# MiniMax H3 prompt guide

URL: https://dreambench.ai/guides/minimax-h3-prompt-guide

To write a MiniMax H3 prompt, describe the clip along its timeline: the style and opening frame, then each shot’s action, camera move and spoken lines, with a cut time for every later shot. Then add the soundscape, and any music only the audience hears. H3 makes up to 15 seconds, always with stereo sound.

Written and reviewed by the DreamBench team. Published 2026-11-01; updated 2026-11-01.

## MiniMax H3 prompts at a glance

- **The structure:** Shots on a timeline, then the soundscape, then the music.
- **One generation:** 4 to 15 seconds in MiniMax’s API, 5 to 15 on DreamBench, always in stereo.
- **Dialogue:** Stable in 11 languages.
- **Prompt length:** Up to 7,000 characters in MiniMax’s API; up to 5,000 on DreamBench’s Video page.
- **On DreamBench:** The Video page sends your prompt as written: five frames, 720p or 1080p.

This guide follows MiniMax’s [prompt writing guide](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_base_en.md) for H3 and its [API guide](https://platform.minimax.io/docs/guides/video-generation). For the model, see the [MiniMax H3 page](https://dreambench.ai/models/minimax-h3).

## The MiniMax H3 prompt structure: description, soundscape, music

MiniMax’s guide splits a prompt into three parts: a description along the timeline (style, composition, subjects, action, camera, speakers and the sounds tied to an action), a soundscape for the whole clip, and music the characters can’t hear.

**Style and frame → subjects → action → camera → speakers → soundscape → music**

|  | Why it matters | Strong | Weak |
|---|---|---|---|
| Style and frame | MiniMax opens with the look and composition. | Live-action, cinematic, a medium-wide shot of a bakery at dawn. | A nice bakery video. |
| Subjects | Named details are the ones H3 keeps. | A middle-aged baker in a green apron, on the left | A baker |
| Action, in order | Each detail is seen or heard, in order. | lifts the shutters, sets a loaf on the counter, then slices it | baking bread |
| Camera | A move, with size and speed if needed. | The camera pushes in with small amplitude at slow speed. | Dynamic camera. |
| Speakers and lines | A voice described once stays the same speaker. | The baker, a calm, raspy voice, says in English: “First batch.” | He says something friendly. |
| Soundscape | Ambience and physical sound, briefly. | Trays clink; the doorbell rings once. | Nice sounds. |
| Music | Instruments and tempo, not moods. | A soft acoustic guitar, fading out. | Emotional music. |

1. Open with the style and first composition: “Live-action, cinematic, a medium-wide shot frames…”.
2. Name each subject with the details that must stay: age, clothes, colors, position.
3. Write the action in order, with the camera move inside the sentence.
4. Start each later shot with its cut time: “[Shot 2] At 00:05.000, the camera cuts to…”.
5. Describe each voice once, then the line in quotes, with its language.
6. Close with the soundscape, then the music, or say there is none.

MiniMax’s guide writes the parts as labeled fields, the form its prompt rewriter, H3-Context-IR, produces; its [API examples](https://platform.minimax.io/docs/api-reference/video-generation-v2-create) are plain sentences. Either way, keep the order.

## MiniMax H3 camera movement prompts

MiniMax’s guide builds a move from its type, how far it goes (amplitude) and how fast (speed), written as an action in the sentence, not a label at the end.

