to use these prompts in AI-0
- On the Home screen, tap Video, then tap the model row and pick the model named in the note.
- Paste the prompt and swap in your subject. For photo-to-video prompts, and any marked Grok Imagine or Runway Gen-4 Turbo, add an image as a reference first.
- Check the credit cost on the Generate button and make one clip before any variations. Video costs far more than images, as the credits guide explains.
AI video prompts for product shots
Extreme close-up, 100mm macro lens, slow slider move left to right. A drop of water runs down a frosted perfume bottle on wet black slate. Ends with the bottle centred and still. Sound: one soft drip.
One small moving detail is easy to get right. Leave out labels, which models garble. MiniMax H3 MAX.
Medium shot, 50mm lens, slow orbit on a dolly track. White running shoes on a pale concrete block, dust floating in side light. Ends on a three-quarter view of the left shoe.
An orbit shows the whole shape in one move. MiniMax H3 MAX at 2K, with your product photo as a reference.
Close-up, 85mm lens, locked-off tripod. A hand slides a steaming coffee cup across a café table toward the camera. Ends with the cup filling the lower half of the frame. Sound: cup on wood, café chatter.
One hand, one simple move, and count the fingers. Grok Imagine, from a still.
Prompts for talking-to-camera creator clips
All three need sound, so use MiniMax H3 MAX.
Medium close-up, 35mm lens, handheld phone selfie with slight shake. A freckled woman walks down a sunny street and says to the camera, "Okay, I found the best croissant in the city and it's three euros." Ends on her grin. Sound: her voice, street noise.
Keep it to one short line; longer speeches need a longer clip.
Medium shot, 50mm lens, static tripod at eye level. A young man on his bed leans toward the camera and says in a low voice, "Nobody tells you this about your first apartment." He pauses and raises one eyebrow. Ends slightly closer. Sound: his voice.
AI video softens expressions, so name the one you want. MiniMax H3 MAX.
Close-up, 85mm lens, slow push-in on a gimbal. A woman in a sunny kitchen holds up a jar of chili oil and says, "Three ingredients. That's it." She laughs. Ends with the jar and her face sharing the frame. Sound: her voice, a pan sizzling.
A prop gives the push-in a reason. MiniMax H3 MAX.
Cinematic establishing shot prompts
Extreme wide shot, 24mm lens, drone flying slowly forward and rising. White houses on a green cliff above a turquoise sea. Ends with the whole bay in frame and the horizon in the top third. Sound: wind, gulls.
"Horizon in the top third" leaves room for a title. MiniMax H3 MAX at 2K.
Wide shot, 35mm lens, slow dolly pull-back. A quiet Tokyo street at 7 a.m. as a baker rolls up the shutter of a tiny shop. Ends on the full shopfront. Sound: the shutter rattling, footsteps.
A pull-back reveals the place. Grok Imagine, from a still.
Wide shot, 50mm lens, static tripod, real-time. A red tractor crosses a golden wheat field under a blue sky. Ends with the tractor small in the middle of the frame.
A still camera with movement inside the frame rarely goes wrong. Wan 3.0 Prime at 1080p.
AI video prompts for action
Medium tracking shot, 35mm lens, camera car alongside, fast. A cyclist in a yellow jersey sprints downhill on a mountain road, leaning into a bend. Ends as he pulls ahead and leaves frame right. Sound: wind roar, tyres on tarmac.
Tracking alongside keeps the rider's size steady, hiding motion errors. Grok Imagine, from a still.
Wide shot, 24mm lens, handheld, quick. A parkour runner vaults a low wall on a sunny rooftop and lands in a roll toward the camera. Ends on a medium shot as she stands up.
Chains of flips bend bodies wrong, so ask for one move. Wan 3.0 Prime, with a real clip of the move as a video reference.
Close-up, 100mm lens, static, slow motion. A football boot strikes a wet ball and water sprays off it in a fan. Ends as the ball leaves frame top right. Sound: a heavy thud, muffled crowd.
A tight frame puts the detail into the spray. MiniMax H3 MAX.
Food video prompts
Extreme close-up, 100mm macro lens, slow push-in. Honey pours from a wooden dipper onto a stack of pancakes and runs down the sides. Ends on the glistening top pancake. Sound: a quiet pour.
Pours are forgiving: nobody knows exactly how honey should fall. MiniMax H3 MAX.
Top-down shot, 35mm lens, static overhead rig. Two hands pull a hot cheese toastie apart over a wooden board, the cheese stretching into strings. Ends as the strings snap. Sound: bread crackling.
Everyone waits for the cheese pull. Watch the hands. Grok Imagine, from a still.
