1つのMiniMax H3ワークスペースで、テキスト・画像・動画・音声を組み合わせた2K動画を制作できます。ネイティブのステレオ音声と自然言語による編集で、最大15秒のクリップを生成できます。
Switch between models from MiniMax, Hailuo, Seedance, Google Veo, Wan, Grok, Kling, and PixVerse in the same generator.
マルチモーダルなリファレンス、ネイティブ オーディオ、モーション転送、正確なクリエイティブ ディレクションによって形成された MiniMax H3 の例をご覧ください。
@image1's female star as the main subject, reference the camera movement method of @video1 for rhythmic push-pull-pan-tilt movements. The female star's movements also reference the dance movements of the woman in @video1, performing energetically on stage.
例
@image1 [Camera View] Official television sports broadcast feed. High-angle professional zoom lens, smooth tripod-mounted tracking, no handheld jitter. The video features a permanent digital scoreboard overlay in the top-left (MATCHWEEK 24, NAS 2, SHB 1) and a "SSC 1 LIVE" television logo in the top-right, perfectly static. [Candid Discovery] The broadcast camera cuts from the game to scan the crowd in natural, ambient daylight. It locks onto a stunning woman in a dark brown silk halter dress. She is surrounded by a diverse audience wearing everyday casual clothes (hoodies, jackets, t-shirts). She is casually eating a burger and holding a blue soda can. She incidentally glances up, catches her own image on the stadium's giant screen, and realizes the live camera is on her. She gives a surprised, cool smirk to the camera. [The Transition] She quickly but calmly sets her burger and drink on her seat, stands up with athletic poise, and coolly vaults over the professional stadium railing onto the pitch. The broadcast camera pans smoothly to follow her movement from the stands to the grass. [The Interaction] On the pitch, a professional soccer player in a white kit passes the ball directly to her feet. Despite wearing sharp high-heeled shoes, she expertly traps the ball and executes a powerful, professional-level strike. The ball arcs with high velocity into the top corner of the goal net. [Audio & Finish] Audio is authentic stadium environment sound: the distant hum of thousands, the rustle of clothing, the sharp "thwack" of the high heel hitting the ball, and the massive roar of the crowd after the goal. She turns toward the lens with a radiant smile and playfully covers the camera with her hand, ending the broadcast segment. [Technical Specs] Broadcast-quality 4K, 60fps, natural color grading, high-end optical zoom, photorealistic textures, clean TV graphics.
Overall Style and Aesthetics: A high-energy, thrilling first-person action sequence, shot in a 9:16 vertical aspect ratio. The visual style is a raw, high-budget prehistoric adventure film, shot with a 35mm anamorphic lens, retaining a strong film grain and noticeable noise to create a realistic and stunning texture. The camera work simulates a GoPro-style wide-angle lens with peripheral distortion. High-fidelity visual effects, realistic mud and water physics, and 8K resolution. Narrative Timeline (9:16 Vertical Shots): 0:00 - 0:03 (High-Altitude Paragliding Flight) This scene opens from the first-person perspective of a paraglider pilot. Two gloved hands grip the control stick, and brown tactical boots sway at the bottom of the frame. The pilot soars above a vast, ancient jungle filled with giant tree ferns and prehistoric palm trees. In the center of the treetops, a massive, long-necked sauropod dinosaur (Brachiosaurus) stands majestically, its head facing the camera. 0:03 - 0:06 (Canyon Dive - Dynamic Chase) The pilot pulls the control stick, diving into a narrow basalt canyon in a thrilling maneuver. A group of pterosaurs, emitting sharp cries, blurs past the camera. Below, a rushing river flows between jagged rock walls. The camera uses strong motion blur and shaky shots to simulate wind resistance and speed during the descent. 0:06 - 0:10 (Water Glide - Low-Altitude Gliding) The pilot glides close to the water's surface, just inches away. Water spray and splash onto the "lens." The camera follows the winding river at incredible speed. Lush green vegetation and ancient moss-covered rocks rush past on either side. The lighting is natural, overcast sunlight, casting realistic reflections on the water. 0:10 - 0:13 (Fall - Physics and Textures) The paraglider loses altitude and plunges violently into a deep, muddy clearing. The camera angle drops sharply. A large patch of thick, brown mud splashes onto the lower half of the frame. The paraglider's white canopy collapses, hanging above the camera, forming a suffocating, translucent white veil.
