16-Shot Storyboard Prompt Template for AI Music Videos

Nov 16, 2026shaomingchan

A copy-ready prompt template for AI music video storyboards: the five-part prompt formula, a 16-shot framework pre-mapped to song sections, adjustment rules, and the negative prompt list.

Storyboard prompt quality is seven parts structure, three parts wording. Below is a copy-ready template: a five-part prompt formula, a 16-shot framework pre-mapped to song sections, and the negative-prompt list that keeps frames from falling apart. Swap in your song's content and go.

The five-part prompt formula

One storyboard prompt, five segments, fixed order:

[subject] + [scene] + [action] + [camera/framing] + [lighting/style anchor]

Subject references a character anchor instead of describing the face in words ("singer #A1", not "a man in a black suit" — free-text faces re-roll every generation). Scene gets execution-grade granularity ("neon rain-soaked street", never "somewhere atmospheric"). Action is a verb phrase one shot can complete ("turns with an umbrella", not "lives a whole life"). Camera/framing uses industry terms (dolly, handheld, wide, over-the-shoulder) — models respond to these far more stably than prose. Lighting/style anchor is one identical string across every shot (e.g. "Kodak 5207 film grain, low-saturation night palette") — it is the fuse that keeps the whole film in one color world.

Each segment swaps independently. That is the point of a template: a new song changes content, not structure.

The 16-shot framework (copy and fill)

Pre-mapped to a 3-minute pop structure. Replace the five-part content per shot; keep each section's job.

Intro (2 shots): establish the world, no vocals

Shot 01 | Extreme wide | city skyline in night rain, neon in puddles | slow lateral track | overcast top light, Kodak 5207, low-sat
Shot 02 | Medium | protagonist #A1 from behind at a rooftop edge | slow dolly-in | rim light, same style anchor

Verse 1 (2 shots): narrative, low and steady

Shot 03 | Close | #A1 profile at a rain-streaked window, gazing out | locked-off with breathing micro-move | window sidelight, same anchor
Shot 04 | Insert extreme close | guitar strings and fingertips, plucks on beat | macro locked | warm lamp, same anchor

Pre-chorus (1 shot): build

Shot 05 | Medium close | #A1 rises and walks to the door, coat catching air | handheld follow | corridor cool-to-warm, same anchor

Chorus 1 (3 shots): vocal peak, A-roll forward

Shot 06 | Medium close | #A1 sings chorus line one, lip-sync per syllable | slow push-in | stage spot snaps on, same anchor
Shot 07 | Close | eyes and micro-expression at the emotional peak | locked with shake accent | top spot, same anchor
Shot 08 | Wide | full stage, all lights up, cut on downbeat | fast pull-out | full wash, same anchor

Verse 2 (2 shots): story advances, B-roll forward

Shot 09 | Medium | #A1 walks a rain alley under alternating streetlights | lateral tracking | point-source streetlights, same anchor
Shot 10 | Insert | a memory object close-up (old photo / ticket) matching the lyric image | slow tilt-down | warm pool of light, same anchor

Chorus 2 (3 shots): same energy, new angles

Shot 11 | Over-the-shoulder | #A1 sings, crowd bokeh in foreground | 30° orbit | spot + crowd rim light, same anchor
Shot 12 | Extreme close | lip-sync close-up on the highest note frame | locked | hard top light, same anchor
Shot 13 | Wide reverse | from behind the stage toward the crowd's light sea | crane up | backlight silhouettes, same anchor

Bridge (1 shot): hold, or turn

Shot 14 | Medium | #A1 sits still, instruments stop, room tone only | very slow push | single window light, same anchor

Final chorus (2 shots): the peak

Shot 15 | Medium close | #A1 final chorus at full tempo with the drums | handheld light shake | full wash + strobe pulse, same anchor
Shot 16 | Extreme wide | return to Shot 01's skyline, sky lightening | very slow pull-out | cold dawn, same anchor

Ending on Shot 01's framing is a low-cost, high-return cinema trick.

Three adjustment rules

  1. Chorus A-roll only grows. Viewers want lip-sync on hooks; more than one B-roll insert in a chorus deflates it.
  2. B-roll imagery must come from the lyrics. Cutaways are the lyric made visual — "sea" gets a sea, "trapped" gets a cage.
  3. Every shot differs from its neighbor in at least one dimension — framing, camera position, subject count, or lighting. Miss this and the film reads as a slideshow.

The negative prompt list (identical on every shot)

plastic skin, glowing eyes, warped hands, flickering face,
changing hairstyle, text artifacts, watermark, oversaturated

This list rides on every generation; it is a floor, not an option. Face drift is cured at the pipeline level — a character asset sheet gives every shot one geometric truth to reference, and prompts are the second line of defense.

Templates are the starting line

Hand-writing sixteen prompts is how you learn shot language. For actual production, ohmymv's AI Director generates the storyboard and keyframes on this structure automatically, and you adjust per-shot at the Review Gate. Eight finished films built this way are in the showcase — study them shot-by-shot against this template. Suno tracks have a dedicated entrance at Suno to Video.

FAQ

Are longer prompts better?

No. Five segments doing their jobs is enough; padded adjectives dilute weight. Test every word: "what does the frame lose if I cut this?" If nothing, cut it.

Won't identical structure produce identical shots?

Shared structure is not shared imagery. Repetition comes from missing difference dimensions (framing/camera/subject/lighting), which the framework pre-seeds per shot — follow it and shots won't collide.

Does this template work with other tools?

The five-part formula transfers to mainstream text-to-video and image-to-video tools. The style-anchor and negative-prompt conventions bite hardest on ohmymv's pipeline, where character anchors let the subject field reference a sheet instead of prose.