Prompt library

MiniMax H3 Prompts

28 copy-ready MiniMax H3 prompts across cinematic shots, action, anime, product ads, native dialogue and video editing — many paired with the real clip they generated. Copy one, make it yours, or send it straight to the generator.

Try a prompt now See the showcase

Real output

CinematicText to Video5s

Desert Convoy Past the Pyramids

Low tracking shot, camera skimming inches above the sand. A convoy of matte-black off-road trucks races past the Great Pyramids in hard late-afternoon desert light. Dust plumes tear off the wheels and drift across the lens. The camera holds a stable low angle alongside the lead truck, sunlight flaring off its windshield, the pyramids sliding by in the background. Gritty realistic physics, dense coherent motion, cinematic color grade with warm sand tones and deep shadows.

Real output

ActionText to Video15s

Festival Race in Four Shots

A fast festival race told in four tightly cut cinematic shots, same lead character in every shot. Shot 1 — wide: runners burst through a paper banner into a crowded festival street strung with lanterns. Shot 2 — tracking medium: the lead character weaves between market stalls, breathing hard, colored streamers falling. Shot 3 — low close-up: their shoes pound wet cobblestones, splashing through a shallow puddle. Shot 4 — front-on slow push: they cross the finish line into confetti, laughing. Keep the character’s face, hair and outfit consistent across every cut. Natural crowd noise, footsteps and cheering synced to the action.

Real output

Sci-FiText to Video12s

Mech Transformation City Chase

High-speed chase through a dense city at dusk. Two sports cars weave through traffic; mid-drift, their panels split and reconfigure as both vehicles transform into towering combat mechs without breaking stride. Close tracking shots whip around joints locking into place, debris and sparks scattering with correct physics. An explosion blooms behind them as the mechs vault over an overpass. Fast cuts, readable motion, metallic impacts and engine roar synced to every hit.

Real output

AnimeImage to Video8s

Reference-Locked Anime Sequence

Use the uploaded image as the strict identity reference for the character: keep the same face, hair color and length, eye design, outfit and overall art style in every shot. Generate a four-shot anime sequence: an establishing shot of the character standing on a school rooftop at sunset; a medium shot as wind lifts their hair and they turn toward the camera; a close-up on their eyes catching the orange light; a final wide shot as they walk toward the rooftop door. Consistent line weight and cel shading throughout, gentle ambient wind and distant city sound.

Real output

Product AdsText to Video5s

Paris Street Handbag Campaign

A polished 5-second luxury fashion commercial on a Paris street at golden hour. Open on a medium shot of an elegant model walking toward the camera past Haussmann facades, coat swaying naturally. The camera glides down and pushes in until a structured leather handbag on her arm fills the frame in a clean close-up, stitching and hardware sharply detailed. Shallow depth of field, warm cinematic grade, smooth commercial camera movement, no on-screen text.

Real output

ActionText to Video8s

Sunset Surfer, Vertical Cut

Native 9:16 vertical video for Reels and Shorts. A surfer carves through a glowing sunset wave, spray exploding backlit into golden light. The camera tracks alongside at water level, staying locked on the surfer as the wave curls overhead. Water droplets hit the lens. Synchronized ocean audio: the roar of the breaking wave, board hissing across the face, wind. High shutter clarity on the spray, warm orange-to-teal grade.

Real output

CinematicText to Video5s

Dragon Over a Burning Village

Low-altitude tracking shot following a massive dragon as it sweeps over a burning medieval village at night. Its wingbeats bend the flames and drag curtains of smoke and embers across the frame. The camera races just above the rooftops, keeping the dragon large in frame as thatched roofs collapse below. Firelight flickers across its scales. Realistic fire, smoke and heat-distortion physics, epic fantasy color grade dominated by orange flame against deep night blue.

Real output

Audio & DialogueText to Video8s

The Rubber Duck Interrogation

A noir interrogation room lit by a single harsh desk lamp. A weary detective in a loosened tie leans across the table and interrogates the suspect: a small yellow rubber duck sitting under the light. Detective, sternly: “Where were you on the night of the twelfth?” Beat. The duck squeaks once, nervously. The detective pinches the bridge of his nose and mutters: “They never talk.” Native synced dialogue with room reverb, chair creaks and the squeak timed as a punchline. Deadpan staging, cinematic noir lighting.

