by MiniMax

MiniMax H3 Prompts

MiniMax's multimodal video model — start from text, an image, a clip or a soundtrack, and get video with native stereo sound.

video

Create a 15-second hyper-realistic video set in Tokyo at night. OUTPUT REQUIREMENTS Aspect ratio: 3:4 vertical Target dimensions: 552 × 736 pixels Duration: 15 seconds Video model: MiniMax H3 Max Do not increase the resolution beyond 552 × 736 simply to improve quality. Do not create multiple video variations. Do not create previews, alternate takes, upscales, or additional assets unless they are absolutely required to complete the requested video. CHARACTER The protagonist is Woman 1, a stylish Japanese woman in her mid-to-late 20s with model-like features and a contemporary Tokyo appearance. She has a modern dark layered hairstyle with natural movement and face-framing pieces. Her clothing has a strong current Y2K streetwear influence, but it must look fashionable and authentic rather than like a Y2K costume. Use fitted layers, subtle metallic details, a compact shoulder bag, and contemporary footwear. She should look like someone you could genuinely encounter in a fashionable Tokyo neighborhood today. Keep Woman 1's face, body, hairstyle, clothing, and accessories consistent throughout the entire video. Her skin must have realistic natural texture and visible detail. Avoid glossy skin, plastic-looking faces, CGI appearance, or an animated AI-video look. SCENE 1 — approximately 4 seconds Woman 1 walks along an active Tokyo street at night and enters a contemporary convenience store. Keep Woman 1 clearly visible while establishing the environment. The street feels active and lived-in, with pedestrians, traffic, storefront illumination, and authentic Japanese signage. Her movement into the store should feel spontaneous and natural rather than staged. SCENE 2 — approximately 7 seconds Inside the convenience store, Woman 1 selects a cold drink from a refrigerated case and walks to a modern self-checkout terminal. The store should look like an ordinary, contemporary Tokyo convenience store, not a futuristic interpretation: bright practical lighting, organized narrow aisles, refrigerated cases, shelves filled with believable Japanese products, and modern checkout technology. At the self-checkout area, Woman 1 notices an interesting local event flyer posted nearby. She pauses naturally, looks at it with genuine interest, takes out her phone, and photographs it. This is a small incidental discovery, not a dramatic moment. She then scans her drink and completes the purchase using contactless payment. SCENE 3 — approximately 4 seconds Woman 1 puts her phone away, takes her drink, and exits through the automatic doors. She steps back onto the active Tokyo night street and naturally joins the pedestrian flow. End with her continuing through the environment. She should not stop and pose for the camera. DRINK AND PACKAGING The drink is a realistic 500 mL PET bottle with an original fictional Japanese beverage brand. Give it professionally designed contemporary Japanese packaging with believable label hierarchy, restrained typography, realistic product information, and proper Japanese writing. Do not reproduce an existing beverage brand. The product should simply look convincing enough that it could genuinely be sitting in a Tokyo convenience-store refrigerator. TEXT AND ENVIRONMENTAL AUTHENTICITY Any visible Japanese signage, product packaging, event-flyer typography, or other readable environmental text must consist of real, correctly written Japanese or contextually appropriate English. Do not generate gibberish, fake Japanese characters, meaningless words, or obviously distorted AI typography. VISUAL STYLE Prioritize photorealism above everything else. The video should have believable human movement, realistic physics, natural skin texture, ordinary environmental imperfections, and authentic nighttime lighting. It should feel like something captured naturally on a high-end smartphone while following Woman 1 through Tokyo, not like a commercial, fashion shoot, CGI render, or highly polished cinematic production. Maintain the quality and realism of the source imagery throughout the video. Do not introduce an animated or synthetic appearance as the video progresses. COST CONTROL This is also a credit-usage test. Use the minimum number of generation operations necessary to produce this one finished 15-second video. Do not automatically add previews, prompt enhancement, additional generations, alternate versions, upscaling, or other credit-consuming operations. Before generating, tell me whether MiniMax H3 Max can actually produce 3:4 at 552 × 736. If it cannot output those exact dimensions, tell me the closest supported resolution and aspect ratio and wait for my approval before generating anything. Also tell me the estimated total credit cost, if available, before starting the generation. Do not generate anything until I explicitly approve the production after you answer those two questions.

Cinematic action scene on a multi lane city highway surrounded by skyscrapers A sleek aerodynamic crimson red supercar drives at high speed, followed closely by a military helicopter hovering above Suddenly the supercar seamlessly transforms into a massive heavily armored red robotic mech with glowing yellow eyes and a core reactor on its chest a terrifying, monstrous dark grey alien creature with glowing purple eyes and sharp claws drops down from an overpass bridge onto the highway, crashing into the asphalt with a cloud of dust. The red robot mech and the monstrous creature fiercely clash in hand-to-hand combat on the street, exchanging heavy punches and combat maneuvers with fiery jet boosters, leading to a massive explosion in the background as debris flies across the road.

A stressed-out office monkey in a suit furiously types on a computer, then pounds the keyboard when it freezes. He dramatically throws a banana at the screen, it explodes in sparks. He leans back, sighs, and sips coffee through a straw like a jaded CEO.

Create a google-earth style spinning of the globe and zoom in starting with the globe image in the first frame and ending with the farmhouse scene in the second frame.

1980s synthwave retro futurism, neon grid landscape, reflective chrome surfaces, scanlines and soft VHS glow, smooth stylish camera moves timed to a beat. Color grade: hot pink, electric cyan, deep purple, striped sunset orange. A colossal chrome shark with a mirror body reflecting neon light. A retro sports car with pop-up headlights driving along a coastal highway lined with palm trees. SHOT 1 (0 to 3s): Tracking shot, 35mm lens, alongside the sports car speeding down the highway, palm trees streaking past, the sunset glowing ahead. SHOT 2 (3 to 6s): Wide shot, the chrome shark breaches out of the grid ocean, neon water exploding into glowing pixels around it. SHOT 3 (6 to 9s): Close-up, 85mm lens, the shark's chrome eye and gills, the whole sunset and the passing car reflected across its body. SHOT 4 (9 to 12s): Low angle wide shot, 20mm lens, the shark arcs over the highway directly above the car, its shadow sweeping across the road, the car's headlights flipping up. SHOT 5 (12 to 15s): Extreme wide shot, 16mm lens, the shark dives back into the grid sea, sending a ripple of neon light across the whole ocean, the car driving off into the sunset. AUDIO: Engine revving, tires on asphalt, a pulsing synthwave bassline, a digital splash with sparkling arpeggios, a deep metallic whoosh, a gated drum hit on the breach, retro synth lead.

