Cinematic Storytelling AI Video Prompts
Start with the moment the viewer needs to understand. A wide shot can place a character in the room, while a cut to their hands can reveal what they are about to do. The prompts here show how to give a short scene a beginning and an ending. Try changing the last shot and see how it alters the meaning of everything that came before.
192 videos
Ex-Lovers Pause at a Red Light
Cinematic prompt for a 30-second romantic drama: former partners lock eyes at a crowded Japanese crossing, flash through warm shared moments, then walk apart when the light turns green.
Manhattan Penthouse Heist Thriller
Photorealistic crime-thriller prompt set at a midnight Manhattan penthouse auction, covering theft, chase, and a late reveal. Character identity depends on supplied reference images.
Girl Encounters Towering Robot
Cinematic fantasy prompt for a rainy European square where a cloaked child and a towering futuristic machine share a quiet first contact.
Orbit Crash-Zoom Four-Man Entrance
Epic one-take prompt that crash-zooms from Earth's orbit into a vintage street where four differently dressed men halt and each strike a signature pose.
Motionless Woman Amid Rushing City
Seedance 2.5 brief for a young woman frozen on a golden-hour city street while traffic blurs past her calm, distant presence.
Tiny Explorer in an Appliance Showroom
Seedance 2.5 prompt for a photoreal 30-second cinematic short of a miniature woman racing through a busy electronics store.
Orbit Crash-Zoom Street Pose One-Take
Seedance 2.5 on WizstarAI: one continuous 10-second camera path from Earth's orbit down to a cobblestone corner where four women halt and hold poses from Image 1.
Seedance 2.5 Oriental Cinematic Multi-Shot Prompt
Multi-shot oriental cinematic brief for Seedance 2.5: white-Hanfu heroine, misted lakes, courtyard and lotus scenes, then grassland gallop—with locked casting, shot order, and look constraints.
30s Documentary Bar Illustration for Seedance 2.5
Seedance 2.5 prompt for a 30-second 9:16 documentary-style short of ink-sketch life inside a neighborhood bar, where three shaky driver-picking schemes settle into a quiet ending.
Package Pickup Drama in Copenhagen
Cinematic realism brief for a 22-second Copenhagen short: a man collects a parcel, banters with a clerk, then delivers it to a design-studio receptionist, with locked character refs, shot order, lighting, and dialogue.
A Foghorn Mystery in San Francisco
A Japanese sound engineer follows a mystery tied to San Francisco’s foghorns in a documentary-style thriller. The scene notes track character continuity and the city’s atmosphere.
Tiny Heroine’s Stationery Aisle Escape Prompt
Cinematic 30-second prompt of a miniature woman racing through a stationary shop, with consistent scale and grounded physics.
Video Prompt of a Son Feeding His Father
For Seedance 2.5. A 10-second one-shot where a son feeds his elderly father, overlapping with a wall memory of his father doing the same for him as a child.
Seedance 2.5 Care Across Generations
Seedance 2.5 prompt for a tender beat: a son feeds his aged father while a wall memory echoes the reverse years earlier.
Runner crossing eras and planets
A high-energy prompt of a man sprinting while the backdrop jumps every few seconds through medieval villages, Mars, neon rain cities, and other historical or futuristic worlds.
Photorealistic Stealth Office Infiltration
Prompt for a photorealistic 16:9 stealth-game sequence in which a female protagonist recovers a phone from a university office, guided by Image1 character consistency and Image2–Image3 storyboards.
Cinematic Halloween Romance Prompt for Seedance 2.5
Shot-by-shot prompt for a 30-second Halloween romance film: a vintage room, a cherished photograph, and a chance encounter that feels enchanted.
Japan Kindness Photo Rescue Film
Chronological multi-scene Seedance 2.5 prompt in which a Japanese woman helps a man gather scattered photographs across an early-2000s neighborhood.
34-Second Superhero Awakening Video Prompt
Shot-by-shot brief for a 34-second photorealistic origin film covering seven beats from power discovery through first flight.
Topview T1 and Seedance 2.5 Side-by-Side: Four Capability Probes
Identical conditions apply to both models across four scenarios—blackboard writing, finger counting, window opening, and an anime duo dance—to compare how each handles the same functional demands.
Ordinary Morning in 1950s Korea
Cinematic 30-second period piece: a young Korean woman’s quiet domestic day in the 1950s, from folding laundry at home to a brief errand through her neighborhood lane.
Rack-Focus Optics Check
Technical cue for an optical focus pull that moves clarity off a near-field model onto a wall silhouette and a man’s locked gaze.
Seedance 2.5 Person Swap Prompt
Shows a Seedance 2.5 workflow for swapping a person inside a video by replacing the subject with someone from a still image.
Four Gray-Box Scenes to Compare Kling 4.0 and Seedance 2.5
A previs comparison for Seedance 2.5 and Kling 4.0, using four rough clips to test blocking, timing and camera moves in an office, on a road, at a crosswalk and in a cavalry charge.
How to use these video prompts
Find a video
Browse the categories and open an example close to what you want to make.
Open the prompt in the generator
On the video page, select “Make Similar” to load its prompt, model and aspect ratio.
Make it your own
Change the subject, setting or camera movement. Add your own references if needed, then generate your video.
Showcase FAQ
We've answered the most frequently asked questions
LumiYing supports cinematic scenes, action sequences, character animations, product showcases, nature footage, architecture visualizations, and creative storytelling clips from text prompts, images, audio, or video references.
Every video in this gallery is generated using AI models available on LumiYing. Users provide prompts and references, and the AI produces the final video.
Yes. These are real outputs from models available on LumiYing. Sign up and start creating videos using your own prompts and references.
Options vary by model. Most video models support landscape and portrait formats, with output resolution ranging from 720p up to 4K depending on the model.
Create your own AI video
Start with a prompt from the gallery or write one from scratch.