Realistic Hailuo video comes from one clear structure and restraint, not from stacking adjectives. Describe a single subject doing one action in a defined place, name one or two camera moves in brackets, set the light, and add a style word. Keep the whole thing around 40 to 60 words. This guide breaks down the six-part formula for Hailuo 2.3, shows how to steer the Director camera, and gives 10 paste-ready prompts.

Advertisement

The 6-part Hailuo formula

Realistic shots follow this order: Subject + Action + Setting + Camera + Lighting + Style. Give the model exactly one thing to animate and one or two moves for the lens, then stop.

1. Subject. Name who or what, with a few concrete visual details. Example fragment: "a weathered fisherman in a yellow oilskin coat, grey stubble."

2. Action — one main action. Choose a single verb. Hailuo 2.3 has strong physics, but stacking actions still causes artifacts. Example fragment: "slowly hauling a wet rope hand over hand."

3. Setting. Ground the subject in a real place. Example fragment: "on the deck of a small trawler, open grey sea behind him."

4. Camera — bracket move + shot size. Put the Director camera move in square brackets at the very start, then name the framing. Example: "[Push in] medium shot." Combine up to three brackets, applied in order.

5. Lighting. Be specific; vague light reads flat and fake. Example fragment: "overcast morning light, soft and cool with faint rim light on the coat."

6. Style. One or two words steer color and grade. Example fragment: "documentary, muted color."

Put together: "[Push in] A weathered fisherman in a yellow oilskin coat slowly hauls a wet rope hand over hand on the deck of a small trawler, open grey sea behind him. Medium shot. Overcast morning light, soft and cool. Documentary, muted color." That is the whole prompt — one action, one move, roughly 45 words. No duration, no resolution, no aspect ratio in the text.

For faster iteration, keep the formula in a reusable structure. Our Hailuo prompt cheat sheet and Hailuo prompt templates lay out fill-in-the-blank versions of each part.

Driving the Director camera

Hailuo's Director model reads camera moves written in square brackets at the start of the prompt. Named moves are far more reliable than describing motion in prose, which the model often blurs or ignores.

The available moves are [Push in], [Pull out], [Pan left], [Pan right], [Truck left], [Truck right], [Pedestal up], [Pedestal down], [Tilt up], [Tilt down], [Zoom in], [Zoom out], [Tracking shot], [Static shot], and [Shake]. You can combine up to three, and Hailuo runs them in the order you write them — so [Truck left][Pan right] slides left while turning right.

Keep it to one or two moves for realistic footage. Three is possible but raises the odds of drift. When you want a locked, stable shot, write [Static shot] explicitly rather than leaving the camera unspecified.

Text-to-video vs image-to-video

For realism, prefer image-to-video whenever you have a clean reference. Image-to-video locks the subject's look and structure from your source frame, which sharply reduces warping and keeps faces and products consistent.

In image-to-video, the source already carries the scene, so describe only the motion and camera — re-describing the setting fights the reference and introduces artifacts. Text-to-video defaults to 16:9 (9:16 and 1:1 are also available), while image-to-video follows the aspect ratio of your source image.

Use text-to-video when you have no usable frame or need full freedom over the subject. Write the same six-part prompt either way, and if you also work in Kling, compare the approach in how to prompt Kling for realistic video.

10 example prompts

Each prompt below is complete and paste-ready. Drop it into Hailuo 2.3, keep the camera brackets at the start, and set resolution and duration in the panel.

1. Fisherman on deck

[Push in] A weathered fisherman in a yellow oilskin coat with grey stubble slowly hauls a wet rope hand over hand on the deck of a small trawler, open grey sea behind him. Medium shot. Overcast morning light, soft and cool with faint rim light on the coat. Documentary, muted color.

Why it works: one action (hauling) plus a single [Push in] keeps skin and fabric physics believable. (16:9, 1080p, 6s)

2. Barista pouring latte art

[Static shot] A young barista in a black apron carefully pours steamed milk into a ceramic cup, forming a leaf pattern, at a marble cafe counter with a chrome espresso machine behind. Close-up on the cup and hands. Warm window light from the left, soft shadows. Calm, artisanal, natural color.

Why it works: the [Static shot] plus one delicate action keeps liquid physics clean and avoids hand warping. (9:16, 1080p, 6s)

3. Runner at dawn

[Tracking shot] A woman in grey running gear jogs along an empty coastal boardwalk, ocean to her right. Wide shot moving alongside her. Golden dawn light with long shadows and gentle lens flare. Fresh, energetic, cinematic.

Why it works: a single [Tracking shot] matched to the subject's motion reads smooth and real. (16:9, 768p, 10s)

4. Chef flipping vegetables

[Shake] A chef in a white jacket tosses sliced vegetables in a steel wok over a gas flame in a busy restaurant kitchen. Medium close-up. Warm tungsten light with a bright flame flare. Fast, appetizing, energetic.