subject_definitions: is the single female character shown across all views in . Preserve her exact face, hair, body proportions, skin tone, and full outfit from the character sheet. summary: [reference generation] A continuous 6-second full-body dance performance of performing [DANCE] in a clean studio. She dances solo with clear, energetic, style-correct movement. retention_analysis: (appears in [Shot 1]): fully_preserved - face, hair, body, and outfit remain identical to the character sheet. detailed_description: Realistic live-action, cinematic lighting. [Shot 1] Full-body framing of centered. She performs clear [DANCE] with authentic footwork, arms, posture, and energy. Mostly static camera or slow push-in. Continuous motion, full body visible, no cuts, no extra people. overall_soundscape: Soft foot contact, fabric movement, and light room ambience matching the motion. non_diegetic_music: [MUSIC]

STYLE: Hyper-realistic epic dark science fantasy. Colossal naval warfare, realistic ocean physics, violent explosions, immense creature scale, stormy skies, premium blockbuster cinematic visuals. SCENE: A massive coastal fortress stands on black cliffs above a stormy ocean. Three armored battleships fire toward a mysterious presence beneath the waves. One gigantic humanoid formation made of seawater rises to defend the ocean. A single ancient armored leviathan sleeps below. SHOT 1 | 0–2s | EXTREME WIDE SHOT The coastal fortress and three battleships unleash a devastating artillery barrage into the ocean. Explosions throw enormous columns of water skyward. A gigantic shadow moves beneath the waves. SHOT 2 | 2–4s | GROUND-LEVEL LOW ANGLE ONE colossal humanoid water giant erupts beside the leading battleship. Its enormous liquid arm smashes through the ship's armored hull. The vessel breaks in half amid a violent explosion of steel and seawater. SHOT 3 | 4–6s | OVERHEAD SHOT The SAME single water giant turns toward the two remaining battleships. It sweeps one enormous arm through both vessels, throwing them together in a catastrophic collision. Both ships explode. No additional water figures appear. SHOT 4 | 6–9s | EXTREME LOW ANGLE The single water giant collapses completely back into the ocean. Immediately, the sea swells upward and the enormous armored head and shoulders of an ancient leviathan emerge from beneath the surface. Thousands of tons of water cascade from its colossal stone-like body. It dwarfs the fortress. SHOT 5 | 9–12s | DRAMATIC SIDE PROFILE SHOT The leviathan drives one gigantic armored forelimb into the coastal cliff. The entire fortress foundation fractures. With overwhelming force, the creature tears the fortress free and crushes it in its enormous grasp. Towers disintegrate as an avalanche of debris crashes into the sea. SHOT 6 | 12–15s | ULTRA-WIDE CINEMATIC SHOT At 12 seconds, the final fortress debris disappears beneath a tremendous wave. At 13 seconds, the water settles, revealing an empty, shattered coastline where the fortress once stood. The colossal leviathan remains towering offshore. At 14 seconds, the sea becomes calm and sunlight pierces the storm clouds, illuminating the victorious ancient creature. HOLD THE FINAL COMPOSITION THROUGH 15 SECONDS. No further attacks or movement. AUDIO: Aggressive orchestral percussion, thunderous naval artillery, crushing steel, violent waves and powerful brass. Monumental impact as the fortress is destroyed. At 13 seconds, percussion stops. Finish with one deep sustained orchestral chord. CONTINUITY: Exactly six cinematic setups. Only ONE humanoid water formation exists. It performs both ship attacks before collapsing completely into the ocean. The single leviathan emerges afterward, not simultaneously. Maintain consistent ship designs, coastal geography and physical scale. RESTRICTIONS: Keep proportions, style and features consistent. Ratio 16:9. No duplicate water giants, no second leviathan, no extra creatures. No dialogue, subtitles, logos or on-screen text. No famous celebrity faces, recognizable actors or public-figure likenesses. No cartoon styling, camera drift, unnecessary zoom-out or abrupt ending.

Photorealistic live-action medieval fantasy action film, cinematic 35mm photography, 16:9 widescreen, realistic human actors, natural skin textures, practical medieval costumes, physically believable creature effects, realistic gravity and motion, natural film grain, dramatic late-afternoon lighting, professional Hollywood feature-film cinematography. REFERENCE CONTROL: is the four-panel storyboard reference. Follow its four panels in exact chronological order, preserving the camera angles, framing, character positioning, action choreography, and scene transitions. Each panel represents a separate live-action movie shot. is the Red Sonja character identity reference. Preserve her exact facial appearance, long wavy copper-red hair, blue-gray eyes, natural skin texture, athletic body proportions, silver scale armor, leather straps, shoulder armor, bracers, boots, and sword with its leather sheath. Maintain her identity and costume consistently throughout all four shots. Transform the illustrated storyboard into completely photorealistic live-action cinema. The comic determines the action and camera composition, NOT the visual style. SETTING: A medieval fantasy village beneath a warm late-afternoon sky. Timber-framed buildings, thatched rooftops, distant mountains, and a wooden tavern. A massive muscular winged reptilian monster with leathery gray-purple skin, a long crocodilian head, sharp teeth, powerful arms, clawed hands, and enormous bat-like wings carries Red Sonja high above the village. 15-SECOND SHOT SEQUENCE: SHOT 1 — 00:00–00:03 TIGHT CINEMATIC CLOSE-UP — MATCH PANEL 1 Camera positioned close to Red Sonja's upper body and face, matching the first storyboard panel. The flying monster is already gripping and carrying her through the air. Red Sonja twists against its grip, her long red hair whipping violently in the wind. Her face shows fierce anger and determination. She braces herself and draws one leg upward, preparing to strike. The camera remains tight on her face and upper body, preserving the original panel's perspective. HARD CUT. SHOT 2 — 00:03–00:07 DRAMATIC LOW-ANGLE ACTION SHOT — MATCH PANEL 2 Camera positioned below the airborne characters, looking upward, matching the second storyboard panel. The monster flies with its enormous wings extended while holding Red Sonja against its body. Red Sonja violently thrusts her boot upward in a single swift, powerful kick. HER BOOT STRIKES DIRECTLY UNDER THE MONSTER'S JAW. The impact is immediate and forceful. The monster's head snaps sharply backward, its mouth opens in pain, and its wings jerk from the sudden impact. Red Sonja's expression is furious, her teeth clenched and eyes filled with rage. The monster involuntarily opens its gripping hand and releases her. Red Sonja separates completely from the creature and begins falling. Fast, realistic action choreography. Clear physical contact between boot and jaw. Natural impact physics. No slow motion. HARD CUT. SHOT 3 — 00:07–00:10 EXTREME WIDE ESTABLISHING SHOT — MATCH PANEL 3 Camera positioned far below the characters, looking upward toward the sky above the medieval village, matching the third storyboard panel. The monster remains high in the sky, recovering from the kick. Red Sonja is now falling freely, several meters beneath the creature. Her body tumbles naturally through the air, arms and legs reacting instinctively to the fall. Her red hair streams upward in the rushing wind. Below her are the steep thatched rooftops of the medieval village. She rapidly descends toward the roof of a large wooden tavern. Maintain the wide composition, clearly showing the separation between the monster, the falling woman, and the buildings below. HARD CUT. SHOT 4 — 00:10–00:15 DYNAMIC INTERIOR WIDE SHOT — MATCH PAN