Single continuous cinematic shot. Audio begins with daytime Shibuya street noise mixed with 8-bit video game sounds and chaotic Japanese chatter. Background music: shamisen strings. The camera pushes low along the ground. A girl’s leather shoe slams into frame — heavy first step. A Japanese schoolgirl in JK uniform sprints into frame at extreme speed, then suddenly slows on the second beat. She side-rolls and runs along a Shibuya crosswalk, shoulder nearly brushing the lens, skirt sweeping across frame. Obstacle ahead. She suddenly accelerates, steps onto a low wall — explosive jump. The camera lifts into the air with her. City lights streak below. She lands steadily, stabilizes with one hand, immediately rolls forward. The camera rotates half a turn with the roll, then quickly corrects. She sprints again. Full speed. Charges toward a railing. No hesitation. She leaps. The camera follows into weightlessness. Wind roars. Shamisen music briefly cuts midair. She lands on a lower platform. Knees absorb impact. Lowest point. Next beat — she snaps her head up and resumes sprinting into the crowd.
Vertical ASMR video. No music. Macro details. Shot 1: A light blue skincare gel bottle sits on glass. A pale elegant hand gently taps the glass, producing crisp fingernail tapping sounds. Shot 2: The hand picks up the bottle and slowly twists the cap. The cap rotation sound is clearly audible. Shot 3: A spoon scoops a portion of gel. With a soft “plop,” a thick blob of pale blue gel lands heavily in the center of the glass surface. The gel is dense and viscous, containing tiny realistic air bubbles. Dramatic cool lighting from behind makes it glow like a gemstone. Shot 4: The hand presses firmly onto the gel. The pressure flattens it, causing it to spread outward into an irregular circle. Tiny bubbles swirl within the texture. Palm lines and skin details are clearly visible through the glass. The hand slowly spreads the gel in circular motions. The thick gel thins into a semi-transparent oily film. Focus entirely on friction and fluid dynamics between hand, product, and glass.
Based on a storyboard of reference images, a fast-paced flashback montage was created. This is a nostalgic love story, shot on 35mm film in the style of a Kodak Portra 800. Soft grain, dreamlike bokeh, warm highlights, and slight color shifts create a retro cinematic atmosphere. The background music is an English song.
Fisheye Horse Comedy
TikTok-style vertical 9:16 marketing video. A glowing Colombian woman in her late 20s with bronze skin, fresh barefaced complexion, tight curly black hair pulled up in a claw clip, white terry headband, ivory bathrobe. She is in a bright marble bathroom with morning daylight. She turns on the pearl-white sonic facial cleansing brush, gently moves it across her cheek with foam, then turns to camera, beaming, and says in confident English: "My skin has never been this smooth. I am throwing out all my old face washes." Quick TikTok jump cuts: brush close-up, foam macro, before-and-after skin tone shot, sparkle SFX, natural sync sound English voiceover, upbeat clean beat track.
A car sketch on paper. The camera pushes in. The sketch lines rise off the paper, gaining dimensionality and color, transforming into a photorealistic 3D car driving on a road.
@image1's female star as the main subject, reference the camera movement method of @video1 for rhythmic push-pull-pan-tilt movements. The female star's movements also reference the dance movements of the woman in @video1, performing energetically on stage.
