Hailuo Director (MiniMax): Camera Movement Prompts for AI Video

Last checked: August 13, 2026.

Current-model notice: MiniMax’s main catalog now lists MiniMax-H3 as the current video model and places Hailuo 2.3, Hailuo 2.3 Fast, and Hailuo 02 under Legacy Models. The legacy V1 text-to-video and image-to-video references still document Hailuo and Director IDs with bracketed [command] camera syntax. This guide therefore covers legacy-compatible camera-command workflows, not the current H3 V2 API.

MiniMax Hailuo Director Models are built for one of the hardest parts of AI video generation: controlling the camera instead of accepting random motion. The original Director models, T2V-01-Director and I2V-01-Director, were released to improve camera-control adherence and cinematic shot composition; MiniMax’s legacy V1 API references also show that Hailuo models such as MiniMax-Hailuo-2.3, MiniMax-Hailuo-2.3-Fast, and MiniMax-Hailuo-02 support bracketed camera commands in specific text-to-video and image-to-video workflows.

This guide explains which model to use, how Hailuo’s [command] syntax works, and how to write practical camera movement prompts for cinematic AI video without overloading the model.

Quick Answer: What Are MiniMax Hailuo Director Models?

MiniMax Hailuo Director Models are Hailuo video-generation models designed to follow explicit camera movement instructions such as [Push in], [Pan left], [Tilt up], [Tracking shot], and [Static shot]. The original Director family includes T2V-01-Director for text-to-video and I2V-01-Director for image-to-video. MiniMax announced them as models with reduced motion randomness, preset camera settings, and stronger prompt adherence for cinematic storytelling.

Director-style camera control is not limited to the original 01 Director IDs in MiniMax’s legacy V1 references. The V1 text-to-video reference lists command support for MiniMax-Hailuo-2.3, MiniMax-Hailuo-02, and T2V-01-Director; its image-to-video reference lists MiniMax-Hailuo-2.3, MiniMax-Hailuo-2.3-Fast, MiniMax-Hailuo-02, and I2V-01-Director. The H3 V2 reference does not document this bracketed command syntax, so do not assume syntax parity.

For legacy camera-command workflows, the practical question is: which documented V1 model matches the required input type, speed, and compatibility? For a new general video integration, evaluate H3 on V2 first.

Which MiniMax Hailuo Model Should You Use?

For a new general video integration, evaluate MiniMax-H3 through POST /v2/video_generation. Use a listed Hailuo or Director model only when you specifically need the bracketed [command] syntax documented on the legacy POST /v1/video_generation route.

Model	Workflow	Best For	Legacy V1 API status
`T2V-01-Director`	Text-to-video	Starting from a written scene and directing camera movement	Officially released as a camera-control model; the legacy V1 T2V reference still lists it as a supported model value.
`I2V-01-Director`	Image-to-video	Animating a first frame while controlling the camera	The legacy V1 I2V reference lists it as a supported model value and supports camera commands.
`MiniMax-Hailuo-2.3`	Text-to-video and image-to-video	Newer within the legacy Hailuo family; general cinematic shots, human motion, and product scenes	MiniMax’s model table describes Hailuo 2.3 as supporting T2V and I2V, with 1080p 6s and 768p 6s/10s at 24 fps.
`MiniMax-Hailuo-2.3-Fast`	Image-to-video	Faster I2V iteration, product tests, ad variations, social clips	MiniMax lists 2.3 Fast as an I2V model focused on value and efficiency; the legacy V1 I2V reference lists it as supporting camera commands.
`MiniMax-Hailuo-02`	Text-to-video, image-to-video, and some start/end-frame workflows	Legacy projects, 512p options, and start/end-frame use cases where supported	MiniMax describes Hailuo 02 as a legacy video model with 1080p 6s, 768p 6s/10s, and 512p 6s/10s at 24 fps.

Simple recommendation:
Use text-to-video when you are inventing a scene from scratch. Use image-to-video when composition, product shape, character appearance, or brand visuals matter. For controlled commercial visuals, I2V is often the more composition-anchored starting point because the first frame anchors the subject, lighting, and composition.

