Log in
WorkflowBy HypernaturalVersion 7, updated

Multi-style music video

hn/workflow-multi-style-music-video

Make a storytelling music video from a song and photos of the cast, switching looks scene to scene (photoreal, comic-book animation, 3D animation, paper collage) with the cast consistent across styles, lip sync on key lines, a rough cut on the song, and kinetic lyric captions styled to each scene. Use for music videos, lyric videos and mixed-media multi-style shorts.

SKILL.md

The person brings a song, its lyrics and a photo of each person who appears. You deliver a finished 16:9 music video where each scene has its own look, the cast stays recognizable across every look, the singer lip-syncs the key lines, and the captions change style with the scenes. For a drawn video in one or two looks with no lip sync, workflow-animated-music-video.

The song's timing is guide-time-a-video-to-a-song (lyric-slot route); the looks are guide-pick-a-stylized-look. Real cast members appear only from their own photos and with their consent (guide-depict-real-people-and-places). Rounds and deliverables follow the core skill.

Steps

  1. Map the song (guide-time-a-video-to-a-song, Process steps 1–4): stems, the sung words and the timed lyrics. Caption what is sung, not the sheet.
  2. Shot list on the sung lines (guide-time-a-video-to-a-song, Process step 5): per shot its time on the song, its look, the action, BAKED words (drawn into the scene) and LIP (lip-synced). Change looks at the story's turns. Gate: the shot list.
  3. Cast sheets, one per person per look (nano-banana-character-sheet, the person's photo as the only image; each look's line from style-3d-animation or guide-pick-a-stylized-look, references/looks.md). Check every comic sheet for skin tone. Gate: the cast in every look.
  4. Stills, nano banana Pro edit, 16:9, 2K, two takes a shot, only the people in it: their front close-up and full-body views from that look's sheet, then the scene, then "Take each person's identity from their reference images (face, features, skin tone, glasses, hair colour, texture and volume, beard) so they are clearly recognizable, and their outfit as shown, rendered in this frame's style. Do not copy the reference backgrounds.", then the look's line, then "16:9. No captions, subtitles, logos or watermarks." A shot with no cast passes one view with "Image 1 shows the art style to match; do not include this character." Baked words go in quotes with "The only readable words in the frame are "X", spelled exactly." (nano-banana-prompt-format). Conform picks to 1920×1080 (centre crop, Lanczos). Gate: every still.
  5. Animate by look (guide-animate-a-shot, guide-choose-a-video-model), each shot a little past its slot: stylized looks on h3-max-image-to-video at 768P with the look's video phrase; photoreal scenes on kling-v3-image-to-video, generate_audio: false; lip-synced lines on H3 Max with the line's window of the vocal stem as target_audio_url, at least 2 s, and the exact lyric in a <d> block (guide-lipsync-video).
  6. Rough cut on the song (guide-time-a-video-to-a-song, Process step 7, with --seconds): every take on its slot, the song under it. Gate: the cut with the music, before anything is judged clip by clip.
  7. Kinetic lyrics per look (hn4a-compositor-kinetic-lyrics): the type changes with the scene's look; punch words on the hooks. Gate: the live spec (hn4a-compositor, Live spec, then render).
  8. Deliver (guide-prepare-renders-for-delivery).

Choices this workflow makes

  • The lyric-slot route: each shot carries a sung line, so a lip-synced line has its whole line in one shot.
  • H3 Max for every stylized look: its smoothed, painted motion is no problem in a drawn look, and it takes the vocal for lip sync. Kling only where the scene is photoreal.
  • People who are apart in the story stay apart in the frame: a split screen of two rooms, a video call on each laptop, rather than side by side in one place. Edit the picked still into a split screen rather than staging it again.
  • Free shots for the gaps: a flicker through the looks over an instrumental, and a finale split screen of each look's take of the cast waving, played forward then reversed to fill the outro (guide-time-a-video-to-a-song, Choosing).
  • The end title as ransom-note tiles of the song's title over the finale split (hn4a-compositor-kinetic-lyrics).

Dead ends

  • Clips reviewed one by one, silent: they made no sense to the person, who asked for the cut with music as soon as the review opened.
  • Background screens and blimps came back with invented text until the prompt said "plain, no words".
  • Two cast members drifting towards each other read as romantic; a big friendly high-five, then both turning to camera keeping a little space, fixed it.

Worked example

A 2-minute song, four looks (photoreal, ink comic, 3D animation, paper collage), two cast members and 26 shots, about 1,200 credits in all.

StepResult
Cast sheets8 sheets plus 2 redos, about 80 credits
Stills2 takes a shot, about 430 credits; signs, chalkboards, marquees, headlines and neon all spelled right on 2 of 2 takes
Animationabout 700 credits, mostly H3 Max 768P, four Kling shots, one retake; the key lines lip-synced on H3 Max
Cut and captionslocal and free; captions styled per look

Related workflows

FAQs

Make a storytelling music video from a song and photos of the cast, switching looks scene to scene (photoreal, comic-book animation, 3D animation, paper collage) with the cast consistent across styles, lip sync on key lines, a rough cut on the song, and kinetic lyric captions styled to each scene. Use for music videos, lyric videos and mixed-media multi-style shorts.

Install Hypernatural Desktop, a free Mac app, and connect it to Claude Code, Codex or Claude Desktop. Then describe the video you want and your agent finds the workflow itself, or paste this page's link (https://hypernatural.ai/skill/hn/workflow-multi-style-music-video) and it installs hn/workflow-multi-style-music-video and follows it.

It runs fal/fal-ai/demucs, Nano Banana Pro, MiniMax H3 Max and Kling 3.0. Hypernatural Desktop has every one of them built in, so there are no API keys to set up, though you can bring your own.

The workflow is free. You only pay for the AI generations it runs, and your agent shows you the price of every batch before it spends anything. Editing, captions and titles run on your Mac at no cost.

It calls 18 smaller skills for its steps, including Time a video to a song, Pick a stylized look and Style 3d animation. Each one is listed on this page, and your agent loads them as the job needs them.

Yes. Ask your agent to adjust it for your job, or to save its own version. You can publish your version under your own handle so anyone can install it.

Go make something cool.

Install the app, open your agent, describe the video. That's the whole onboarding.

Sticker HypernaturalSticker BannerSticker TVBoy