**Camera moves in MiniMax’s words, and how to write them**

|  | What it does | How to write it |
|---|---|---|
| Push in, pull out | The camera moves forward or back. | The camera pushes in with small amplitude at slow speed toward the letter. |
| Pan | Turns left or right from one spot. | The camera pans right at fast speed, revealing the doorway. |
| Truck | Slides sideways. | The camera trucks right at slow speed along the train window. |
| Tilt | Turns up or down from one spot. | The camera tilts up from her boots to her face. |
| Pedestal | The whole camera rises or sinks. | The camera pedestals up over the market stalls. |
| Arc shot | Moves in an arc around the subject. | An arc shot circles the dancer at slow speed. |
| Tracking shot | Follows a moving subject. | A tracking shot follows the cyclist through the crowd. |
| Static shot | Doesn’t move at all. | The camera holds a static shot as the runner leaves. |
| Shake | A slight or a strong shake. | The camera shakes strongly as the bridge jolts. |
| POV | Sees through the subject’s eyes. | POV of the cyclist, handlebars in frame. |

MiniMax’s [API guide](https://platform.minimax.io/docs/guides/video-generation) also accepts a short form: [pan], [zoom] or [static] straight after the description it applies to. Hailuo’s [camera guide](https://hailuoai.video/pages/knowledge/ai-camera-movements-pan-tilt-zoom-guide) adds: one main move per clip, calmed with slow or subtle.

For a steady frame, Hailuo’s [guide to unwanted motion](https://hailuoai.video/pages/knowledge/ai-video-camera-motion-control) suggests rig words, a locked-off shot or a tripod; handheld adds randomness. If only the distance changes, MiniMax says move the camera rather than cut.

## How to write dialogue, sound and music for MiniMax H3

Every H3 clip has stereo sound made with the picture. MiniMax puts lines and action sounds in the description, where they happen; the soundscape covers the rest; music goes last. See [native audio](https://dreambench.ai/glossary/native-audio).

- Describe a voice once: age, pitch, timbre, pace or accent. MiniMax gives each speaker an ID, (S1) or (S2), kept across shots.
- Keep the delivery outside the quotes and only the spoken words inside: she says in French, “Bonjour, les croissants sortent du four.”
- For a voice-over, write “says in an off-screen voiceover”, then add that the on-screen character’s lips stay closed.
- Write the soundscape in one to four sentences: wind, footsteps, breathing; no dialogue.
- Write the music as instruments, tempo and volume, not moods, or say there is none.

*Diagram: Sound made in the same pass as the picture: lines, effects and ambience timed to the action.*

MiniMax’s [model card](https://huggingface.co/MiniMaxAI/MiniMax-H3) lists dialogue that holds up in 11 languages: Arabic, Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish; others work less consistently. Put any sign or label text in double quotes, exactly as it should read. See [lip sync](https://dreambench.ai/glossary/lip-sync).

## How long a MiniMax H3 prompt should be, and timing shots in 15 seconds

MiniMax’s API takes up to 7,000 characters, and its rewriter’s prompts run to hundreds of words: spend them on what is seen and heard, moment by moment.

**How much an H3 clip holds**

|  | What fits | Prompt length |
|---|---|---|
| 5 to 6 seconds | One shot: an action and its result, or one short line. | 60 to 120 words |
| 8 to 10 seconds | One shot with a build-up and a payoff, or two shots and one cut. | 120 to 250 words |
| 12 to 15 seconds | Two or three timed shots, or a short exchange between two speakers. | 200 to 500 words |

Time cuts as MiniMax does: no time on the first shot, a later time on each one after, and something new at each cut. Hailuo’s general [prompt guide](https://hailuoai.video/pages/blog/effective-ai-video-prompt-guide) suggests 20 to 50 words for most platforms; H3’s own examples run longer, so try both.

On DreamBench, H3 renders 21:9, 16:9, 9:16 for Reels and Shorts, 1:1 and 4:3; MiniMax’s [Hailuo AI](https://hailuoai.video/tools/minimax-h3) app also offers 3:4.

## MiniMax H3 image-to-video and first and last frame prompts

MiniMax’s API takes a first frame, a last frame or both; the picture sets the clip’s shape. See [image to video](https://dreambench.ai/glossary/image-to-video) and [first and last frame](https://dreambench.ai/glossary/first-and-last-frame).

