Guides

    Hailuo 02: silent plate, score later (1080p, 8cr)

    Hailuo 02 has audio off. Plan music, TTS, or a talking model before you call this the hero.

    Versely Team5 min read

    Hailuo 02 has audio off. Plan music, TTS, or a talking model before you call this the hero.

    Hailuo 02 is MiniMax's Hailuo 02 video generator: expressive physical motion and strong prompt-to-action fidelity. A plain-language action description becomes dynamic footage. Category is text-to-video. No image is required. Audio is false. Max output resolution is 1080p. The catalog lists 8 credits. Styles on the row: Cinematic, Documentary, Abstract. The catalog does not publish a duration ladder for this slug.

    That last line matters. You cannot honestly brief "8 seconds of Hailuo 02" from this page. You can honestly brief a silent 1080p action plate at 8 credits listed, and you must plan the sound elsewhere.

    Physical motion, empty soundtrack

    The description is prompt-to-action. Write the verb: the cloth snaps, the bottle tips, the runner plants. Hailuo 02 is the silent plate for that verb. It will not also hum a bed, cast a VO, or lipsync a line. Audio is off. If you generate this and then act surprised at the mute export, you skipped the brief.

    Plan the score before you call the clip the hero:

    • Music bed → generate Hailuo, then add music.
    • Narration → a TTS row, then lay it under the plate.
    • A mouth saying a specific line → this is the wrong hero. Use a talking or lipsync model, or generate this as B-roll only.

    The AI video generator will still let you type. The mix is not in the model. Put the mix on the calendar.

    Eight credits is the listed figure. Billing on the matrix is per second, with a floor of 40 credits in that matrix — do not treat 8 as a complete-job fantasy if the take runs long. The honest habit is: silent action plate, then sound, then decide if this is the hero.

    1080p action from a sentence

    No image is required. That is a real text-to-video row. If a pack must match a photo, you are on the wrong Hailuo (or the wrong family). Lock the still, then image-to-video. Hailuo 02 from text will invent the pack.

    1080p is the ceiling. There is no 4K on this row. Cinematic / Documentary / Abstract are looks, not soundtracks and not identity locks. Documentary will not add location audio. Cinematic will not add a score. Abstract will not hide a mute.

    No durations are listed. Do not copy a 4 / 6 / 8 ladder from another model onto this slug. If the UI later offers a length, treat that as the product surface; this article only quotes the catalog, which leaves the ladder empty. What the catalog does say: 1080p, audio off, 8 credits listed, text-to-video, physical motion.

    Hailuo's roster and best Hailuo model sit next to this row. Best text-to-video is the wider map. This page is the mute rule.

    Mute on purpose, or pick another hero

    A hero with no sound plan is a file you cover with a stock bed at 2am. Fine as B-roll. Not fine as a talking launch film you spent the 8-credit row on because the motion looked expensive.

    Use Hailuo 02 for action you can score later: product physics, a named body move, a silent insert. Do not use it for a presenter, a joke that lives in the line, or a mix you expected for free because "video models have audio now." This one does not. Captions still need words you wrote.

    Play it with the sound off

    That is the whole file. If that is the film, add music or TTS on purpose. If someone is supposed to speak, name the second pass before you generate — or pick another hero.

    FAQ

    Does Hailuo 02 generate native audio?

    No. Audio is marked false. Plan music, TTS, or a talking model. A silent prompt here does not get a surprise soundtrack. It gets silence.

    How long can a Hailuo 02 clip be?

    The catalog does not list durations on this slug. Do not invent a ladder. What is published: text-to-video, 1080p, 8 credits listed, per-second billing in the matrix, prompt-to-action motion. Set length in the product UI, and budget as a timed silent plate, not as a talking take.

    Can I upload a still?

    This row does not require an image. It is text-to-video. If the still is the contract, use an image-to-video model. Feeding a sentence and hoping Hailuo 02 finds your SKU is how a silent plate also becomes a wrong product.

    When is a silent 1080p plate the right hero?

    When the action is the story and the mix is a later, separate credit: music, VO, or a talking cutaway. When the story is the line, Hailuo 02 is B-roll at most. Do not discover that after you have already named it the hero.