PixVerse VibeMV: The Essentials

Covers PixVerse VibeMV (model_pixverse-vibemv)

PixVerse VibeMV turns a finished song into a full music video. Upload an audio track, pick a visual style, and the model generates a synced video that matches your song's length automatically, no editing timeline required.

At a Glance

How to Use the Model

How VibeMV Works

There is no text prompt. Instead, VibeMV reads your song's audio and shapes the video around it: energy, tempo, and structure inform scene pacing, while the style parameter sets the visual direction and musicStyle nudges the mood to match the genre. Set both for the most accurate result.

Open this asset in Scenario

Custom style, character photo plus a style reference image, 16:9, 1080p, genre Others. Both the character and the style reference are honored at once: the performer stays on-model while the environment picks up the reference's gold and green marble palette.

Choosing a Visual Style

The style preset is the single biggest driver of look and feel. Some presets stay photoreal (Cinematic), others fully restyle the scene into illustration, claymation, or voxel art, even when the character photo you supply is a realistic headshot. Pick the preset for the aesthetic you want, not for how realistic your reference photo is.

Left: Voxel style, Reggae genre. Right: Woolen Felt style, Hip Hop genre. Open Voxel example · Open Woolen Felt example

Adding a Character

Upload one clear, front-facing photo in image to give the video a consistent performer. Avoid busy backgrounds, group shots, sunglasses, or extreme angles, the same guidance PixVerse gives for its own character consistency feature. The performer's identity holds up across the whole clip, even through fast cuts and style changes.

Open this asset in Scenario

Impressionism style, Soul genre, character photo supplied. The performer's likeness carries through a full jazz-club performance scene.

Leaving image empty does not produce an abstract, performer-free video. VibeMV invents its own on-screen performer instead, and keeps that invented identity consistent for the whole clip. If you want full control over who appears, always supply a character photo.

Open this asset in Scenario

Cinematic style, Pop genre, no character photo supplied. VibeMV generated its own performer and kept her consistent throughout.

Custom Style References

Set style to Custom and supply a styleImage (a texture, mood board, or color palette reference) to art-direct the video beyond the 15 presets. A style reference alone, with no character photo, keeps the video an environment or object shot with no invented performer, useful for abstract or instrumental-feeling videos.

Open this asset in Scenario

Custom style, styleImage only (a cracked asphalt and lightning texture), no character photo, Metal genre. The result stays a pure environment shot, storm clouds over mountains, matching the reference's mood and palette with no invented performer.

Music Genre and Lyrics

musicStyle nudges pacing and visual energy toward the genre you pick, Metal cuts faster and harder than Ambient, for example. lyrics accepts up to 5,000 characters of text, but in testing across 23 generations at every length from a single line to a full multi-verse song, it never rendered as an on-screen caption in the output video. Treat it as reserved for a future captioning feature rather than a working subtitle field today.

Open this asset in Scenario

Futuristic style, Electronic genre, character photo supplied, a full multi-verse lyrics block (over 2,000 characters) passed to the lyrics field. No caption text appears anywhere in the video, confirming the field does not drive on-screen subtitles.

Parameters

Every parameter VibeMV exposes, and what it actually does based on testing.

audio

Required. The song to turn into a music video, 10 seconds to 6 minutes. The output video runs exactly as long as this file. Trim to your strongest section, verse plus chorus is usually enough, since the video length follows the audio 1:1.

image

Optional. A reference photo of the character who appears in the video. See the Adding a Character section above. Use a single, clear, front-facing, well-lit photo.

style

Default Cinematic. One of 15 presets: Cinematic, Lo-fi, Dreamscape, Woolen Felt, Candy, Golden Age, Voxel, Retro Game, Claymation, Woodland Tale, Impressionism, Decadence, Futuristic, Chromatic Clash, and Holiday, or Custom to follow a styleImage reference instead. Presets range from photoreal (Cinematic) to fully illustrated or stylized (Woodland Tale, Claymation, Voxel), independent of how realistic your character photo is.

styleImage

Optional, only used when style is Custom. A texture, mood board, or color reference image. Works alongside image: supplying both at once (see the Custom style example above) keeps the character on-model while the environment adopts the reference's palette and materials.

musicStyle

Default Others. One of Pop, Rock, Hip Hop, R&B, Jazz, Reggae, Country, Folk, Electronic, Classical, Soul, Funk, Metal, or Ambient. Guides pacing and visual energy to match the genre. Set it alongside style for the most accurate result.

lyrics

Optional, up to 5,000 characters. Documented as displaying lyrics as subtitles, but did not render any on-screen caption text in any tested generation regardless of length. See the Music Genre and Lyrics section above.

aspectRatio

Default 16:9. One of 16:9, 9:16, 1:1, 4:3, or 3:4. All five rendered cleanly in testing at both resolutions.

resolution

Default 720p. 720p or 1080p. 1080p costs more and takes longer to render but showed no quality or reliability tradeoff in testing. Consider drafting a style and character combination at 720p, then re-rendering the keeper at 1080p.

Use Cases

• Artist promos: turn a finished track into a shareable music video without booking a shoot, using a character photo of the artist.

• Lyric videos and fan edits: pair a song with a consistent visual identity across a full runtime for YouTube or fan content.

• Social clips: the 9:16 and 1:1 aspect ratios cover TikTok, Reels, and Shorts directly.

• Fictional performers and virtual artists: combine a generated character photo with a Custom style reference to build a distinctive, repeatable visual identity for a virtual musician.

• Game and brand soundtracks: visualize an original score or jingle for a trailer, ad, or in-game cutscene without production.

• Mood boards and pitch decks: quickly generate a visual direction for a track before committing to a full production.

Tips for Better Results

1. Trim to your strongest section. Video length follows audio length exactly, so a tight 20 to 40 second verse-and-chorus clip renders faster and stays focused.

2. Set both style and musicStyle. Leaving either on default produces a less accurate match to your song's genre and mood.

3. Supply a character photo if you want a specific performer. Without one, VibeMV invents its own, consistent but not under your control.

4. Combine image and styleImage under Custom style for maximum art direction. Both references are honored together, character identity plus environment palette and materials.

5. Use a front-facing, well-lit, single-subject photo for image. Busy backgrounds, group shots, and extreme angles weaken identity consistency.

6. Draft at 720p, then re-render the keeper at 1080p. Both resolutions are equally reliable, so save the higher cost for your final pick.

7. Do not rely on lyrics for on-screen captions. It did not render as subtitles in any tested case. Add captions with a dedicated captioning model afterward if you need them.

Known Limitations

• The lyrics field does not produce on-screen subtitles. Confirmed across 23 test generations spanning a single line to a full multi-verse song. If you need captions, generate them separately.

• No character photo means no control over who appears. VibeMV invents its own consistent performer rather than staying abstract. Supply image if the performer's identity matters.

• No explicit prompt field. Creative direction comes entirely from style, styleImage, and musicStyle, not free text. There is no way to describe a scene directly.