How Hailuo Camera Command Syntax Works

Hailuo camera control uses bracketed commands inside the prompt:

[Push in] A luxury ceramic coffee cup on a stone table, morning window light, steam rising slowly, shallow depth of field, cinematic product shot.

MiniMax’s current docs say prompts can be up to 2,000 characters, and camera movement can be controlled with [command] syntax in supported models. The docs also state that free-form natural language can work, but explicit commands usually produce more accurate results.

The Three Command Patterns

1. Single movement

Use one command when the shot has one clear camera idea.

[Push in] A chef places a finished dessert on a marble counter, soft warm kitchen light, shallow depth of field, elegant commercial food cinematography.

2. Combined movement

Put multiple commands inside one bracket when they should happen at the same time. MiniMax recommends a maximum of three combined movements.

[Pan left,Pedestal up] A wide mountain valley at sunrise, mist moving between pine trees, cinematic landscape reveal, natural golden light.

3. Sequential movement

Place commands in the order you want them to happen.

[Push in] A detective studies a torn photograph under a desk lamp, then [Static shot] the camera holds as his expression changes from doubt to recognition.

For short 6–10 second clips, do not try to direct a full scene with five camera beats. One primary camera move and one subject action usually works better than a crowded prompt.

Official Hailuo Camera Movement Commands

MiniMax’s API docs list 15 supported camera commands. Use the exact bracket format for official command prompting.

Command	What It Does	Best Use Case	Prompt Tip
`[Truck left]`	Moves the camera left	Sideways product reveal, moving through a space	Keep the subject’s direction clear so lateral motion does not feel random.
`[Truck right]`	Moves the camera right	Fashion walk, car reveal, interior scan	Use stable background details to help depth.
`[Pan left]`	Rotates the camera left from a fixed position	Revealing a landscape or room	Best for wide scenes where the camera surveys the environment.
`[Pan right]`	Rotates the camera right from a fixed position	Stage reveal, product lineup, city view	Avoid too much subject movement during the pan.
`[Push in]`	Moves the camera toward the subject	Tension, focus, product hero shot	Strong for emotional close-ups and premium product shots.
`[Pull out]`	Moves the camera away from the subject	Revealing scale, isolation, context	Start with a clear close subject and describe what appears as the frame widens.
`[Pedestal up]`	Moves the camera vertically upward	Hero reveal, architecture, tall object	Works well when the subject has vertical scale.
`[Pedestal down]`	Moves the camera vertically downward	Descending reveal, tabletop product, overhead shift	Use with simple compositions to avoid warping.
`[Tilt up]`	Tilts the camera upward	Revealing height, buildings, trees, heroic posture	Keep the camera position stable and describe the reveal.
`[Tilt down]`	Tilts the camera downward	Revealing an object on a table, ground detail, clue	Good for suspense, product details, and overhead transitions.
`[Zoom in]`	Simulates lens zoom toward detail	Detail emphasis, UI/product feature, face reaction	Use when you want magnification rather than physical camera travel.
`[Zoom out]`	Simulates lens zoom away	Context reveal, ending shot, environment reveal	Works best with a strong central subject.
`[Shake]`	Adds shaky movement	Impact, fear, action, handheld energy	Use sparingly; too much action plus shake can look unstable.
`[Tracking shot]`	Follows a moving subject	Walking, running, vehicle movement, performer motion	Describe the subject’s path clearly.
`[Static shot]`	Holds the camera still	Interviews, product display, delicate animation	Useful when the subject moves but the camera should not drift.

Important: Do Not Treat Every Cinematic Term as an Official Command

Terms like “dolly zoom,” “orbit,” “crane shot,” “steadicam,” “handheld,” and “drone shot” can be useful natural-language descriptors, but they are not part of the 15 official bracket commands unless MiniMax’s current docs list them as commands. Use them as descriptive language, not as verified bracket commands.

Example:

[Pull out] A lone astronaut stands on a silent red desert, the camera pulls away to reveal a vast crater city in the distance, slow drone-like feeling, cinematic sci-fi lighting.

Here, [Pull out] is the official command; “drone-like feeling” is descriptive style language.