Aggressive tracking shot through a gigantic abandoned industrial facility. The camera races between steel columns while electrical transformers overload one after another, sending violent arcs of electricity across the environment. Sparks shower toward the lens. A huge turbine begins spinning at impossible speed before exploding apart. Thousands of glowing metallic fragments freeze in midair for one dramatic beat, then slam together with brutal force to construct the word “POWER” in enormous heavy steel letters. Blue-white electricity crawls across every letter as the camera performs a fast 180-degree orbit. Dark industrial blockbuster aesthetic, practical sparks, volumetric smoke, realistic physics.

Ultra-cinematic action title sequence opening on a colossal reinforced wall submerged in darkness, its surface covered in pressure fractures and condensation. A deep impact hits from behind, deforming the structure outward before a second strike tears through the center and launches slabs of steel, concrete and shattered glass toward the camera. The lens pulls back through the expanding debris field as compressed air, sparks and dust create violent pressure rings. A powerful reverse shockwave suddenly catches every fragment and drags them back toward the rupture point. The pieces collide, rotate and interlock with brutal mechanical force, progressively constructing the monumental word "BREACH" from scorched structural steel and fractured composite material. The camera sweeps rapidly across the face of the letters as internal orange light leaks through the seams, then a final pressure pulse punches straight through the center of the title, blowing dust, fragments and glowing particles toward the lens while the word remains readable for one hard hero beat. Premium destruction physics, tactical blockbuster energy, sharp cinematic typography.

Create an 18-second, 16:9 realistic ASMR skincare video. A beautiful young woman with long brunette hair wears soft pink pajamas in a cozy pink bedroom at night. Use one completely fixed iPhone-style camera. No camera movement, zoom, dialogue, voiceover, or BGM. Only natural ASMR sounds. 0–3s: Open on the finished look. She wears a glossy translucent sheet mask fully covered with colorful rhinestones, pearls, hearts, stars, flowers, and crystals. She slowly turns her face so the gems catch the light. 3–5s: Cut to her bare face. She picks up and unfolds a plain wet sheet mask. 5–8s: She carefully applies the mask and smooths it over her forehead, cheeks, and chin. 8–13s: She brings in a tray of colorful rhinestones and presses them onto the mask one by one. Focus on satisfying gem clicks, fingertip taps, and soft pressing sounds. 13–16s: She places the final heart-shaped gem on her cheek, then slowly looks into the camera and tilts her head to reveal the finished jeweled mask. 16–18s: She taps three rhinestones on her cheek: tap, tap, tap, then gives a small satisfied smile. On-screen text: “skincare but make it fun ✨” “would you try it? 💎🎀” Style: intimate, realistic, pinkcore, tactile, satisfying ASMR. Natural skin texture, realistic wet sheet mask, clear varied rhinestones, no CGI sparkle, no excessive retouching, no dramatic transitions.

pours a wide stream of wrapped Halloween candy into a plastic jack-o'-lantern bucket until it overflows

green light spreads from the bend out to both ends and then stays lit, steady and bright

falls open like the petals of a flower around a lit candle standing where the core was

Cinematic 10-second video photorealistic, smooth camera movement, warm natural sunlight.Wide establishing shot of a sunny orange orchard, camera slowly pushes forward down a dirt path between rows of lush green trees heavy with ripe bright oranges.Close-up of two large juicy oranges hanging on a branch with water droplets, then a fair-skinned hand gently picks one orange.Cut to a woven basket filled with fresh oranges and green leaves, hand places the picked orange into the basket. Soft bokeh orchard background.Transition to bright modern kitchen. Hands squeeze a halved orange, juice streams into a glass bowl. Then juice is poured through a strainer into a measuring cup.Hands open a cute pink-to-purple gradient stainless steel tumbler with handle, cute whale, starfish jellyfish charms and purple monster keychain Ice cubes are dropped in, fresh orange juice is poured over the ice, two orange slices are added, lid is closed. Final hero shot of the filled tumbler standing on the kitchen counter with oranges and a glass of juice around it.

Step 1 — 0 to 3 Seconds: A beautiful woman wearing an elegant white flowing gown with delicate lace sleeves rides on the back of a majestic red dragon. The dragon flies gracefully above a snow-covered forest surrounded by towering pine trees and majestic snowy mountains. Its enormous wings move naturally as it glides through the cold winter air. The camera captures a cinematic side view of the woman and the dragon in flight. Step 2 — 3 to 5 Seconds: The red dragon begins descending toward a snowy clearing in the forest. The woman remains seated on its back, holding on gently as the dragon lowers its wings. Several wild deer and other animals are visible on the snowy ground below. The camera follows the dragon's smooth descent. Step 3 — 5 to 8 Seconds: The dragon lands gracefully in the snowy clearing, spreading its enormous wings. The woman carefully dismounts and walks toward a beautiful brown deer with large antlers. The dragon stands behind her, watching calmly. The camera captures the magical winter environment and the realistic interaction between the characters. Step 4 — 8 to 12 Seconds: The woman gently embraces the deer and lovingly strokes its head. The deer calmly approaches her and rests its head near her shoulder. She smiles warmly while the majestic red dragon watches from behind. The camera moves closer to capture the emotional connection between the woman and the deer, with soft snowfall and cinematic lighting. Step 5 — 12 to 15 Seconds: The woman continues embracing the deer as the red dragon raises its head and opens its mouth, releasing a powerful stream of bright orange fire into the snowy clearing. The fire illuminates the surrounding snow, the dragon's red scales, and the woman's white gown. The deer remains beside her as the scene ends with a dramatic cinematic fantasy atmosphere. Visual Style: Ultra-realistic fantasy cinematography, detailed dragon scales, realistic deer fur, natural character movements, flowing white dress, cinematic camera motion, realistic fire simulation, snowy mountain environment, soft atmospheric lighting, high detail, vertical 9:16 format, consistent character appearance, no text, no watermark.