**What to write when H3 starts or ends on a picture**

|  | What the picture fixes | What the prompt says |
|---|---|---|
| A first frame | Everything at 0 seconds. | How the action begins, develops and ends. |
| First and last frames | The opening and the closing state. | The path between, ideally one shot. |
| A last frame only | The final moment. | An earlier state, then how it lands on the picture. |

1. Start from the picture: the young woman by the rainy train window.
2. Say how the action begins: the camera trucks right slowly as she looks up from the letter.
3. End on a result: she folds the letter along its crease.

*Diagram: A first and a last picture, and the motion the model fills in between them.*

MiniMax’s own example asks for change: “Pull focus to the people in the background and add more steam to the ramen bowl.” DreamBench’s composer has no start or end slot: in a film it draws each opening frame from your references, and the prompt says how the clip ends.

## Keeping a character the same across MiniMax H3 shots

MiniMax’s reference mode takes up to 9 images, 3 videos and 3 audio clips, 12 files in all; its [feature examples](https://platform.minimax.io/docs/guides/video-prompt) name each in the prompt: “Image 2: character. Image 3: bag.” References and first or last frames can’t be mixed.

- Describe each character in the same words in every shot: “a middle-aged baker with a gray beard and a green apron”.
- Keep each speaker’s voice description, and their ID, the same across shots.
- Keep the clothes, props, time of day and light.

DreamBench doesn’t send H3 reference images, videos or audio. In a film, each character, place and prop is drawn once as a sheet, and each clip’s opening frame is drawn from the sheets it needs. See [character consistency](https://dreambench.ai/glossary/character-consistency) and [reference to video](https://dreambench.ai/glossary/reference-to-video).

## Multi-shot MiniMax H3 videos: cuts in one clip, or several clips

MiniMax trained H3 for multi-shot video, and its guide times cuts inside one clip. Past 15 seconds, or to redo one shot alone, make each shot its own clip. See [multi-shot video](https://dreambench.ai/glossary/multi-shot-video).

*Diagram: One idea, several shots: each its own generation, hard cuts at the joins, any one redone alone.*

**One 12-second idea, as three shots inside one H3 clip**

|  | What the shot shows | Camera | Sound |
|---|---|---|---|
| Shot 1 | The baker sets the first loaf on the counter | Medium-wide, pushing in slowly | “First batch of the morning.” |
| Shot 2, at 00:05.000 | Steam rises as a knife slices the loaf | Static close-up | The knife through the crust |
| Shot 3, at 00:09.000 | A customer takes the paper bag | Static, over the counter | The doorbell, a coin on the wood |

- Inside one clip: number the shots, time each cut after the first, describe subjects identically.
- Across clips: one moment each, new framing at every join, ambience running across it.

On DreamBench, [describe the film](https://dreambench.ai/) and pick MiniMax H3: it writes each shot as one continuous clip of up to 12 seconds, draws its opening frame and cuts the clips together; redo any one with a note. On the Video page, your shot labels reach H3 as written.

## MiniMax H3 prompt examples to copy

Paste these into DreamBench’s Video page or any MiniMax H3 tool. Each is one clip.

#### Three timed shots: the morning bakery

12 seconds, 16:9. [Shot 1] Live-action, cinematic, a medium-wide shot of a street bakery before sunrise. A middle-aged baker with a gray beard and a green apron sets a loaf on the counter as the camera pushes in slowly. The baker, a calm, raspy voice, says in English: “First batch of the morning.” [Shot 2] At 00:05.000, the camera cuts to a static close-up of steam rising as he slices the loaf. [Shot 3] At 00:09.000, the shot cuts to a customer’s hand taking a paper bag. Soundscape: trays clink, the doorbell rings once. Music: a soft acoustic guitar, fading at the end.