A Practical Prompt Formula for Cinematic Hailuo Video

A strong Hailuo Director-style prompt usually follows this structure:

[Camera command] + subject + subject motion + environment + visual style + lighting + mood + constraint

Example:

[Push in] A glass perfume bottle stands on a wet black stone surface, tiny water droplets glisten as soft mist drifts behind it, dramatic side lighting, luxury commercial cinematography, shallow depth of field, no text, no logo.

Why this works:

The camera command comes first, so the model receives the motion instruction early. The subject is concrete. The action is simple. The environment gives depth cues. The lighting and style guide the final look. The negative constraint removes unwanted text or logos.

Keep One Primary Motion

For short AI video clips, the safest rule is:

One camera move + one subject action + one atmosphere

Weak prompt:

[Pan left,Push in,Tilt up,Shake] A model walks through a futuristic city while rain explodes, lights flash, the camera spins, zooms, shakes, and follows her dramatically.

Improved prompt:

[Tracking shot] A model in a silver raincoat walks through a neon city alley, wet pavement reflecting pink and blue signs, slow confident movement, cinematic cyberpunk lighting, stable camera, realistic motion.

The improved version gives the model a clear subject, one dominant camera behavior, one environment, and a realistic motion target.

Copyable MiniMax Hailuo Camera Movement Prompts

Use these as starting templates. Replace the subject, environment, style, and action to fit your scene.

1. Cinematic Product Push-In

Camera move: [Push in]
Best for: Ads, hero shots, premium product visuals.

[Push in] A matte black wireless speaker sits on a dark wooden table, soft dust particles floating in a beam of warm window light, the speaker vibrates subtly as music plays, cinematic product photography, shallow depth of field, premium minimal mood, no text, no logo.

2. Luxury Detail Zoom

Camera move: [Zoom in]
Best for: Jewelry, watch details, texture, packaging.

[Zoom in] A silver mechanical watch rests on deep blue velvet, the second hand moves smoothly, tiny reflections travel across the polished metal, macro lens look, elegant studio lighting, crisp detail, luxury commercial mood.

3. Wide Establishing Pan

Camera move: [Pan right]
Best for: Landscapes, architecture, world-building, scene openings.

[Pan right] A quiet coastal village at sunrise, fishing boats floating near the harbor, orange light reflecting on calm water, birds crossing the sky, cinematic wide establishing shot, natural colors, peaceful atmosphere.

4. Vertical Hero Reveal

Camera move: [Pedestal up]
Best for: Tall products, statues, buildings, fashion reveals.

[Pedestal up] A modern electric motorcycle parked in a clean industrial studio, the camera rises from the wheels to the handlebars, soft overhead reflections on the metal body, dramatic commercial lighting, precise and stable motion.

5. Suspenseful Push-In

Camera move: [Push in]
Best for: Thriller scenes, emotional beats, dramatic reveals.

[Push in] A young investigator sits alone in a dim archive room, staring at an old map under a green desk lamp, dust floating in the air, her expression shifts from confusion to realization, quiet suspense, cinematic shadows, realistic face movement.

6. Static Shot for Subject Stability

Camera move: [Static shot]
Best for: Character consistency, delicate animation, product clarity.

[Static shot] A ceramic tea cup on a small wooden tray, steam rises slowly and curls in the air, morning light passes through white curtains, calm minimal composition, realistic steam motion, no camera drift.

7. Tracking a Moving Subject

Camera move: [Tracking shot]
Best for: Walking scenes, dancers, vehicles, animals.

[Tracking shot] A cyclist rides along a tree-lined road in early autumn, the camera follows smoothly from the side, golden leaves move in the wind, natural sunlight, realistic speed, cinematic outdoor commercial style.

8. Pull-Out Scale Reveal

Camera move: [Pull out]
Best for: Revealing environment, loneliness, scale, final shots.

[Pull out] A lone hiker stands on a narrow snowy ridge, the camera slowly moves backward to reveal a vast mountain range under pale morning light, wind blowing snow gently, epic cinematic scale, realistic atmosphere.

9. Controlled Handheld Energy

Camera move: [Shake]
Best for: Impact, action, emergency, chaos.