THE WRONG TABLE — EPISODE 1 — CLIFFHANGER FREEZE-FRAME ENDING CRITICAL: NO ON-SCREEN TEXT OR SUBTITLES. No captions, names, titles, logos, or watermarks. REFERENCE IMAGES: — CHEN YICHEN: Use the provided restaurant image of the male lead as the ONLY visual reference for the male character. Strictly preserve his original facial appearance, hairstyle, navy-blue suit jacket, light-blue shirt, calm expression with a slight smile, and the original restaurant background, lighting, and camera angle. Do not redesign the character. — SHEN RUOXI: Use the provided shocked close-up image of the female lead as the ONLY visual reference for the female character. Strictly preserve her original facial appearance, long hair, gold hoop earrings, dark-blue silk blouse, widened eyes, slightly parted lips, and the original background, lighting, and camera angle. Do not redesign the character. integrated_multimodal_description: 9:16 vertical format, modern Chinese urban romantic short drama, cinematic photorealistic live-action visuals, classic Chinese short-drama suspenseful freeze-frame ending. The entire sequence uses ONLY the two original character images from and . Do not create new camera shots or change the characters' faces, expressions, clothing, compositions, or backgrounds. Create suspense through frozen images, rapid cuts, white flashes, soft misty light blooms, and fade-out effects. The character appearance order must strictly be: Shen Ruoxi → Chen Yichen → Shen Ruoxi. [Shot 1] At 00:00.000, immediately show the shocked facial close-up of Shen Ruoxi from . Her eyes are widened and her lips are slightly parted, preserving the exact shocked expression from the original image. The image freezes completely. No blinking, no head movement, and no lip movement. A soft, misty white glow gradually appears around the edges of the frame, followed by a rapid white flash at 00:00.850. [Shot 2] At 00:01.000, immediately after the white flash, hard cut to the original restaurant image of Chen Yichen from . Preserve his original seated posture, subtle smile, clothing, and background. The image remains completely frozen, creating a strong contrast with the female lead's shocked expression. Hold briefly, then introduce another rapid white flash. [Shot 3] At 00:02.000, immediately after the flash, cut back to the original shocked close-up of Shen Ruoxi from . The image instantly freezes, preserving her exact original facial expression and composition. Add a subtle, soft white misty glow around the edges, a hazy light bloom, and a brief ghosted afterimage overlay to create the suspenseful atmosphere of a classic Chinese short-drama cliffhanger ending. Her face must remain clearly visible at all times and must not be obscured by the white glow. [Shot 4] At 00:03.000, continue holding the completely frozen close-up of Shen Ruoxi. The soft white glow briefly intensifies, then gradually diminishes. During the final half-second, smoothly fade the image to pure black. End at 00:04.000. VISUAL CONTINUITY: Strictly preserve the characters' identities, original facial structures, hairstyles, clothing, natural skin textures, lighting, backgrounds, and camera compositions from the two reference images. All character images must remain completely motionless freeze-frames. Only editing transitions and subtle optical effects are permitted. Do not generate new actions, change facial expressions, animate lips, introduce additional characters, change locations, rotate the camera, zoom in, or zoom out. No subtitles, no text, no graphic titles, and no watermarks. overall_soundscape: Begin with a short, deep cinematic suspense impact sound. Each rapid white flash is accompanied by a subtle, clean air-whoosh sound. During the final freeze-frame, maintain a brief lingering low-frequency suspense resonance that quickly disappears as the image fades out, ending in complete silence. No character dialogue, no voiceover, and no restaurant conversations. non_diegetic_music: None. Use only brief suspense impact sound effects and white-flash transition sounds. No continuous background music.

A young woman walking slowly down a quiet forest path at dusk, trees tall on both sides, soft fog drifting low across the ground. A cold breeze moves her hair and rustles leaves. She holds her jacket closed, tense but trying to stay calm. Sound: soft wind through trees, distant insects, occasional twig snap, quiet footsteps on dirt and leaves. Music: minimal cinematic tension bed, low sustained drones, very faint pulsing bass every few seconds, subtle high ambient tones that slowly rise, no big hits or jump-scare stingers. She quietly says, “I don’t know why this place still scares me…” She pauses and listens to the silence. “My mom used to say nothing bad happens out here.” She takes a few slow steps forward, breathing softly. “But every time I come back… it feels like someone’s watching.” The fog thickens slightly as shadows stretch between the trees. She turns slowly, heart racing, but sees nothing. She whispers, “This is stupid… I’m just tired.” She exhales and keeps walking as the forest grows darker, the music slowly tightening as the clip ends.

High-speed chase shot through an abandoned post-apocalyptic city in daylight. The camera races backward ahead of a lone runner while a gigantic dust tornado destroys the avenue behind them. Cars roll through the air, windows explode and debris fills the sky. The tornado suddenly tears an entire building facade apart. Thousands of concrete, steel and glass fragments are sucked upward, rotating violently inside the vortex. From above, the debris unexpectedly aligns into enormous floating letters spelling “SURVIVE.” The runner passes underneath as the tornado instantly destroys the word again. Massive disaster-movie VFX, intense speed, photorealistic debris and atmospheric destruction.