#### Product with a label: a tin of tea

8 seconds, 1:1. Live-action, a close shot of a tin of tea on an oak table, morning light from the left. The lid reads “HIGHLAND BREAKFAST” in gold capitals. A hand lifts the lid and a spoon scoops dark leaves; the camera pushes in slowly, the lettering sharp. Soundscape: the lid’s soft pop, leaves rustling. Music: none.

#### Dialogue: two speakers on a night bus

10 seconds, 16:9. Live-action, a static two-shot on the top deck of a night bus, city lights streaking past. A young man in a yellow raincoat, a quick, nervous voice, says in English: “I think we missed our stop.” The woman beside him, a low, amused voice, doesn’t look up from her book: “Two stops ago.” Soundscape: the engine’s hum, rain on the roof. Music: none.

#### A line in Spanish: the fruit stall

6 seconds, 9:16. Live-action, a static medium shot at a market stall in midday sun. An older woman in a striped apron holds a mango up to the camera and, in a warm, brisk voice, says in Spanish: “¡Prueba este, está dulcísimo!” Soundscape: market chatter, a paper bag crackling. Music: none.

#### Voice-over: a letter read aloud

10 seconds, 16:9. Live-action, a close shot of an old man in a brown cardigan reading a handwritten letter by lamplight. The camera trucks left slowly. The old man, a soft, low voice, says in an off-screen voiceover: “I still remember the road to your house.” His lips stay closed. Soundscape: a clock ticking, rain at the window. Music: sparse piano at a slow tempo.

#### Close-up sound: bread on a board

6 seconds, 9:16. Live-action, a static extreme close-up of a sourdough loaf on a board. A serrated knife saws through it in three slow strokes and the slice falls away. Soundscape: the crust crackling under the blade, close and crisp. Music: none.

#### Stereo travel: a tram across the frame

10 seconds, 21:9. Live-action, a wide static shot of a cobbled square at dusk. A red tram enters on the left, rattles across and leaves on the right, ringing its bell once. Soundscape: the rumble travels from left to right in stereo, wheels squeal on the bend. Music: none.

#### Animation: a clay robot

8 seconds, 1:1. Claymation, a small blue clay robot with one round eye on a workbench. It picks up a bolt, screws it into its own chest and gives a proud bounce. An arc shot circles it slowly. Soundscape: clay squeaks, a metallic click. Music: a light xylophone at a bouncy tempo.

#### App motion: a banking screen

8 seconds, 9:16. 3D CG, a phone screen on a lilac background. A balance card slides up reading “£2,480.00”, three payments drop in, and a green check mark appears beside “Rent paid”. The camera pushes in slowly. Soundscape: soft taps, one chime on the check mark. Music: a minimal electronic pulse.

#### Sport: a rower at dawn

8 seconds, 16:9. Live-action, a sculler in a white tank top rows across a misty lake at dawn. A tracking shot follows at water level, then the camera pedestals up as the boat glides past. Soundscape: oars dipping, the seat sliding, a bird calling. Music: none.

#### Vertical trailer: the lighthouse

15 seconds, 9:16. [Shot 1] Live-action, cinematic, a wide shot of a lighthouse as storm waves break over it. [Shot 2] At 00:04.000, the camera cuts to a young keeper with cropped red hair and an oilskin coat at the rail; she shouts in English, in a hoarse voice: “The light’s gone out!” [Shot 3] At 00:09.000, the shot cuts to her hand striking a match in the dark lamp room. Soundscape: wind howling, waves booming. Music: low strings building, stopping on the flame.

#### A one-line prompt with a camera instruction

5 seconds, 16:9. A fox trots across a snowy field at dusk [static], stops, then dives head first into the snow [zoom]. Soft wind, a muffled thump.

## Weak MiniMax H3 prompts, rewritten

Most weak H3 prompts are short, not wrong.