[Shake] A rescue worker runs through a smoky industrial corridor, warning lights flashing, the camera shakes slightly as sparks fall in the background, urgent documentary style, dramatic but readable motion.

Use [Shake] carefully. If the subject is already moving fast, add words like “slightly,” “controlled,” or “readable” to reduce the chance of chaotic motion.

10. Combined Reveal Shot

Camera move: [Pan left,Pedestal up]
Best for: Scenic reveal, architecture, fantasy environments.

[Pan left,Pedestal up] A grand glass greenhouse filled with rare tropical plants, the camera rises while slowly panning left to reveal sunlight pouring through the roof, soft mist, lush cinematic atmosphere, smooth controlled motion.

Text-to-Video vs Image-to-Video: Which Is Better for Camera Control?

Both workflows can use camera prompts, but they solve different problems. MiniMax’s video guide describes text-to-video as generating directly from a text description, while image-to-video uses a first_frame_image as the opening frame and the prompt describes how the scene evolves from that image.

Use Case	Better Workflow	Why
You want to invent a new scene from scratch	Text-to-video	No source image is needed; faster for ideation.
You need product shape consistency	Image-to-video	The first frame anchors the product’s appearance.
You need a character to remain visually consistent	Image-to-video or subject-reference workflow	A reference helps reduce identity drift.
You need many ad variations from one approved visual	Image-to-video	Reuse one master frame and test different camera moves.
You need a precise beginning and ending frame	First-and-last-frame workflow, where supported	Hailuo 02 supports start/end-frame features in specific contexts.

A practical production workflow is:

Create or choose a clean first frame.
Decide the emotional purpose of the camera move.
Use one official bracket command.
Add only the subject motion needed for the shot.
Generate several variations.
Pick the most stable result.
Refine the prompt, not everything at once.

Legacy V1 API Notes for Developers

Legacy V1 example: the examples below use Hailuo models on the older V1 route because that is where MiniMax documents bracketed camera commands. H3 uses a different V2 content-array schema.

MiniMax’s video-generation workflow is asynchronous: create a task, poll the task status with the returned task_id, and retrieve the file once the task succeeds.

A simplified text-to-video request looks like this:

{
  "model": "MiniMax-Hailuo-2.3",
  "prompt": "[Push in] A ceramic coffee cup on a stone table, steam rising slowly, warm morning light, cinematic product shot.",
  "duration": 6,
  "resolution": "1080P",
  "prompt_optimizer": false
}

In MiniMax’s API docs, prompt_optimizer defaults to true, and setting it to false can give more precise control when you want the model to follow your prompt more literally.

For image-to-video, you also provide a first_frame_image. MiniMax’s I2V docs say the image can be a public URL or a Base64 data URL, with accepted formats including JPG, JPEG, PNG, and WebP, file size below 20MB, short edge above 300px, and aspect ratio between 2:5 and 5:2.

{
  "model": "MiniMax-Hailuo-2.3-Fast",
  "first_frame_image": "https://example.com/product-frame.png",
  "prompt": "[Push in] The camera moves toward the product as soft reflections travel across the surface, premium studio lighting, clean commercial style.",
  "duration": 6,
  "resolution": "768P",
  "prompt_optimizer": false
}

How to Choose the Right Camera Movement

Camera movement should serve the story. Do not choose [Push in] because it sounds cinematic; choose it because the viewer needs to feel closer to the subject.

Goal	Use This Move	Why It Works
Make a product feel premium	`[Push in]` or `[Zoom in]`	Draws attention to texture, shape, and details.
Reveal a large environment	`[Pan left]`, `[Pan right]`, or `[Pull out]`	Expands context and scale.
Keep a product stable	`[Static shot]`	Reduces unnecessary camera drift.
Follow a moving subject	`[Tracking shot]`	Keeps movement readable.
Reveal height or power	`[Tilt up]` or `[Pedestal up]`	Creates vertical emphasis.
Create unease	`[Push in]` or controlled `[Shake]`	Builds tension without needing complex action.
Show before/after context	`[Pull out]`	Starts intimate, ends wide.
Make a scene feel observational	`[Static shot]`	Lets subject motion carry the scene.