Real output

CinematicText to Video5s

Rainy Night in Three Shots

A compact three-shot night story, moody and rain-soaked. Shot 1 — wide: an empty street corner at night, rain falling through the cone of a streetlamp, neon reflections trembling in puddles. Shot 2 — insert: gloved hands slide a document across a wet café table, rain drumming on the awning above. Shot 3 — long lens: a distant figure under an umbrella turns and disappears into the alley haze. Preserve one continuous atmosphere across all three shots: cold blue-green grade, film grain, quiet tension.

Real output

Sci-FiText to Video5s

Spacecraft Push-Through

One deliberate camera move, no cuts. The camera drifts through the interior of a spacecraft, pushes slowly toward a porthole window as Earth glows outside, then passes through an open hatch into the next module. Crisp depth of field shifts as each bulkhead slides past the lens. Instrument panels blink softly in the dark; dust motes float in a shaft of sunlight. Controlled, weightless camera motion with clean cinematic detail and deep space silence broken only by a low hull hum.

ActionText to Video

Speeder Chase, One Continuous Shot

Single continuous shot, no cuts. From high above a monumental city carved into a sea cliff, the camera dives toward a streak of light racing along a narrow ledge road — a speeder hugging the rock wall at insane speed. The camera slingshots ahead of it, whips back to face the rider, then drops in tight behind the thrusters: heat haze, grit snapping off the ledge, warning lights strobing. A collapsing balcony rains debris; the rider swerves under a falling arch at the last inch, threads through hanging laundry lines, then bursts out over open air — revealing a vast waterfall-fed valley below, mist breaking into rainbow. Engine scream and wind shear throughout.

Product AdsText to Video

Luxury Headphones Macro Showcase

A 15-second luxury product film for premium wireless over-ear headphones, 16:9. [0–4s] Extreme macro tracking shot gliding across the memory-foam ear cushion, fabric weave and stitching in razor focus. [4–8s] The full headphone rotates slowly on a dark reflective surface, a single softbox highlight sweeping across the aluminum yoke. [8–12s] Exploded view: driver, magnet and housing separate in mid-air with engineering precision, each part catching rim light. [12–15s] The parts glide back together and the assembled headphones land on a hero pedestal. No hands, no text, no logos. Deep black background, precise studio lighting, subtle mechanical sound design.

Product AdsText to Video

Perfume Bottle Orbit Commercial

A 15-second cinematic fragrance commercial, 16:9. Scene 1 (0–7s): elegant macro close-up of a faceted glass perfume bottle standing on a glossy black reflective surface between two vertical neon light bars, soft pink and purple reflections playing across the glass. The camera performs one slow, smooth 180-degree orbit around the bottle. Scene 2 (7–15s): a stylish model in a fitted black leather jacket enters from the side, lifts the bottle toward the camera with a confident motion, label facing the viewer. End on a hero shot: bottle sharp in the foreground, model softly out of focus behind it. Ultra-photorealistic, realistic glass refraction, shallow depth of field, luxury campaign aesthetic.

Product AdsText to Video

Matcha Social Ad with Kinetic Type

A 9:16 vertical short-form social ad for a ceremonial matcha drink. A young athlete in a bright blue outfit and cap moves through a sunlit outdoor skate bowl mid-stride, holding a green matcha drink. Bold layered kinetic typography spelling the product name sweeps across the frame in sync with her movement. High-energy daylight, crisp shadows, punchy saturated color, fast but readable motion. Leave clean space at the top and bottom of frame for captions added in post. End on a strong action pose with the drink clearly visible.

CharacterImage to Video

Identity-Locked Autumn Walk

Use the uploaded image as the identity, hairstyle and wardrobe reference for the same woman in a belted camel-brown leather coat. She walks steadily toward the camera along an empty tree-lined boulevard, golden leaves scattered on the pavement behind her. Begin with a low front tracking shot as she approaches, then settle into a steady medium shot. Preserve her exact face, hair, coat design, belt and body proportions throughout. Confident natural walking motion, soft warm daylight, shallow depth of field. Do not change the coat and do not add other people.

CharacterText to Video

Velvet Lounge Close-Up with Sound Design

Photoreal cinematic live action, frontal close shot. A man lounges on a red velvet sofa in a dim, luxurious red-and-black interior, background softly defocused. He leans back with effortless composure; beside him, a cut-crystal tumbler of amber whiskey sweats fine condensation. He lifts the glass, ice turning slowly, and takes an unhurried sip. Sound design carries the scene: ice tapping crystal, a faint cigar burn, subtle room air, clothing movement, controlled breathing. Low-saturation warm grade, intimate and unhurried.

