Prompt field guide

MiniMax H3 prompts for your next video

Watch a real sample, choose a template, and change the subject and action. Create an H3 video with sound in English or Chinese. Explore 8 full templates and the official prompt structure below.

All eight templates are written by this site, not published by MiniMax. Each card shows a clip we generated with MiniMax H3 from that template's English prompt, so you can judge the result before you spend credits. The generator here runs text-to-video and first-frame image-to-video only.

Last updated 2026-10-15

Real H3 output from the original English food template. Your edited prompt produces a new result.

Make a 5-second draft for your idea

Choose a scene, replace the subject and action, then edit the complete prompt below.

Too vague: “A nice food video.”

Give H3 a visible action, a camera move and matching sound. The draft includes all three.

A separate single-shot draft, 16:9. The full 8-template library below keeps every original shot.

Generate your H3 video here

Edit the loaded prompt before continuing. Full templates keep their original duration and format; image templates require your own upload.

0/7000

By generating, you acknowledge that this submitted prompt may be securely retained for product analytics for up to 30 days when analytics consent and protected collection are enabled. Never include passwords, keys, personal contact details, or other secrets.

Duration

Resolution

Aspect ratio

Sign in to generate

Costs 20 credits · 0 welcome credits · 5 seconds = 20 credits

Your H3 video will appear here

Copy, then make it yours

MiniMax H3 prompt examples for eight common shots

Eight templates built on MiniMax's published structure, covering text-to-video and first-frame image-to-video. Copy one, then replace the subject, action, timing, dialogue, and sound with your own shot.

Fills the generator with this prompt, mode, duration, and aspect ratio.

Filter by mode
Filter by duration

Every clip on this page was generated on our own RTX 5090 running MiniMax H3, from the English prompt printed on its card, and nothing was re-edited afterwards. Each template got a single generation — no seed re-rolls and no best-of picks, so what you see includes the model's misses. The three image-to-video clips start from first frames we generated with FLUX, which makes them motion added to a still rather than a re-shot scene.

T2VA10s16:9

Cinematic product reveal

A two-shot landscape product film with controlled rotation, macro texture, rain ambience, and a restrained electronic score.

Prompt language
integrated_multimodal_description: [Shot 1] Live-action product film, a medium-wide shot frames a matte-black wireless speaker on a wet stone plinth at blue hour. Fine rain beads on the metal grille while a narrow amber light travels across its edge. The camera arcs clockwise with small amplitude at slow speed as the speaker turns one quarter rotation, keeping the logo sharp and readable. [Shot 2] At 00:06.000, the camera cuts to a macro close-up of water droplets vibrating with the bass, then pulls out slowly as the product settles at the center of the frame.

overall_soundscape: Light rain taps against stone while a low, clean bass pulse makes the water tremble. A soft mechanical turntable hum remains underneath.

non_diegetic_music: A restrained electronic beat at a moderate tempo with deep sub-bass and sparse metallic percussion, fading cleanly at the end.
T2VA10s16:9

Two-person dialogue scene

Stable speaker IDs, language-tagged dialogue, a motivated cut, and environmental sound for a short dramatic exchange.

Prompt language
integrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot frames two sisters waiting beneath the awning of a closed train station at night. Rain falls beyond the warm pool of light. The older sister with a low, steady voice (S1) looks toward the empty tracks and says: <d>[English] We missed the last one.</d> [Shot 2] At 00:05.000, the camera cuts to a close-up of the younger sister with a quick, bright voice (S2). She lifts a bicycle key, smiles, and replies: <d>[English] Then we take the long way home.</d>

overall_soundscape: Rain strikes the metal awning, distant traffic passes behind the station, and the bicycle key gives a small metallic jingle.

non_diegetic_music: Sparse piano notes at a slow tempo, joined by a soft sustained cello after the second line.
T2VA10s16:9

Tracking action sequence

A fast lateral tracking move, one informative cut, physically linked action, and synchronized impact sounds.

Prompt language
integrated_multimodal_description: [Shot 1] Live-action sports commercial, a low tracking shot follows a mountain biker accelerating along a narrow forest trail after rain. The rear tire throws small arcs of mud while the rider leans into a left turn. The camera trucks beside the bicycle at fast speed, maintaining the rider in the right third of the frame. [Shot 2] At 00:06.500, the camera cuts to a front three-quarter close shot as the rider clears a shallow stream, lands firmly, and exits toward a bright opening between the trees.

overall_soundscape: Tires grind over wet gravel, the chain clicks under load, water splashes on the landing, and the rider breathes sharply beneath the wind.

non_diegetic_music: Fast hand percussion and a short distorted bass pattern build through the jump, then stop on the landing.
T2VA5s1:1

Five-second macro food shot

One simple action, a small camera move, material detail, and no background music in a square composition.