Troubleshooting: Why Hailuo Ignores or Weakens Camera Commands

Even with Director-style commands, AI video generation is probabilistic. Camera instructions can be weakened by prompt conflicts, excessive subject motion, unclear environments, or too many commands.

Problem: The Camera Move Is Ignored

Likely causes:

The command appears too late in the prompt.
The prompt contains several competing camera instructions.
The subject action is too complex.
The model’s prompt optimizer rewrote or diluted the command.

Fix:

[Push in] A single red sports car parked in an empty studio, the camera slowly moves toward the front grille, glossy reflections, dramatic controlled lighting, no other motion.

Put the command first. Use one subject. Use one camera direction. Consider setting prompt_optimizer to false in API workflows when exact control matters.

Problem: The Shot Warps or Melts

Likely causes:

Too much motion in a close-up.
The subject has complex hands, faces, reflections, or text.
Camera movement and subject movement conflict.

Fix:

[Static shot] A handmade ceramic vase on a simple wooden table, soft window light, only the flower petals move gently in the breeze, clean background, realistic texture.

Reduce the camera motion first. If the subject still warps, simplify the subject action.

Problem: The Scene Feels Random

Likely causes:

No clear subject hierarchy.
Too many style words.
The prompt describes mood but not motion.
The environment lacks spatial anchors.

Fix:

[Tracking shot] A delivery robot moves slowly down a bright hospital corridor, the camera follows from behind at waist height, blue floor lines guide the path, clean white walls, realistic lighting, smooth motion.

Add stable spatial cues: corridor lines, table edges, windows, shelves, horizon, road, or background lights.

Problem: The Camera Moves But the Subject Drifts

Likely causes:

The prompt does not say what the subject should keep doing.
The camera instruction is clear, but the subject instruction is vague.
The first frame is not clean enough in I2V.

Fix:

[Tracking shot] A golden retriever trots steadily along a park path, always centered in frame, the camera follows smoothly from the side, trees and benches pass slowly in the background, natural daylight.

Use phrases such as “centered in frame,” “stays in place,” “moves steadily,” or “camera follows from the side” when subject position matters.

Common Prompt Mistakes to Avoid

1. Stacking Too Many Commands

Weak:

[Push in,Pan left,Tilt up,Zoom in,Shake] A fashion model walks through a futuristic hallway.

Better:

[Tracking shot] A fashion model walks through a futuristic hallway, the camera follows smoothly from the front, soft blue light strips reflect on the floor, confident pace, clean cinematic motion.

MiniMax recommends no more than three commands inside one combined movement bracket; in practice, one or two is often easier to control.

2. Mixing Opposite Motions

Weak:

[Push in] The camera pulls far away from the subject while zooming in closely.

Better:

[Push in] The camera moves slowly toward the subject’s face as her expression becomes serious, dim office light, cinematic drama.

3. Making the Subject and Camera Compete

Weak:

[Pan right] A dancer spins rapidly, jumps, flips, turns toward camera, changes outfit, and runs through a crowd.

Better:

[Pan right] A dancer performs one slow turn on an empty stage, warm spotlight, fabric moving gently, elegant cinematic performance, smooth readable motion.

4. Asking for Logos, Text, or Brand Marks

AI video models can struggle with exact text and logos. For commercial work, use clean visuals, add official logos in post-production, and avoid generating brand assets you do not have rights to use.

Best Practices for Cinematic Hailuo AI Video

Start With the Shot Type

Decide whether the shot is:

Establishing shot
Product hero shot
Character close-up
Action follow shot
Reveal shot
Detail insert
Static atmospheric shot

Then choose the camera command.

Use Physical Language

Instead of “make it cinematic,” describe what the viewer sees:

soft backlight, shallow depth of field, slow steam movement, warm highlights on glass, realistic shadows

Anchor the Environment

Good camera motion needs depth cues. Add objects that help the model understand space:

foreground leaves, long corridor lines, window reflections, road markings, shelves, lamp glow, mountain horizon

Keep the Clip Length in Mind

Hailuo clips are short. Current MiniMax model documentation lists common 6-second and 10-second output options for newer Hailuo models, depending on model and resolution.