Why it works: [Shake] adds handheld life without a second move, and one flame source anchors the lighting. (16:9, 1080p, 6s)

5. Vintage car on a mountain road

[Pedestal up][Pan right] A red 1960s convertible drives along a winding mountain road with pine forest on both sides. Wide aerial view revealing the valley. Late-afternoon sun, warm and hazy with soft atmosphere. Nostalgic, cinematic.

Why it works: two ordered moves build scale while the car does the only action. (16:9, 768p, 10s)

6. Rain on a city window

[Static shot] Rain streaks slide down a cafe window with a blurred neon street beyond, a coffee cup steaming on the sill in the foreground. Close-up, shallow depth of field. Cool blue night light with warm neon reflections. Moody, quiet, teal-and-orange grade.

Why it works: a locked [Static shot] and shallow focus let the rain and steam carry realistic micro-motion. (16:9, 1080p, 6s)

7. Product hero: sneaker turntable

[Truck left] A white leather sneaker sits on a matte grey pedestal in a studio. Slow arc around the shoe. Soft, even softbox lighting with a subtle gradient background. Clean, premium, minimal.

Why it works: a single slow [Truck left] on a still subject is the safest realistic product move; start from a product photo in image-to-video for best results. (1:1, 1080p, 6s)

8. Child blowing bubbles

[Pull out] A small child in a striped shirt blows a stream of soap bubbles in a sunlit backyard, green grass and a wooden fence behind. Medium shot. Bright afternoon sunlight, warm and soft with bokeh. Joyful, tender, nostalgic.

Why it works: one gentle action and a single [Pull out] keep the face stable and expressive. (16:9, 1080p, 6s)

9. Blacksmith at the forge

[Static shot] A muscular blacksmith hammers a glowing orange blade on an anvil in a dark stone workshop, sparks flying. Medium close-up. Firelight from the forge, deep shadows with a strong orange glow on the face. Intense, gritty, dramatic.

Why it works: firelight as the only source plus one repeating action produces convincing sparks and skin. (16:9, 1080p, 6s)

10. Astronaut looking out a window (vertical)

[Push in] An astronaut in a white suit slowly turns to look out a spacecraft window at Earth below. Medium shot toward the visor. Cool blue Earthlight on the helmet with soft interior fill. Awe, calm, cinematic.

Why it works: a slow [Push in] and one subtle turn read as weightless and real; the 9:16 frame suits vertical feeds. (9:16, 768p, 10s)

Advertisement

Common mistakes to avoid

Most unrealistic results trace back to a handful of habits. Fix these first, then browse the best Hailuo prompts roundup for more ready-made shots.

Stacking multiple actions. "Walks, waves, and sits down" forces the model to guess timing and it warps. Keep one main action per shot.

Vague camera direction. Prose like "the camera moves around" is unreliable. Use a bracketed Director move such as [Truck left] at the start instead.

Over-long prompts. Past roughly 60 words the main action gets buried and artifacts creep in. Trim to one or two tight sentences.

Expecting audio. Hailuo 2.3 is silent. Describing sounds wastes words; add music and effects in an editor afterward.

Combining 1080p with 10 seconds. 1080p is capped at 6 seconds. If you need 10 seconds, render at 768p — the panel will not let you have both.

Negative phrasing. "No blur, not dark, without warping" can confuse the model into producing the very thing you named. Describe what you want positively instead.

Re-describing the scene in image-to-video. The source frame already sets the look. In I2V, prompt only the motion and camera, or the text fights the image. For finished setups built on these rules, see our Hailuo cinematic b-roll prompts.

Frequently Asked Questions

How long should a Hailuo prompt be?

About 40 to 60 words. Cover the six parts of the formula in one or two tight sentences with a single main action and one or two camera moves. Over-stuffing a prompt causes artifacts and drift, so trim anything that is not doing work.

How do I control the camera in Hailuo?

Use the Director model's bracketed camera moves at the very start of the prompt, such as [Push in] or [Tracking shot]. You can combine up to three, and Hailuo applies them in the order written. Named moves are far more reliable than describing motion in plain prose.

Does Hailuo generate audio?

No. Hailuo 2.3 outputs silent video. Add music, dialogue, or sound effects in a separate editor afterward. Do not waste prompt words describing sounds, since the model ignores them.

Should I render at 768p or 1080p?

Use 1080p for final delivery and hero shots where detail matters, and 768p for fast drafts and iteration. Remember that 1080p is capped at 6 seconds, so a 10-second clip must be rendered at 768p.

Should I pick 6 seconds or 10 seconds?

Choose 6 seconds for a single clean beat such as one action and one camera move, and 10 seconds when the shot needs room to breathe. Note that 1080p only supports 6 seconds, so 10-second clips render at 768p.

Text-to-video or image-to-video for realism?

Image-to-video gives the most reliable realism because the source image locks the subject's look and structure. In that mode, describe only the motion and camera, not the scene. Use text-to-video when you have no usable reference frame.

What is the difference between the Standard and Fast variants?

Standard prioritizes maximum quality and physics accuracy, while Fast trades a little fidelity for quicker, cheaper generations. Draft with Fast, then finalize your chosen shot on Standard at 1080p.

Advertisement