Prompt language
integrated_multimodal_description: [Shot 1] Live-action macro food cinematography, an extreme close-up frames a spoon breaking through the caramelized top of a small crème brûlée. The camera pushes in with small amplitude at slow speed as the brittle sugar shell cracks into irregular amber pieces and the pale custard folds around the spoon. Warm side light reveals steam and fine texture; the dessert remains centered against a dark neutral background.

overall_soundscape: The sugar crust gives a crisp crack, followed by the soft scrape of a metal spoon against ceramic and quiet room tone.

non_diegetic_music: N/A
T2VA10s9:16

Vertical fashion film

A 9:16 walk-and-pause sequence with disciplined lighting, garment motion, framing, and echoing footsteps.

Prompt language
integrated_multimodal_description: [Shot 1] Live-action vertical fashion film, a full-body shot frames a model in a structured red coat walking through a concrete gallery. Hard morning light creates long rectangular shadows across the floor. The camera tracks backward at the model's pace while the coat hem moves naturally with each step. [Shot 2] At 00:06.000, the shot cuts to a tight profile as the model stops beside a mirrored wall, turns toward her reflection, and adjusts one cuff without looking at the camera.

overall_soundscape: Firm footsteps echo through the gallery while fabric shifts softly and distant city noise enters through an open doorway.

non_diegetic_music: A minimal drum-machine rhythm at a steady moderate tempo with one dry synth note repeating every two beats.
I2VA5sadaptive

Animate a portrait first frame

Preserve identity and composition while adding only a breath, an eye-line change, and a small natural expression.

Prompt language
For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.

integrated_multimodal_description: [Shot 1] Live-action, cinematic, the person shown in <Picture 1> keeps the same facial features, hairstyle, clothing, lighting direction, and background composition. The camera holds a static close shot as the person takes a quiet breath, shifts their gaze from the window toward the camera, and forms a small natural smile. Hair and loose fabric move only slightly in the existing breeze; no new objects enter the frame.

overall_soundscape: Soft room tone continues with a faint breeze and one quiet breath.

non_diegetic_music: N/A
Edit & generate this template

Needs a first-frame image — upload one after tapping Use this prompt.

I2VA5sadaptive

Animate a product first frame

Keep labels and geometry stable, limit the rotation, and explicitly prevent unsupported new props from entering the frame.

Prompt language
For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.

integrated_multimodal_description: [Shot 1] Live-action product cinematography, the product in <Picture 1> keeps its exact shape, materials, label text, color, position, and background. The camera arcs clockwise with small amplitude at slow speed while the existing highlight travels naturally across the surface. The product rotates no more than fifteen degrees, then returns to a stable hero angle with the label facing the camera. Do not add hands, packaging, liquid, smoke, or extra props.

overall_soundscape: Quiet studio room tone with a subtle mechanical turntable hum.

non_diegetic_music: A single low synth note rises gently and fades before the final frame.
Edit & generate this template

Needs a first-frame image — upload one after tapping Use this prompt.

I2VA10sadaptive

Animate a landscape first frame

Move only elements already present in the image while preserving time of day, geography, palette, and composition.

Prompt language
For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.

integrated_multimodal_description: [Shot 1] Cinematic landscape, the mountains, lake, shoreline, clouds, and color palette in <Picture 1> remain consistent. The camera pushes forward with small amplitude at slow speed above the existing shoreline. Ripples travel across the lake, the visible grass bends gently in the wind, and the existing clouds drift gradually from left to right. Preserve the time of day and introduce no people, buildings, animals, or weather effects that are absent from the first frame.

overall_soundscape: A light breeze moves through grass while small waves touch the shore and a distant bird calls once.

non_diegetic_music: Sustained soft strings at a slow tempo, remaining quiet and even throughout.
Edit & generate this template

Needs a first-frame image — upload one after tapping Use this prompt.

The three-field H3 prompt structure

integrated_multimodal_description:
[Shot 1] style + composition + subject + action
[Shot 2] At 00:05.000, cut + camera + result

overall_soundscape:
ambience + physical sounds + non-verbal sounds

non_diegetic_music:
instruments + tempo + dynamics, or N/A

Write for time, not a still image

What a strong MiniMax H3 prompt controls

Start with the visible and audible timeline. Every detail should map to a shot, an action, a camera decision, dialogue, or a sound that can happen during the clip. Specific, compatible instructions are more useful than a long pile of visual adjectives.

Chronological shot plan

Open with [Shot 1] and add later cuts with increasing timestamps such as [Shot 2] At 00:05.000. A cut should reveal new information; use camera movement when only framing changes.

Camera type, range, and speed

Write camera motion as part of the action: push in, pull out, pan, truck, tilt, arc, track, hold static, or use POV. Add small or large amplitude and slow or fast speed only when they matter.

Sound in the right field

Put dialogue and shot-synchronized sound in the timeline, summarize ambience and physical sounds under overall_soundscape, and reserve non_diegetic_music for music the characters cannot hear. Use N/A when no score is wanted.

Five official prompt modes

Choose the H3 structure that matches your inputs

The opening instruction changes with the assets you provide. Do not paste reference labels into a plain text-to-video request, and do not describe a first-frame image as though the model has never seen it.

T2VA

Text-to-video begins directly with the three core fields and builds the complete audiovisual timeline from text.