**Three weak prompts and their strong versions**

|  | Weak | Strong | What changed |
|---|---|---|---|
| An ad | Luxury perfume ad, cinematic, 8K, beautiful lighting. | 8 seconds, 16:9. Live-action, a perfume bottle reading “NOIR 9” on black stone; warm light sweeps across it as the camera pushes in slowly. Music: a slow cello note. | A style and frame, readable text in quotes, one move, light and sound. |
| A talk | Two friends chatting in a café. | 10 seconds, 16:9. A static two-shot in a café. A woman in a red scarf, a bright voice, says: “You’re late.” The man, a tired voice: “The trains hate me.” Soundscape: cups clinking. | Each speaker described, their line in quotes, a soundscape. |
| A story | A storm hits a lighthouse and the keeper relights the lamp. | 15 seconds, 9:16: the storm wide; at 00:04.000 the keeper close; at 00:09.000 the match. Wind and waves throughout. | One idea became timed shots. |

## Common MiniMax H3 prompt mistakes and how to fix them

**MiniMax H3 prompt mistakes and fixes**

|  | What goes wrong | Fix |
|---|---|---|
| Mood words for music | “Emotional music” is guessed. | Name instruments, tempo and changes in volume. |
| A timestamp on the first shot | The timeline starts wrong. | No time on Shot 1; later times after. |
| A cut for a small change | A jump where a move would do. | Move the camera; cut for something new. |
| Camera labels stacked at the end | The move isn’t tied to the action. | Write it in: the camera pans as she turns. |
| A speaker with no voice | The wrong face speaks. | Describe each voice once. |
| Dialogue in the soundscape | The line lands at the wrong moment. | Put lines where they are spoken. |

## What a MiniMax H3 prompt can’t fix

- One prompt run twice gives two clips.
- MiniMax says visual detail can still improve in some scenes: keep key details large in frame, at 1080p.
- Every clip has sound; for silence, ask for room tone and mute it in an editor.
- A real person or product can’t be known from words; it takes a picture you have the right to use.
- One generation stops at 15 seconds, and on DreamBench’s Video page a prompt past 5,000 characters is refused.

## MiniMax H3 prompts on DreamBench

- One clip: on the Video page, pick MiniMax H3, paste your prompt, choose 5, 6, 8, 10, 12 or 15 seconds, the frame and 720p or 1080p. It goes to H3 as written.
- A prompt: the [AI video prompt generator](https://dreambench.ai/tools/ai-video-prompt-generator) writes one for any model in shot-list order; it has no H3 preset and writes the prompt only.
- A film: [describe it](https://dreambench.ai/), add up to 8 reference pictures and pick MiniMax H3; DreamBench writes, draws and cuts every shot, and you can redo any one.
- Not offered: first or last frames, references sent to H3, video editing, or 3:4.

MiniMax H3 is on every plan. The composer shows the price in credits before you generate; the first time, create a free account and choose a plan. For another model, see the [Grok Imagine 1.5](https://dreambench.ai/guides/grok-imagine-1-5-prompt-guide), [Veo 3.1](https://dreambench.ai/guides/veo-3-1-prompt-guide) or [Kling 3.0](https://dreambench.ai/guides/kling-3-0-prompt-guide) prompt guide; for prompts from words alone, [text to video](https://dreambench.ai/glossary/text-to-video).