AnimeImage to Video

Wuxia Heroine Storyboard Sequence

Use the uploaded image as the locked character reference: half-up long black hair, openwork silver crown, indigo ribbon, layered pale hanfu with a translucent blue outer robe, deep-blue sash and long tassels — preserve every element. A wuxia sequence in three beats: she stands on a misty bamboo rooftop at dawn; leaps into the air as her robes and ribbon trail behind her in slow motion; lands silently on a temple eave, robes settling, and looks toward the horizon. Natural camera movement with seamless transitions between beats — flowing shots, never a slideshow. Ink-wash-inspired color palette, soft mist, distant guqin notes.

AnimeText to Video

Hand-Drawn Creature in a Real Kitchen

15 seconds, 16:9. Blend live-action footage of a small kitchen at dusk with hand-drawn, softly glowing 2D animation. The last sunset light lingers at the window. A tiny luminous hand-drawn creature peeks out from behind a mug, pads across the counter, and warms itself by the kettle’s steam. Shoot it like someone filming one-handed on a phone: subtle hand tremor, hesitant close-focus pulls, backlit exposure breathing, slightly coarse shadow noise. Audio: room tone, cloth friction, a soft mug clink, faucet drips, and gentle electronic chirps from the creature. Warm, quiet, wholesome — no jump scares.

ActionText to Video

First-Person Shooter Gameplay

First-person camera at eye level, handheld gameplay feel. Simulate a player in a modern-warfare FPS holding an assault rifle, advancing slowly along the perimeter of a military base at dawn. The reticle sweeps across the passage ahead; the player pauses, fires several controlled rounds at a distant target — muzzle flash, shell casings, recoil kick — then keeps pushing forward like authentic player-controlled footage. Cool natural light mixed with drifting smoke and firelight, HUD-free clean frame, rifle foley and echoing gunshots.

AnimeText to Video

Claymation Fox Canyon Leap

Claymation style throughout. A sprinting clay fox reaches the edge of a cliff and launches without hesitation into a dramatically tense, heroic slow-motion leap across a vast lava canyon. While the fox is airborne, the camera rushes at high speed beneath its belly, revealing the terrifying depth of the glowing chasm and the fully stretched pose of its clay body — visible fingerprints in the clay, slight stop-motion stutter at 12 frames per second. It lands in a tumble on the far rim, shakes itself, and strikes a tiny victorious pose. Warm lava underlight against a cool dusk sky.

CinematicText to Video

Space Opera Title Sequence

An epic theatrical space-opera teaser with title cards, 15 seconds, 16:9. Keep the pace fast and the scale enormous: sharp hard cuts, a shaking command deck, white-hot flashes, split-second black frames, and a violent jump-to-warp impact. Title typography is wide-tracked and cinematic — not pure white — with restrained material texture and a faint edge glow. Titles emerge from deep-space shadow, catch a sweep of starlight, open their letter spacing, then flash against black. Thunderous low-end hits sync to every cut; end on the main title holding over silence.

Sci-FiText to Video

Cyberpunk Character Load-In

Game-cinematic sequence with timed beats, 15 seconds. [0–2s] High-angle overhead shot: a cyberpunk heroine sits on a vivid purple floor and looks up at the camera; a game menu slides in on the right. [2–6s] Push in to her right arm: a panel slides open and her mechanical hand reconfigures, cyan LEDs flaring brighter with each click. [6–10s] She flexes the rebuilt hand; holographic stat readouts orbit her wrist. [10–15s] As she stands, the full world loads in around her — a dense cyberpunk slum with flickering neon and rain-wet streets — and HUD elements fade in: minimap, health, ammo, then a mission marker. UI motion crisp and game-authentic, synth score building to the marker ping.

Audio & DialogueReference to Video

Voice-Cloned Character Line

The character rides a motorcycle along an empty coastal road at sunset, hair whipping in the wind, and speaks directly over the engine: “Follow the wind, live free. Leave worries behind, enjoy the moment.” Use the uploaded audio as the voice-timbre reference and match its tone exactly. Natural lip sync, wind and engine bed under the line, warm backlit haze, the camera tracking alongside at speed.