A 6-second clip should not include a full narrative arc. It should capture one visual beat.

Generate Variations Intentionally

Do not rewrite the whole prompt after one failed result. Change one variable at a time:

Camera move
Subject action
Motion intensity
Lighting
Model
Prompt optimizer setting
First frame

This makes it easier to learn what improved the result.

Commercial Use, Watermarks, and Rights: Be Careful

Hailuo’s paid subscription terms say free users can download videos with a watermark, while paid users receive additional benefits such as watermark-free downloads and priority generation. The same subscription terms state that Hailuo does not claim ownership over content generated and downloaded while on a paid subscription plan, and that users retain intellectual property rights including commercial use rights as between the user and Hailuo.

That does not mean every output is automatically safe for every commercial use. Hailuo’s Terms of Service also say users are responsible for their submitted and generated content, and content must not infringe copyright, trademark, publicity, privacy, or other rights.

For client or ad work:

Check the latest Hailuo and MiniMax terms before publishing.
Avoid celebrity likenesses, private individuals, copyrighted characters, and third-party logos.
Add exact brand marks, text, disclaimers, and legal copy in post-production.
Keep records of prompts, source images, rights, and subscription status.
Treat this as production guidance, not legal advice.

MiniMax Hailuo Director Prompt Templates

Use these templates as reusable building blocks.

Template 1: Product Hero Shot

[Push in] A [product] placed on [surface], [small natural motion], [lighting], [background], premium commercial cinematography, shallow depth of field, clean composition, no text, no logo.

Example:

[Push in] A white ceramic skincare jar placed on a smooth limestone slab, soft mist moving behind it, warm sunrise light, minimal beige background, premium commercial cinematography, shallow depth of field, clean composition, no text, no logo.

Template 2: Character Close-Up

[Static shot] A [character description] sits/stands in [environment], [subtle facial or body motion], [lighting], [mood], realistic expression, stable camera.

Example:

[Static shot] A tired botanist sits in a greenhouse at night, gently touching a glowing leaf while her eyes widen with curiosity, soft green reflections on her face, quiet magical realism, realistic expression, stable camera.

Template 3: Tracking Shot

[Tracking shot] A [subject] moves through [environment], the camera follows from [angle], [background movement], [lighting/style], smooth readable motion.

Example:

[Tracking shot] A small delivery robot moves through a modern airport terminal, the camera follows from a low front angle, travelers pass softly in the background, bright natural skylight, smooth readable motion.

Template 4: Establishing Reveal

[Pan right] A wide view of [location], [foreground detail], [background detail], [time of day], cinematic establishing shot, natural atmosphere.

Example:

[Pan right] A wide view of a desert research station, red dust moving across the foreground, solar panels stretching toward distant cliffs, late afternoon sun, cinematic establishing shot, natural atmosphere.

Template 5: Suspense Beat

[Push in] A [subject] notices [small clue], [environment], [subtle reaction], low-key lighting, tense cinematic atmosphere, controlled motion.

Example:

[Push in] A museum guard notices a tiny crack spreading across a glass display case, empty gallery at midnight, his breathing slows as he leans closer, low-key lighting, tense cinematic atmosphere, controlled motion.

MiniMax Hailuo Director Models FAQ

Are T2V-01-Director and I2V-01-Director still available?

MiniMax’s current API docs still list T2V-01-Director as a supported text-to-video model and I2V-01-Director as a supported image-to-video model. The newer docs also show Director-style command support on MiniMax-Hailuo-2.3, MiniMax-Hailuo-2.3-Fast for I2V, and MiniMax-Hailuo-02 in relevant workflows.

Which Hailuo model is best for camera movement prompts?

For a new video integration, evaluate MiniMax-H3 through POST /v2/video_generation. If you specifically need the bracketed [command] syntax documented on POST /v1/video_generation, use a listed legacy Hailuo or Director model. Do not imply that the H3 V2 reference supports the same syntax.

What are the supported Hailuo camera commands?