STYLE: Hyper-realistic epic science fiction. Immense cosmic scale, colossal ancient alien machinery, brutal orbital warfare, devastating explosions, realistic metallic debris, blinding plasma energy, cinematic lighting, premium blockbuster visual effects. Dark steel, electric blue and ominous crimson. Relentless action with a monumental final resolution. SCENE: High above Earth, a gigantic artificial black sphere surrounded by rotating mechanical rings eclipses the Sun. A fleet of angular human battleships attacks the mysterious object. The Moon hangs in the background. SHOT 1 | 0–2s | EXTREME WIDE SHOT Hundreds of human warships surge toward the enormous black sphere, firing concentrated blue plasma beams and missiles. Explosions race across its armored surface. The surrounding mechanical rings begin trembling violently. SHOT 2 | 2–4s | LOW ANGLE TRACKING SHOT A gigantic human dreadnought accelerates through burning wreckage and fires its main cannon. The enormous energy beam punches through the sphere's armored shell. A violent explosion sends colossal metal fragments spinning outward. SHOT 3 | 4–6s | OVERHEAD SHOT Inside the shattered shell, gigantic interlocking gears and ancient mechanical rings awaken. Crimson energy races through their channels. The entire mechanism begins rotating violently, pulling nearby wreckage inward. SHOT 4 | 6–9s | SIDE PROFILE SHOT A colossal red energy beam erupts from the awakened machine and strikes the Moon. Enormous cracks split its surface. The Moon fractures into gigantic sections, sending mountains of lunar debris outward. Nearby warships are hurled aside. SHOT 5 | 9–12s | DRAMATIC REAR ANGLE The black sphere unfolds six colossal mechanical arms. They seize the fractured lunar sections and pull them into a blazing circular mechanism. The captured debris is crushed into glowing material. The machine redirects a gigantic ring of energy toward Earth, initiating a rapidly expanding transformation across its surface. SHOT 6 | 12–15s | ULTRA-WIDE COSMIC SHOT The transformation spreads across Earth. Oceans and continents disappear beneath enormous interlocking metallic plates. Brilliant crimson energy races through a vast planetary grid. At 13.5 seconds, the transformation completes. Earth has become a gigantic mechanical planet covered in glowing geometric structures. The black sun's machinery locks into place above it. At 14 seconds, all moving mechanisms stop. The energy beams extinguish. A single surviving human warship hangs motionless in the foreground, dwarfed by the completed artificial world. HOLD THIS FINAL COMPOSITION UNTIL THE VIDEO ENDS. AUDIO: Aggressive orchestral percussion, massive low brass, powerful plasma cannons, deep metallic impacts and thunderous explosions. Music builds relentlessly through the lunar destruction and planetary transformation. At 13.5 seconds, deliver one immense final orchestral impact. The music resolves into a deep sustained note. No additional explosions or musical build after the resolution. CAMERA AND CONTINUITY: Six distinct cinematic setups. Maintain consistent human ship designs, sphere geometry, Moon and Earth positions. One continuous orbital battle. Strong physical cause and effect. Clear progression from fleet attack to machine awakening, lunar destruction, mechanical expansion and final planetary transformation. Fast cinematic cuts, dramatic scale, readable destruction. No camera drift. RESTRICTIONS: Keep proportions, style and features consistent. Ratio 16:9. No dialogue, subtitles, logos or on-screen text. No famous celebrity faces, recognizable actors or public-figure likenesses. No cartoon styling. No additional planets or alien creatures. No abstract ending. No unnecessary zoom-out.

Dark cinematic VFX title sequence beginning with a single drop of glossy black liquid falling onto polished metal. Instead of splashing normally, the fluid violently spreads outward in branching organic tendrils. The camera tracks the liquid at macro scale as it races across the surface, splitting, reconnecting and forming increasingly complex structures. Thousands of black droplets suddenly leap upward against gravity and spiral into a living fluid vortex. The viscous material stretches into sharp ribbons and rapidly constructs the word "VENON", every letter constantly moving beneath a wet reflective surface as if the typography itself were alive. The camera performs a sinister close orbit while black tendrils whip between the letters and droplets float around the lens. The title suddenly loses structural stability and violently liquefies, exploding toward the camera as an enormous black fluid splash. Hyper-realistic viscosity, organic procedural motion, glossy reflections, dark premium theatrical VFX.

【PROJECT】 MiniMax-H3, 15 seconds, 16:9, one photoreal live-action cinematic musical-theater pop-ballad test. Use @Audio1 as the master timing source: female resonant vocal, soft opening, orchestral rise, brief dip, explosive final climax, clean ending. Synchronize cuts, breath, gaze, gestures and type to the actual audio. Do not invent a different song or singer. 【REFERENCES】 @Image1 locks the singer’s recognizable face, long wavy hair, expressive performance, sleeveless black outfit, silver jewelry and proportions. @Audio1 is the uploaded 15-second song; preserve its vocal phrasing, sustained notes, dynamic build and final repeated climax. One female performer throughout. 【CREATIVE】 A solitary singer moves through one connected theatrical world: shadowed backstage corridor, velvet stage, intimate spotlight. Longing becomes confrontation and then release. Photoreal skin and hair, realistic fabric, brass microphone, velvet, light haze, polished studio image, cinematic orchestral-pop atmosphere, no film grain. 【PERFORMANCE / MUSIC】 Show breath preparation, controlled jaw and throat tension, expressive eyes, subtle head lifts, one hand near the sternum, and gradual expansion from vulnerability to chest-led power. Mouth and facial articulation match the uploaded vocal; no generic silent lip motion. No dialogue, rap, crowd vocals, comedy or dance troupe. Keep natural breath gaps. Final intensity arrives only on the last sustained phrase. 【VISUAL / TYPE】 Black velvet, warm ivory spotlight, deep burgundy fabric, muted metallic silver jewelry and microphone. Ultra-black condensed grotesk, ALL CAPS, tight tracking; secondary narrow sans used sparingly. Editorial stage typography, not subtitles or karaoke. Use paper-edge reveal, restrained double-image echo, halftone offset and one brief black-frame word. Words must be real, short, readable before transforming; never gibberish or logos. 【SHOT 1 — 0.0–1.8s】 Medium close-up, slow lateral track through a backstage doorway. The singer enters from shadow, shoulders forward, hand near sternum, preparing breath. Action: threshold entry. Space: backstage corridor. Hair and gown edge follow one beat late. Typography: “LONGING”; Layer A large condensed hero on wall, Layer B smaller silver print at doorway seam, Layer C thin foreground letters partly hidden by her shoulder. Rhythm / Cut: opening vocal reveals the word; paper edge wipes into her face. 【SHOT 2 — 1.8–3.6s】 Close-up tracking beside her as she turns toward the microphone. Action: shoulder-led half-turn. Space: corridor opening to stage. Eyes lift toward the audience; breath shows in chest and throat. Typography: “STAY”; Layer A large word behind head, Layer B narrow letters on glass reflection, Layer C cropped lower-frame letter clear of eyes and mouth. Rhythm / Cut: vocal consonant hard-cuts profile to three-quarter view. 【SHOT 3 — 3.6–5.5s】 Medium close-up, slow push to the vintage microphone. Action: grounded reach and contact. Space: stage threshold. She grips the stand and leans into the emotional phrase; eyes close on the sustained note, then reopen. Typography: “ENOUGH”; Layer A hero on velvet panel, Layer B halftone photocopy offset behind mic, Layer C thin type reflected in brass grille. Rhythm / Cut: hand contact lands on beat with restrained metallic resonance. 【SHOT 4 — 5.5–7.3s】 Medium shot, quarter arc as the curtain opens. Action: curtain release and chest expansion. Space: front stage. She releases the stand, opens her ribcage into the rise; gown follows late and settles. Typography: “HEAR ME”; Layer A split hero across curtain, Layer B offset print registration, Layer C torn-paper foreground briefly masks lower torso. Rhythm / Cut: orchestral lift opens curtain; words remain readable before folding away. 【SHOT 5 — 7.3–9.1s】 Extreme close-up on eyes, cheek and upper lip, nearly locked with breathing drift. Action: still-point eye performance. Space: center spotlight. Music dips; eyes move from floor to camera, uncertainty becoming resolve. Typography: “WAIT”; Layer A heavy word beside eye line, Layer B faint mirrored print behind it, Layer C no obstruction over face. Rhythm / Cut: half-second visual suspension, no shock zoom or lyric subtitles. 【SHOT 6 — 9.1–11.0s】 Medium close-up, camera pulls back as she takes one deliberate step into full light. Action: step-forward power transfer. Space: center stage. Rear foot pushes, shoulders settle, chin rises; microphone stays behind. Typography: “I RISE”; Layer A projected hero backdrop, Layer B narrow print fragments trail one step, Layer C foreground black letter exits frame. Rhythm / Cut: returning pulse triggers step and brief double-image echo, resolving to one sharp performer. 【SHOT 7 — 11.0–13.0s】 Medium to close-up, controlled tilt as strings widen. Action: upward head lift, open-arm release, then pivot freeze. Space: downstage spotlight. She lifts her head and opens one arm from the sternum; hair and gown sweep and settle. She turns three-quarter to frontal and freezes on the high vocal. Typography: “FOR ME”; Layer A large hero on black stage plate, Layer B silver registration offset, Layer C narrow strip only across lower frame. Rhythm / Cut: vocal peak triggers the pivot; hold one breath with mouth clear. 【SHOT 8 — 13.0–15.0s】 Extreme frontal close-up, slow final push into face and microphone grille. Action: held-note concentration and breath release. Space: intimate center-stage afterglow. Eyes glisten; pleading becomes self-possession; jaw and mouth match the final sustained vocal. Typography: “ENOUGH” then “FOR ME”; Layer A same condensed hero returns clearly, Layer B faint silver duplicate under it, Layer C no face obstruction. Rhythm / Cut: final p