Audio & DialogueReference to Video

Dialogue Replacement with Re-Performance

In the reference video, replace the woman’s line “There’s no way we can be together — it’s not that we don’t love each other, we just can’t make it to the end” with the line from the uploaded audio: “Please don’t go. This time, let’s not let go of each other.” Keep the scene, lighting and camera identical, and subtly adjust her facial performance, breathing and eye movement so the emotion matches the new line. Preserve natural lip sync throughout.

Video EditingReference to Video

Six Edits in One Pass

In the reference video, make all of the following edits at once: replace the newspaper with a green hardcover book; change the chair the character sits on to a red sofa; remove the character’s sunglasses to reveal a clear face; remove the burning-car effect and restore the vehicle to normal; replace the photo he takes from his coat with a small black notebook; and add a tree on the left side of the frame. Keep the character’s motion, the camera move and the lighting untouched.

Video EditingReference to Video

Subject Swap: Capybara Formation

Match the action in the reference video from a locked-off wide camera. Replace the three suited men with three highly photorealistic capybaras. Preserve the original movement path exactly: all three drop quickly to the floor, then rotate positions to form a pyramid. Keep the camera fixed and integrate fur, lighting and contact shadows realistically — the capybaras must have correct weight and momentum, not floaty composited motion.

Video EditingReference to Video

Green Screen to Fairy-Tale World

Remove the green-screen background of the reference video and replace it with a fairy-tale forest: giant glowing mushrooms, drifting spores, a warm cottage light in the distance. The new background must track perfectly with the original camera motion and parallax, and every background element must react to the character’s actions. Re-light the character to match the scene — soft teal ambient fill with a warm rim from the cottage side — so the composite is seamless.

Video EditingText to Video

Coffee-to-Desert Match Cut

One continuous shot with no visible edit. Push in rapidly toward the surface of a fresh coffee: milk foam, cocoa particles and dark liquid texture fill the frame. At the exact moment the swirling cocoa ridges resemble wind-carved dune ridges, transition seamlessly into a real desert landscape at the same scale and camera speed, sand grains replacing cocoa particles mid-motion. Continue pulling up and out to reveal endless dunes under a hard sun. No tearing, black frames, hard cuts or compositing seams — one fluid morph.

How to write MiniMax H3 prompts

The strongest H3 prompts share the same habits. Steal these six before you write your own.

Describe change, not a picture

A video prompt should explain what happens over time: how the subject moves, what the camera does, and how the shot ends — not just what the scene looks like.

Give every reference a job

When you upload images, audio or video, assign each one a role: "use the image for identity, the audio for voice timbre, the video for motion timing." Unassigned references drift.

Write edits as substitutions

For video editing, use explicit "replace X with Y" instructions instead of vague requests. H3 can apply several precise substitutions in a single pass.

Say what must not happen

Negative direction prevents drift: "no cuts", "do not add other people", "keep the packaging identical". State the failure you want to rule out.

Timecode longer sequences

For 10–15 second clips with several beats, block the prompt into time ranges — [0–4s], [4–8s] — so pacing and shot order land where you want them.

Lock identity in words

Even with a reference image, name the details that must stay fixed — hair, outfit, accessories, proportions. Written identity plus a reference is far more stable than either alone.

MiniMax H3 prompt questions

What is a MiniMax H3 prompt?

A MiniMax H3 prompt is the text instruction (optionally combined with image, video or audio references) that tells the Hailuo 3 video model what to generate — the scene, the motion, the camera work, the dialogue and the sound. Better prompts produce dramatically better videos.

How do I write a good H3 prompt?

Describe what changes over time, not just what the scene looks like: subject action, camera movement, timing and sound. For longer clips, block the prompt into time ranges, and add negative direction ("no cuts", "don't change the outfit") to prevent drift.

Can I copy and use these prompts?

Yes. Every prompt in this library is free to copy, edit and reuse — commercially included. Click Copy to grab the text, or Use Prompt to open it directly in the TryH3 generator.

Are the example videos real H3 outputs?

Yes. Cards marked "Real output" play an actual MiniMax H3 generation, unedited apart from compression. Cards without a clip are copy-ready prompts following the same patterns.

Which generation modes do these prompts cover?

Three: text-to-video (pure text), image-to-video (a reference image locks identity or the first frame), and reference-to-video (video or audio references for motion transfer, dialogue replacement and precise editing). Each card is labeled with its mode and opens the matching tool.

Stop reading prompts. Run one.

Pick any prompt above and render it with MiniMax H3 in minutes — 2K, native audio, no watermark on paid plans.

Open the generator