## Questions

- **How do you write a prompt for MiniMax H3?** Describe the clip along its timeline: the style and opening frame, each shot’s action, camera move and lines, and a cut time for every later shot. Then add the soundscape and any music only the audience hears.
- **Is Hailuo 3 the same as MiniMax H3?** Yes. MiniMax calls its third-generation video model MiniMax H3; its Hailuo AI app says creators also call it Hailuo 3 or Hailuo H3. It is one model, prompted the same way.
- **What is a good MiniMax H3 prompt example?** 8 seconds, 1:1. Live-action, a close shot of a tea tin reading “HIGHLAND BREAKFAST”. A hand lifts the lid as the camera pushes in slowly. Soundscape: the lid’s pop. Music: none. A style, exact text, one move and the sound.
- **How long should a MiniMax H3 prompt be?** About 60 to 120 words for 5 or 6 seconds, 120 to 250 for 10 and up to 500 for 15. MiniMax’s API caps a prompt at 7,000 characters; DreamBench’s Video page takes up to 5,000.
- **How do I make MiniMax H3 characters talk?** Describe the speaker’s voice once, then write the line in quotes with its language: the baker, a calm, raspy voice, says in English, “First batch of the morning.” H3 voices it and moves the lips in the same generation.
- **What languages can MiniMax H3 speak?** MiniMax lists stable dialogue in 11 languages: Arabic, Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish. Others work to varying degrees, so check each take.
- **How do I write camera movement in a MiniMax H3 prompt?** Name the move, then its size and speed, inside the sentence: “The camera pushes in with small amplitude at slow speed.” MiniMax’s API guide also accepts [pan], [zoom] or [static] after the description it applies to.
- **Can MiniMax H3 make several shots in one clip?** Yes. MiniMax’s prompt guide numbers the shots and times each cut after the first, “[Shot 2] At 00:05.000, the camera cuts to…”, inside one clip of up to 15 seconds. Each cut should bring something new.
- **Does MiniMax H3 have a negative prompt?** MiniMax’s API for H3 has no negative prompt field, so describe what you want: “an empty platform”, not “no people”. For no score, write “Music: none”; MiniMax’s guide writes N/A.
- **How do I write a MiniMax H3 image-to-video prompt?** Start from what the first frame shows, then say how the action begins, develops and ends, with the camera and sound. With first and last frames, describe the path between them, ideally as one shot.
- **Can I use MiniMax H3 reference images on DreamBench?** No. MiniMax’s app and API take up to 9 images, 3 videos and 3 audio clips. DreamBench sends H3 your words, and in a film an opening frame drawn from your pictures, so describe each character identically in every shot.
- **Can I paste my own prompt into MiniMax H3 on DreamBench?** Yes. DreamBench’s Video page sends your prompt to MiniMax H3 as written, at 5 to 15 seconds in 1:1, 16:9, 9:16, 4:3 or 21:9, with stereo sound. First, create a free account and choose a plan.

## Sources

- [MiniMax: H3 video prompt writing guide](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_base_en.md), MiniMax: read 2026-11-01
- [MiniMax: MiniMax H3 model card](https://huggingface.co/MiniMaxAI/MiniMax-H3), MiniMax: read 2026-11-01
- [MiniMax API: Video generation guide](https://platform.minimax.io/docs/guides/video-generation), MiniMax: read 2026-11-01
- [MiniMax API: Create video generation task](https://platform.minimax.io/docs/api-reference/video-generation-v2-create), MiniMax: read 2026-11-01
- [MiniMax API: H3 feature highlights](https://platform.minimax.io/docs/guides/video-prompt), MiniMax: read 2026-11-01
- [MiniMax: MiniMax H3 launch post](https://www.minimax.io/blog/minimax-h3), MiniMax: read 2026-11-01
- [Hailuo AI: MiniMax H3 model](https://hailuoai.video/tools/minimax-h3), Hailuo AI: read 2026-11-01
- [Hailuo AI: Pan, tilt or zoom? Basic AI camera movements](https://hailuoai.video/pages/knowledge/ai-camera-movements-pan-tilt-zoom-guide), Hailuo AI: read 2026-11-01
- [Hailuo AI: How to stop unwanted camera motion in AI prompts](https://hailuoai.video/pages/knowledge/ai-video-camera-motion-control), Hailuo AI: read 2026-11-01
- [Hailuo AI: How to write effective AI video prompts](https://hailuoai.video/pages/blog/effective-ai-video-prompt-guide), Hailuo AI: read 2026-11-01

DreamBench is an independent platform and isn’t affiliated with or endorsed by the companies named on this page.