Create a 15-second, single continuous cinematic 3D animated video. A young man is busy working at his computer when his adorable orange cat approaches and lovingly rubs against his face. He casually pushes the cat away without looking. Big mistake! The cat notices a water cup beside the computer. With a mischievous look, it swipes the cup, spilling water all over the keyboard. The owner turns around, furious. But the cat instantly switches to its most innocent, pitiful expression. The owner's anger melts. He picks up a cloth and starts cleaning. Meanwhile, the cat slowly develops a smug smile. In the final moment, the owner picks up the cat and gently pats its head. The cat looks delighted and says: "He actually loves me!" Visual Style: Pixar-inspired 3D animation, expressive character movements, warm golden-hour lighting, soft clay-like textures, cinematic depth of field and hilarious facial expressions. Settings: 15 seconds | 16:9 | 768P | MiniMax-H3-Max

Create a 15-second, 16:9, 24fps fashion MV based on the attached single-character reference image. Character Lock Use only ONE young Korean-style female character throughout the entire video. Strictly preserve her: face and facial proportions skin tone age hairstyle, bangs, and hair color body proportions overall identity She must remain the exact same person in every shot. No face drift, no second character, no background people, no Westernized facial features, no random hairstyle or body changes. Overall vibe: Korean beauty / Seoul fashion girl / sweet-cool / polished / light Y2K / cute but premium. Visual Style PINK PATTERN COLLAGE POP × KOREAN FASHION EDITORIAL × FASHION MV Create a polished Korean fashion-magazine world using: pink graphic patterns, checkerboards, stripes, grids, photo-booth frames, geometric color blocks, collage cards, editorial typography, stickers, and clean 2D graphic elements. Keep the character photorealistic. Graphics should feel sharp, stylish, clean, premium, and professionally art-directed — not handmade or childish. Main palette: cream pink, soft pink, rose pink, peach pink, cream white, silver gray, with small black accents. Wardrobe Use Korean sweet-cool / light Y2K fashion. Possible looks: light denim, fitted white or pink tops, mini skirts, small bags, jewelry, platform boots, fashion accessories. Minor outfit changes are allowed, but the character identity must remain consistent. Motion This is a real fashion MV, not a slideshow. Each shot should include: starting pose → one clear fashion movement → new ending pose Use subtle editorial actions: walking half a step, turning back, adjusting hair, touching accessories, holding a bag, sitting, crouching, shifting weight, looking into camera, reaching slightly toward the lens. No dancing or exaggerated acting. Use a mix of: close-up, medium, full-body, low-angle, detail shots, smooth lateral movement, subtle push-in/pull-out, and editorial crop transitions. Typography Use large fashion typography throughout. Main title: NAME Supporting text may include: SEOUL POP / SOFT CHIC / LOVELY MOOD / TRENDY LOOK / KOREAN GIRL / FASHION CUT Typography can be oversized, repeated, vertical, diagonal, or placed inside graphic frames. Keep all text readable and never cover the face. Timeline 0:00–0:01.6 | LOOK 01 Close-up or medium shot. She raises her chin and looks into camera. Large pink collage graphics and bold title typography. 0:01.6–0:03.2 | LOOK 02 Full or medium shot. Half-step forward, touch skirt or bag, then turn back. Stripes, dots, and geometric blocks. 0:03.2–0:04.8 | LOOK 03 Sitting or crouching pose. Hand near face. Checkerboard, grid, and photo-frame graphics. 0:04.8–0:06.4 | LOOK 04 Medium close-up. Turn slightly, adjust hair, look back at camera. Strong magazine-style layout. 0:06.4–0:08.0 | LOOK 05 Low-angle or wide shot. Lean or reach slightly toward camera. Graphic bands sweep across frame. 0:08.0–0:09.6 | LOOK 06 Beauty close-up. Lower head, then look up at camera. Cleaner background and repeated editorial text. 0:09.6–0:11.2 | LOOK 07 Full-body fashion pose with subtle lateral camera movement and geometric magazine composition. 0:11.2–0:12.8 | LOOK 08 Medium shot. Adjust hair or accessories. Large color blocks and enlarged patterns. 0:12.8–0:15.0 | FINAL HERO Strong final cover pose. Previous graphics and typography reassemble into one polished Korean fashion-magazine composition. Hold the final frame for 0.5 seconds. Music Use a stylish sweet-cool K-pop / fashion-pop instrumental. 115–125 BPM, no lead vocals, clear beat points for transitions. Subtle whoosh, click, paper slide, and fabric swish sound effects are allowed. Avoid No multiple people, identity drift, face changes, stiff motion, static slideshow, messy backgrounds, unreadable text, text covering the face, childish collage, cheap filters, logos, watermarks, subtitles, exaggerated expressions, or excessive dancing. Final result: Korean fashion girl × Pink Pattern Collage Pop × Editorial Fashion MV