Ultra epic wow visual with a thrilling clear crisp great lighting scene with a frozen grassland under a pale orange dawn, where a rider in layered wool armor and carved antler shoulder plates faces a long-necked ice predator with translucent fins and six running limbs, intense anticipation across the open plain. The camera begins at horse level, racing beside the rider as the beast charges from the right. The rider throws a rope harpoon into the beast’s shoulder fin, circles wide to avoid its bite, then uses the horse’s turn to pull the beast off line. Snow trails and breath plumes show direction clearly. The camera speeds up with a clean lead shot in front of the horse, a wide side sweep showing the rope tension, and a gradual overhead arc as the beast recovers and sprints again. Climax: the rider dismounts onto a boulder, anchors the rope, and the beast’s own speed brings it down in a slide of ice and snow. Ending: the horse slows, the beast stops struggling, and the rider cuts the rope loose as dawn light warms the plain
The performance erupts within the grand, shadowed theater of **Figure 3**. The protagonist (**Figures 1 and 2**) sits at the center, a single high-contrast spotlight cutting through the darkness to reveal swirling dust. Following the cinematic rhythm of **Figure 4**, the scene unfolds with breathless intensity. The camera utilizes handheld techniques, creating a gritty, documentary-style urgency with slight motion blur as it pivots from wide shots to aggressive close-ups. We see the pianist’s fingers strike the keys with percussive force, capturing the vibration of strings and the spray of sweat under the harsh light. The visual narrative transitions from a top-down "God's eye" perspective to a claustrophobic close-up of the performer’s face, etched with strain and desperation. The color grading is desaturated and cool, emphasizing the stark, modern-action aesthetic. Every shot is framed with an imperfect, shaky composition that heightens the tension, turning a solo recital into a visceral, high-stakes battle.
00:00 - 00:02 | The Disappointment (Cold Realism) Visual: A man in a dimly lit, sterile kitchen stares sadly at a piece of dry, factory-processed white bread. The atmosphere is melancholic. Prompt: Medium shot, a middle-aged man looking disappointed at a piece of bland, industrial sliced bread on a cold granite counter, desaturated colors, cinematic moody lighting, 4k. 00:02 - 00:05 | Nostalgia Flashback (Vintage 16mm) Visual: Rapid-fire montage of "The Golden Age of Baking." Hands covered in flour, a wooden table, dough rising, and a stone oven glowing. Prompt: Fast rhythmic cuts, 16mm film grain, vintage sepia tones, close-up of hands kneading dough on a wooden board, flour dust flying in the air, golden sourdough loaf emerging from a wood-fire stone oven, nostalgic 1970s aesthetic. 00:05 - 00:08 | The "What Happened?" (Modern Grittiness) Visual: Quick, shaky shots of bread in plastic bags on a conveyor belt and a squashed, soggy loaf. Prompt: Handheld camera, gritty texture, quick zoom on a crushed loaf of bread in a plastic grocery bag, industrial factory lighting, fast-paced editing, motion blur. 00:08 - 00:12 | The Craftsmanship (High-End Food Porn) Visual: Extreme macro shots of the new handmade bread. The "crack" of the crust, steam rising from the center, and a baker smiling while handing it over. Prompt: Extreme close-up macro shot, a knife slicing through a crusty artisan loaf, audible crunch texture, steam rising from the soft white interior, warm golden hour lighting, high-speed camera (slow motion), a friendly baker in a flour-dusted apron smiling. 00:12 - 00:15 | The Resolution (The New King) Visual: The man from the first scene takes a massive, happy bite of the fresh bread. Bold text appears: "REAL BREAD. FOR THE REAL KING." Prompt: Cinematic close-up, the man taking a satisfying bite of a thick slice of handmade bread, eyes closing in delight, warm sun-drenched lighting, high-end color grading, bold gold text overlay 'THE REAL BREAD IS BACK,' logo reveal.