Medium shot, 50mm lens, slow slider. A street food cook in Mexico City flips tortillas on a hot griddle, smoke rising. Ends tight on the griddle. Sound: sizzling, chatter.
The sizzle does half the work. MiniMax H3 MAX.
Travel video prompts
Medium shot from behind, 35mm lens, gimbal following at walking pace. A woman in a straw hat walks down a narrow street of blue-painted houses. Ends as she turns the corner into bright sun. Sound: footsteps, distant voices.
We never see a face that could break. Grok Imagine, from a still.
Extreme wide shot, 24mm lens, slow drone orbit. A white camper van on a cliff above a sandy beach at golden hour, two friends on its roof. Ends with the van small against the sea.
Tiny people mean faces can't go wrong. Wan 3.0 Prime at 1080p.
Point-of-view shot, 16mm lens, head-mounted camera, steady. Two hands rest on a ferry rail as it pulls into a Greek island harbour. Ends with the harbour filling the frame. Sound: engine hum, wind, a horn.
POV puts the viewer on the boat. MiniMax H3 MAX.
Anime-style video prompts
2D anime style, clean line art, cel shading, painted background. Wide shot, slow pan right. A girl in a red scarf cycles along a seaside road as petals blow past. Ends on her silhouette against the sea. Sound: a bicycle bell, waves.
Naming the look first stops a slide toward 3D. Grok Imagine, from an anime still.
2D anime style, bold outlines, speed lines. Low-angle medium shot, fast push-in. A boy with spiky silver hair clenches his fist on a windy rooftop, eyes narrowing. Ends on an extreme close-up of his eyes. Sound: rising wind, a musical sting.
Naming the expression stops the face going blank. MiniMax H3 MAX.
Keep the art style of the reference image exactly. Static camera. Her hair and the curtains move in a breeze; she blinks once and looks up from her book. Ends on the same framing as the start.
Make the still in Seedream 5.5 or GPT Image 2.5 Flare; the matching end gives a loop. Wan 3.0 Prime.
Photo-to-video motion prompts
Add your photo as a reference image and describe only the motion. Our image-to-video guide goes deeper.
Static camera, real-time. The person in the photo turns their head slightly toward the camera and smiles. Keep the face, clothes and background the same. Ends on the same framing as the photo.
Less motion looks more real; animate only people who agree to it. Runway Gen-4 Turbo, or Grok Imagine, which adds sound itself.
Slow push-in, steady. Clouds drift, the water ripples and the trees sway gently. Everything else in the photo stays still. Ends slightly closer on the centre.
Moving only the scenery is the safest way to animate a view. Wan 3.0 Prime.
Medium shot, slow orbit 30 degrees to the right. The product in the photo stays still and sharp while light slides across its surface. Ends on a three-quarter view.
Keep the orbit small, or the model invents the back of your product. MiniMax H3 MAX.
Which AI-0 video model suits which prompt?
Options at the time of writing; the app shows the current ones.
| Model | Clip length | Sound | Best for |
|---|---|---|---|
| MiniMax H3 MAX | 5 to 15 s, up to 2K | Native audio | Talking to camera, products, wides, sizzle |
| Grok Imagine | 6 or 10 s, from an image | Added automatically | Food, action and travel from a still |
| Wan 3.0 Prime | 4 to 17 s, up to 1080p | Add in the edit if missing | Wides, copied moves, loops |
| Runway Gen-4 Turbo | 5 s, 720p, from one image | Add in the edit | Gentle portrait motion |
If you only learn one model, we'd pick MiniMax H3 MAX, because it covers voices and detail.
Frequently asked questions
What is the best prompt for AI video?
The best AI video prompt names the shot size, lens, camera rig and move, speed, one subject action, the end framing and the sound. That beats any list of adjectives.
How long should an AI video prompt be?
The prompts here run about 30 to 50 words, enough for the camera, one action and the sound. Longer prompts tend to pile up actions, and each one can go wrong.
How do I write an AI video prompt with sound?
End the prompt with a line starting "Sound:" and put any dialogue in quotation marks, with a note on how it is spoken. Then use a model that makes audio, such as MiniMax H3 MAX, or Grok Imagine if you start from an image.
Can I use the same prompt in different AI video models?
Yes. Plain descriptions of camera and action work in any model, which is why these prompts use no model-specific syntax. Results still differ, so test one clip per model first.
Start with one prompt
Pick the closest prompt and run one clip before changing anything. AI-0 is free on the App Store and Google Play. New accounts start with 15 credits, and each optional ad adds 5 more.
Create on your phone with AI-0
Watch a short ad when you need credits, choose a video or image model, and create without a monthly subscription.