lifts a little and slides slowly away to the right, as if something underneath is walking

a crack opens in its side by itself, and warm molten chocolate oozes out and pools slowly around it

creaks open by itself, slowly, and thick fog rolls out across the floor toward the camera

15-second continuous one-shot in 1990s Japanese retro horror 2D cel animation style, 16:9 widescreen, deep indigo night with cyan-blue shadows and pink-magenta side lighting. Clean frame with no subtitles, no on-screen text, and no Japanese script or lettering. 00:00–00:04 (OPENING TWO-SHOT): In the narrow night alley, the curly-haired man with rectangular glasses leans casually against the blue plaster wall on the left with both hands in his pockets. The woman with long wavy platinum-blonde hair in the dark school blazer and plaid skirt walks slowly past him in the foreground. The man tilts his head with a casual smirk and speaks in Japanese: "Motto waratta hou ga ii yo." The blonde woman stops walking. 00:04–00:09 (SMOOTH CAMERA PUSH-IN TO CLOSE-UP & MONSTER JAW REVEAL): The woman turns her face directly toward the camera as the camera smoothly pushes in to a tight frontal close-up of her face under cyan-blue and pink-magenta split lighting. She stares with calm, half-lidded eyes and forms a tiny, polite closed-mouth smile. Suddenly, her lower jaw violently unhinges and elongates vertically into a terrifying, impossibly wide black supernatural mouth lined with long needle-sharp fangs while her eyes widen in an unblinking stare. 00:09–00:15 (CAMERA PULL-BACK TO WIDE ALLEY & ENDING): The camera rapidly pulls back down the dark alley as the streetlight flickers. The blonde woman's dark silhouette lunges at the man, dragging him down into the dark shadows until the man completely disappears. She slowly rises to stand upright alone in the middle of the pink-purple lit alley, pauses, and calmly walks away down the alley with her long wavy blonde hair trailing behind her, matching the last frame. Sound: quiet nighttime crickets and footsteps (00:00–00:02), a young man casually speaking in Japanese saying "Motto waratta hou ga ii yo" (00:02–00:04), an eerie rising flute drone (00:04–00:06), a piercing female banshee shriek with a sharp horror sting (00:06–00:09), a brief muffled gasp and scuffle (00:09–00:11), and a deep ominous dark ambient bass drone (00:11–00:15).

Create a 15-second ultra-realistic cinematic travel vlog featuring a beautiful girl with long, straight black hair wearing an elegant sleeveless cream-colored dress and carrying a small white shoulder bag. The video begins with the girl sitting inside a modern car with a panoramic sunroof, smiling at the camera as she arrives at a breathtaking lakeside destination surrounded by lush green mountains and sparkling blue water. She gently adjusts her hair while recording herself in selfie-vlog style. Next, she walks along a charming stone pathway surrounded by traditional European houses decorated with colorful flowers, turning toward the camera with a warm smile and a gentle wave as the camera smoothly follows her movements. She then arrives at a beautiful outdoor lakeside café with wooden tables, cream-colored umbrellas, and flower decorations, where she sits down and enjoys a cold iced coffee while admiring the peaceful scenery. She records another selfie, smiles naturally, and turns the camera to capture the panoramic view of the crystal-clear lake, green mountains, and elegant café terrace. In the final moments, she stands beside a wooden lakeside railing during golden hour, smiling at the camera before gently turning toward the sunset reflecting on the water, ending with a breathtaking cinematic view. Use realistic facial expressions, natural skin texture, consistent facial features and clothing, gentle hair movement, smooth camera transitions, authentic handheld vlog angles, beautiful European scenery, warm cinematic lighting, and premium 4K quality. Vertical 9:16 format, exactly 15 seconds long, with no subtitles, no text overlays, and no watermarks.

Create a 10-second, 16:9 photorealistic UGC ad for Ensure High Protein Creamy Caramel Nutrition Shake using the exact reference product image. Character: A beautiful 25-year-old woman with long dark brown hair, natural makeup, and a cream top in a bright modern kitchen. Shot 1 (0–2s): She smiles and presents the bottle. Dialogue: “Looking for a delicious protein boost? Check this out!” Shot 2 (2–4s): Bottle close-up, label facing camera. Dialogue: “Sixteen grams of protein in every bottle!” Shot 3 (4–5s): She naturally opens the bottle with realistic sound. Shot 4 (5–7s): She takes a sip and smiles. Dialogue: “And that creamy caramel flavor? So good!” Shot 5 (7–8s): She holds the bottle beside her face with an approving smile. Shot 6 (8–9s): Premium product hero shot on the kitchen counter, girl blurred in the background. Shot 7 (9–9.5s): On-screen text: “16g PROTEIN | 160 CALORIES | CREAMY CARAMEL.” Shot 8 (9.5–10s): Clean final product shot with soft highlights. Style: Authentic UGC, cinematic lighting, natural expressions, smooth transitions, realistic skin, accurate lip-sync. Product: Preserve the exact bottle shape, blue cap, branding, white-and-purple packaging, and Creamy Caramel variant. Audio: Natural female American English voice, soft upbeat music, realistic opening and sipping sounds. Output: Exactly 10 seconds, 16:9 landscape, 1920×1080. Consistent character and outfit, realistic hands, no distorted packaging, flickering, misspelled text, or artificial expressions.

About MiniMax H3

MiniMax H3 is the flagship of MiniMax's Hailuo video line, and its defining trait is that it is genuinely multimodal on the way in and complete on the way out. You can start a generation from a text description, a reference image, an existing video clip, an audio track, or a combination of them — and the result arrives with native stereo sound generated alongside the picture, not dubbed on afterwards.