The official command list includes [Truck left], [Truck right], [Pan left], [Pan right], [Push in], [Pull out], [Pedestal up], [Pedestal down], [Tilt up], [Tilt down], [Zoom in], [Zoom out], [Shake], [Tracking shot], and [Static shot].

Can I combine multiple Hailuo camera commands?

Yes. MiniMax’s docs say multiple commands inside one bracket take effect simultaneously, with a recommended maximum of three, and commands can also be used sequentially in the order they appear.

Should I use `[Push in]` or `[Zoom in]`?

Use [Push in] when you want the camera to feel like it physically moves toward the subject. Use [Zoom in] when you want the lens to magnify the subject from a fixed position. In AI video, both may sometimes look similar, so test both for important shots.

Why does Hailuo ignore my camera prompt?

Usually because the prompt has too many competing instructions, the command appears too late, the subject action is too complex, or the prompt optimizer changes the instruction. Put the command first, simplify the scene, use one main camera move, and consider setting prompt_optimizer to false in API workflows when exact control matters.

Can I use MiniMax Hailuo Director videos commercially?

Hailuo’s paid subscription terms state that paid users retain intellectual property rights to generated and downloaded content, including commercial use rights as between the user and Hailuo. Still, this does not mean every output is automatically legally cleared for every commercial use. Commercial use depends on your plan, the current terms, your source images, likeness permissions, brand assets, and whether the output infringes third-party rights. Always check the latest terms before client or ad use.

Is image-to-video better than text-to-video for product videos?

Often, yes. Image-to-video lets you start from an approved product render or photo, which can help preserve product shape, lighting, and composition. MiniMax’s image-to-video workflow uses the provided first frame as the video’s opening frame, then uses the prompt to describe how the scene evolves.

More on Hailuo: see the Hailuo AI hub for every model compared side by side, or Is Hailuo AI free? for what each access route actually costs.

Quick Answer: What Are MiniMax Hailuo Director Models?

Which MiniMax Hailuo Model Should You Use?

How Hailuo Camera Command Syntax Works

The Three Command Patterns

Official Hailuo Camera Movement Commands

Important: Do Not Treat Every Cinematic Term as an Official Command

A Practical Prompt Formula for Cinematic Hailuo Video

Keep One Primary Motion

Copyable MiniMax Hailuo Camera Movement Prompts

1. Cinematic Product Push-In

2. Luxury Detail Zoom

3. Wide Establishing Pan

4. Vertical Hero Reveal

5. Suspenseful Push-In

6. Static Shot for Subject Stability

7. Tracking a Moving Subject

8. Pull-Out Scale Reveal

9. Controlled Handheld Energy

10. Combined Reveal Shot

Text-to-Video vs Image-to-Video: Which Is Better for Camera Control?

Legacy V1 API Notes for Developers

How to Choose the Right Camera Movement

Troubleshooting: Why Hailuo Ignores or Weakens Camera Commands

Problem: The Camera Move Is Ignored

Problem: The Shot Warps or Melts

Problem: The Scene Feels Random

Problem: The Camera Moves But the Subject Drifts

Common Prompt Mistakes to Avoid

1. Stacking Too Many Commands

2. Mixing Opposite Motions

3. Making the Subject and Camera Compete

4. Asking for Logos, Text, or Brand Marks

Best Practices for Cinematic Hailuo AI Video

Start With the Shot Type

Use Physical Language

Anchor the Environment

Keep the Clip Length in Mind

Generate Variations Intentionally

Commercial Use, Watermarks, and Rights: Be Careful

MiniMax Hailuo Director Prompt Templates

Template 1: Product Hero Shot

Template 2: Character Close-Up

Template 3: Tracking Shot

Template 4: Establishing Reveal

Template 5: Suspense Beat

MiniMax Hailuo Director Models FAQ

Are T2V-01-Director and I2V-01-Director still available?

Which Hailuo model is best for camera movement prompts?

What are the supported Hailuo camera commands?

Can I combine multiple Hailuo camera commands?

Should I use [Push in] or [Zoom in]?

Why does Hailuo ignore my camera prompt?

Can I use MiniMax Hailuo Director videos commercially?

Is image-to-video better than text-to-video for product videos?

Should I use `[Push in]` or `[Zoom in]`?