Available here

I2VA

Image-to-video first anchors Picture 1 at 0.00 seconds, then preserves its subjects, composition, lighting, and objects while describing forward motion.

Available here

FL2VA

First-and-last-frame video aligns two pictures to the opening and ending timestamps, then describes a continuous, physically plausible path between them.

Official guide only

L2VA

Last-frame video infers a compatible opening state and gradually converges on the supplied final picture at the exact ending time.

Official guide only

Ref2VA

Full-reference video defines subjects and picture, video, and audio labels before describing what is preserved, transferred, edited, or reused across the target timeline.

Official guide only

Chinese prompts

H3 takes Chinese values, but the field names stay English

The three timeline fields are a fixed schema, so integrated_multimodal_description, overall_soundscape, and non_diegetic_music are written exactly as MiniMax defines them. Everything inside those fields can be Chinese: scene description, camera movement, dialogue, ambience. MiniMax's own skill and API documentation are written in English, which is why most published examples are English too.

  • Keep the three field names in English. Translating them breaks the schema the model was trained on.
  • Write the values in whichever language your audience speaks. Keep camera terms recognisable — Arc, Push In, Pull Out, Truck, Tracking — either in English or with the English term in brackets.
  • Tag dialogue with its language so the model casts the right voice: <d>[Chinese] 台词</d> or <d>[English] line</d>. Give each speaker one stable ID such as (S1).
English value
[Shot 1] Live-action macro food cinematography, an extreme close-up frames a spoon breaking through the caramelized top of a small crème brûlée. The camera pushes in with small amplitude at slow speed...
Chinese value
[Shot 1] 实拍微距美食摄影,极端特写拍一把勺子压碎一小份焦糖布丁的糖壳。镜头以小幅度、慢速推进(Push In)...

Both lines describe the same shot, and the field name above them is identical. Switch any template card between 中文 and EN to see a full side-by-side.

Choose one change for your next draft

Keep the same source and scene while you test one adjustment. These are prompt-writing starting points, not measured fixes.

What you noticeWhat to try nextWhat to keep
The clip rushes through too many actionsReduce the prompt to one action, such as lifting a cup, before adding a second beat.The subject and location.
An image-to-video result drifts from the referenceState what stays fixed, then describe a small movement rather than redesigning the person or product.The same reference image.
The camera direction is inconsistentChoose a static camera or one movement; remove simultaneous push-in and pull-back instructions.The action and sound.

Before you generate

A five-point H3 prompt check

  • Keep the action achievable in 5, 10, or 15 seconds; shorten the story before adding more cuts.
  • Give every speaking person one stable ID such as (S1), and keep only the spoken words inside the <d> block.
  • Use exact timestamps only from the second shot onward; [Shot 1] starts at zero without one.
  • For image-to-video, preserve what the upload already establishes and describe the motion that develops from it.
  • Remove conflicting camera commands, duplicate sound instructions, and decorative adjectives that do not change the shot.

Sources and evidence boundary

The structure, field names, and mode definitions follow MiniMax's published H3 prompt material and API documentation. The eight templates and every clip on this page are produced by this site; none of them is presented as an official MiniMax prompt or as a controlled quality benchmark.

MiniMax H3 prompt FAQ

Five things people ask before writing an H3 prompt

Can I write a MiniMax H3 prompt in Chinese?

Yes. The three field names stay in English because they are a fixed schema, but the values inside them can be Chinese — scene, action, camera movement, dialogue, and ambience all work. Tag dialogue with its language, for example <d>[Chinese] 我们错过了最后一班。</d>, so the model picks the right voice. Every template on this page has a 中文 version you can switch to.

My prompt looks fine but the video ignores it. What usually goes wrong?

Three useful starting checks: shorten an action chain that does not fit the clip; keep one compatible camera move per moment; and, for image-to-video, describe what changes instead of rewriting everything already visible in the upload. For example, replace “the person walks across the room, picks up a cup, drinks, turns, and exits” with one beat: “the person lifts the cup; the camera holds still.” Generate another draft with the same source image and change one instruction at a time so you can compare. This is a planning example, not a guaranteed repair or a measured output.

Does H3 have a negative prompt field?

There is no separate negative-prompt input in the documented H3 format. Write exclusions as a normal sentence in the timeline instead, in the same place the action lives — for example Do not add hands, packaging, liquid, smoke, or extra props. The three image-to-video templates on this page do exactly that.

Can I paste these templates into Veo, Kling, or another video model?

Not as-is. The field names, the [Shot N] timeline, the At 00:05.000 cut timestamps, and the <d> dialogue tags are MiniMax H3 conventions. Other models read a plain prose paragraph better. Reuse the shot plan — subject, action, camera move, sound — and rewrite the wrapper in that model's own format.

Should the prompt state resolution, duration, or aspect ratio?

No. Those are request parameters you select in the generator, and H3 ignores them inside the prompt text. Use the prompt for content and time: what happens, in what order, seen from where, and heard how. Writing "15s, 16:9" at the start of a prompt only spends characters that could describe the shot.

Generate your H3 video here

Use this 5-second draft