That combination makes it the model to reach for when the clip has to feel finished: ad spots where the ambience sells the product, story scenes where footsteps and room tone carry the mood, image-to-video shots where a still photograph needs to come alive with believable motion and sound. It inherits the physical realism the Hailuo line is known for, and follows camera direction — dolly, orbit, handheld — with unusual obedience.

If you have used Hailuo 02, H3 will feel familiar but less forgiving of vagueness. The older line rewards a short scene description and fills in the rest; H3 gives you six or seven independent levers — subject, action, setting, camera, light, sound, and the input modality itself — and hands each unpulled lever back to its own defaults. More control is only an advantage if you actually spend it.

The practical consequence is that a MiniMax H3 prompt is closer to a shot list than to a caption. A prompt that reads "a woman walking in the rain, cinematic" gets you a competent, anonymous clip. A prompt that names the lens, the direction of the walk, the camera's move relative to her, the quality of the light, and what the rain sounds like on what surface gets you a shot you could actually cut into something.

Because H3 accepts so many kinds of input, the prompt is what tells it which job to do — and vague prompts waste exactly the control that makes it strong. The gallery above collects tested MiniMax H3 prompts with their real generated videos, so you can copy a structure that already works and swap in your own subject, motion, and soundscape.

Everything on this page is free to copy. There is no sign-up, no paywall, and no watermark on the prompt text — take the structure, change the nouns, and run it wherever you have access to the model.

How to write MiniMax H3 prompts

1Structure the prompt as subject → action → setting → camera → light and mood → sound. H3 reads all six layers; leaving one out means the model decides it for you.
2Write it as a shot, not as a caption. "Cinematic" and "high quality" are the two least useful words you can spend — replace them with the specific thing you mean: the lens, the grade, the time of day, the weather.
3Use real camera vocabulary: "slow dolly-in", "handheld tracking shot", "orbit around the subject". H3 follows named camera moves far more precisely than vague phrases like "dynamic shot".
4For slow motion, name it as a property of the footage rather than of the subject: "filmed in slow motion, water droplets hanging in the air" works; "she moves slowly" just makes her walk slowly at normal speed. If you want a speed change mid-clip, say where it happens — "real time, then ramps into slow motion as the glass hits the floor".
5Direct the audio explicitly — sound is native, so it listens: "rain on a tin roof, distant thunder, no music" or "upbeat synth track builds under the voiceover". Unprompted audio defaults to generic ambience.
6Name the surface, not just the sound. "Footsteps" is generic; "boots on wet gravel" gives the model the material to synthesize. The same trick works for everything from "rain" to "engine".
7If you want no sound at all, say so — "no music, no dialogue, ambient room tone only". Silence is a direction H3 will follow, but only if you give it.
8When starting from a reference image, describe only the motion and the camera — the image already fixes subject, style, and framing, and re-describing them invites drift.
9Give each generation one continuous action beat. A clip that tries to cover three story moments blurs all of them; chain separate generations when the story needs more beats.
10Freeze your look in a reusable clause — film stock, color grade, lens character — and keep it identical across a campaign so every clip cuts together.
11To hold a character across shots, feed the same reference image into each generation rather than re-describing the person. Text descriptions of faces drift between runs; an image does not.
12When a result misses, change one layer and regenerate rather than rewriting the whole prompt. Rewriting everything makes it impossible to learn which clause was actually responsible.

Frequently asked questions

How do I write a good MiniMax H3 prompt?

Write it as a shot list, in this order: subject, action, setting, camera move, light and mood, sound. H3 reads each of those as a separate instruction, so every layer you leave blank is a layer the model fills in with its own default. Keep one continuous action per generation, use real camera vocabulary for the move, and name the audio explicitly. Every prompt in the gallery above follows that shape — the fastest way to learn it is to copy one and swap in your own subject.

Is MiniMax H3 free to use?

The prompts on this page are free — copy any of them with no account and no paywall. Access to the model itself runs through MiniMax's Hailuo AI app and the MiniMax open platform API, and their current plans and free-tier terms are set on MiniMax's side, so check there for what a run costs today.

What is MiniMax H3 best at?

Clips that need to arrive finished: video and stereo audio generated together, from whatever input you have — text, a still image, an existing clip, or a soundtrack. Its physical realism and precise camera-following make it a strong pick for ads, cinematic single takes, and image-to-video work.

Does MiniMax H3 generate sound?

Yes — stereo sound is generated natively with the picture, not added afterwards. You get the most out of it by directing the mix in the prompt: name the ambience, the effects, and whether you want music. If you say nothing, you get generic ambience.

How do I get slow motion out of MiniMax H3?

Describe slow motion as a property of the footage, not of the subject. "Filmed in slow motion" or "high-speed camera, slow-motion playback" changes how the clip is shot; "she moves slowly" only makes the subject sluggish at normal speed. Pair it with a detail that only reads in slow motion — spray, dust, fabric, hair — so the effect has something to show, and name the moment if you want the speed to change mid-shot.

Can I start from an image or an audio track instead of text?

Yes. H3 is multimodal on input: a reference image can define the subject and framing, a clip can be extended or restyled, and an audio track can drive the scene. When an image carries the look, keep the text prompt focused on motion and camera only.

How is MiniMax H3 different from Hailuo 02?

They are two generations of the same Hailuo video line. Hailuo 02 takes a short scene description and fills in most of the decisions for you. H3 is multimodal on input, generates its stereo audio natively with the picture, and exposes far more of those decisions as things you control — which means it rewards a detailed prompt and punishes a lazy one more than the older line did.

Why did MiniMax H3 ignore part of my prompt?

Usually because the prompt asked for more than one thing at once. H3 handles a single continuous action beat well; a prompt covering three story moments gets averaged into a blur. The other common cause is re-describing what a reference image already fixed — when the image says one thing and the text says another, the result drifts. Cut the prompt to one beat, change one layer at a time, and regenerate.

Where can I use these MiniMax H3 prompts?

H3 is available through MiniMax's Hailuo AI app, and developers can access it via the MiniMax open platform API. Copy any prompt from this page, paste it there, and swap in your own subject, camera move, or soundscape.

Related models

Hailuo 02

MiniMax's Hailuo video model with standout physical realism.

Speech 2.5

MiniMax's multilingual voice synthesis model.

Seedance 2.5

Seedance 2.5 is a next-gen AI video model that generates continuous, synchronized 30-second 4K clips from a single prompt.

Seedance 2.0

ByteDance's #1-ranked video model — multi-shot direction with native lip-sync and spatial audio.

Browse AI Video Prompts All models