MiniMax H3 モデルの概要
MiniMax H3 は、2026 年 7 月 31 日にリリースされた MiniMax AI の汎用マルチモーダル生成モデルです。テキストから動画、画像から動画、モーション参照、およびオーディオを個別のタスクとして扱うのではなく、MiniMax H3 はテキスト、画像、動画、およびオーディオを 1 つのクリエイティブなコンテキストとして理解します。
MiniMax H3 AI 動画 ジェネレーターは、最大 2K 解像度、15 秒のネイティブ ステレオ サウンドを含む動画を作成します。 MiniMax は、広告、電子商取引、製品デザイン、UI/UX、ゲーム、および映画のコンセプトに対する指示のフォロー、正確なテキストとブランドのレンダリング、V2V モーション転送、自然言語の参照と編集関係を強調します。
MiniMax H3 AI 動画 ジェネレーターが、統合コンテキスト、ネイティブ ステレオ サウンド、2K 出力、自然言語制御をどのように組み合わせているかをご覧ください。
MiniMax H3 は、テキスト、画像、動画、オーディオの関係を 1 つのクリエイティブ ブリーフで理解します。リファレンスは、新しいショットの主題、パフォーマンス、カメラ言語、タイミング、またはサウンドをガイドすることができます。
MiniMax H3 AI 動画 ジェネレーターは、サウンドを別個の仕上げステップとして扱うことなく、画像とネイティブ ステレオ オーディオを一緒に作成して、より一貫したダイアログ、アンビエンス、エフェクト、音楽を実現します。
広告、電子商取引、UI/UX、ゲーム、デザインのワークフロー向けに、強力な命令追従、正確なテキストとブランドのレンダリング、V2V モーション転送を備えた最大 15 秒の 2K MiniMax H3 動画を作成します。
MiniMax H3 アプリでのアイデアからエクスポートまでの 3 つの簡単なステップ。
プロンプト、静止画像、または参照クリップから始めます。 MiniMax H3 は、自然言語の説明から被写体、カメラの動き、ペース、ビジュアル スタイルを読み取ります。
テキストから動画、画像から動画、または参照から動画のモードを選択し、解像度とアスペクト比を設定して、MiniMax H3 で最初の映画のようなカットを生成します。
プロンプトを調整し、参照を交換し、バージョンを比較し、クリップを拡張し、MiniMax H3 の結果がクリエイティブ ブリーフと一致する場合はエクスポートします。
プロンプト、製品画像、モーションリファレンス、またはオーディオキューを 1 つの MiniMax H3 ブリーフに組み込みます。生産計画を再構築することなく、方向を比較し、すべての入力間の関係を調整します。
MiniMax H3 アプリを探索するMiniMax H3 画像、モーション キュー、ショット言語をまとめて保存します。
ブリーフ全体を再構築せずに、MiniMax H3 シーンを再実行します。
いくつかの MiniMax H3 方向を生成し、最も強いカットを選択します。
動画のリファレンスを使用して、動き、タイミング、視覚的な方向をガイドします。
MiniMax H3 は、テキストから動画の変換、画像から動画の変換、モーション参照、およびオーディオのタスクを 1 つの創造的なコンテキストとして理解することで、個別のタスクを超えて前進します。
MiniMax H3 は、広告コンセプト、製品デモ、ストーリーボード、ソーシャル メディア コンテンツ、トレーニング 動画、および迅速なクリエイティブ テストに使用します。
TikTok、リール、ショート、ローンチ動画、有料ソーシャルバリエーション用の MiniMax H3 短編コンセプトの草案を作成します。
製品画像を、季節のフック、ヒーロー ショット、プラットフォーム対応作物を備えた MiniMax H3 アニメーション プロモーションに変換します。
MiniMax H3 を使用してシーンを探索し、カメラの動きを事前に視覚化し、制作前の映像を作成します。
資料、説明、トレーニング資料、社内発表、セールス ストーリー用の MiniMax H3 モーション クリップを作成します。
MiniMax H3 ビジュアル スタイル、ミュージック 動画、インスタレーション、超現実的なループ、コンセプト フィルムを試してください。
MiniMax H3 学習アセットを使用して、レッスン、抽象的なアイデア、言語のバリエーション、およびクイック リフレッシュを示します。
動画制作量、キューのニーズ、クレジットの使用状況に合わせて MiniMax H3 プランを選択してください。
MiniMax H3、MiniMax AI、MiniMax Hailuo、Hailuo H3、Hailuo AI、価格設定、入力、マルチモーダル 動画 ワークフローに関する回答。
MiniMax H3 は、2026 年 7 月 31 日に発表された MiniMax AI の汎用マルチモーダル 動画生成モデルです。テキスト、画像、動画、オーディオ コンテキストを統合して、最大 2K 解像度、15 秒のネイティブ ステレオ サウンドを備えた動画を作成します。
MiniMax H3 のテキストから動画の変換、画像から動画の変換、および参照から動画への変換を 1 つのワークフローで作成してみてください。
MiniMax H3 から始める