Finish the audio first, then build visuals around a section. Decide whether viewers need to see exact words being performed: rhythmic B-roll and lyric-accurate close-ups require different production paths.
AI Genjutsu makes silent Seedance 1.5 Pro clips from text or one optional starting image. It does not generate music, accept a soundtrack, or align mouths to lyrics. Check current studio availability before rendering. The storyboard below is a production plan, not a demonstrated result.
Lock the track before choosing the shots
Start with an original vocal recording, lyrics, and a beat you have permission to use. Keep one final audio file. Mark phrase starts, strong accents, and a clean ending in your timeline before cutting visuals.
For a 4/4 beat, one beat lasts 60 / BPM seconds and one four-beat bar lasts 240 / BPM seconds. At 100 BPM, a beat is 0.6 seconds and a bar is 2.4 seconds. Listen beyond this grid: the vocal may enter early or stretch across a beat.
If you need an external songwriting step, Suno's own-lyrics help explains entering lyrics in Custom mode. Its Add Vocals guide describes lyrics-based vocals over instrumentals; this feature requires an eligible paid plan. These are separate services, not AI Genjutsu features; check their current access and usage terms for your project.
Choose mouth performance or rhythmic B-roll
Use a close-up when seeing a specific lyric delivered is central to the idea. Record that delivery against the locked track, or evaluate an external audio-driven lip-sync workflow before committing your storyboard. Silent mouth movement may not match the words.
Use B-roll when the beat, setting, and mood carry the hook. A hand on a rail, wind moving a jacket, or an empty stage can support the track without suggesting synchronized delivery. Do not put “sing these lyrics” into this site's prompt expecting an audio result.
An original fourteen-second storyboard
Plan separate clips within the site's 4–12-second range, then trim them in an external editor. A fourteen-second final edit is not one supported generation request.
| Timeline | Shot brief | Soundtrack decision |
|---|---|---|
| 0–3 seconds | Wide shot of one original performer on a rooftop, camera slowly approaching. | Establish the location over the intro pickup. |
| 3–7 seconds | Medium shot, one shoulder turn toward the skyline; no lyric close-up. | Cut at a strong accent, leaving the vocal uninterrupted. |
| 7–10 seconds | Insert of a hand resting on a weathered rail, sleeve visible. | Use the insert over the densest phrase. |
| 10–14 seconds | Return to the wide framing; performer holds position as the camera settles. | Let the final word finish before the end cut. |
Keep one continuity card: navy work jacket, rooftop rail, pale morning light from frame left, muted color palette. Reuse those details across briefs and select starting images that fit the same scene.
A visual prompt for the opening
Six-second cinematic wide shot of one original performer in a navy work jacket on a quiet rooftop beside a weathered metal rail. Pale morning light enters from frame left. The performer holds a relaxed stance and looks toward the skyline. Camera makes one steady push-in. Muted colors, natural fabric movement, no crowd, no readable text. Visual atmosphere for an original rap edit.
This prompt creates a visual brief, not a beat-synchronized instruction. ByteDance's Seedance 1.5 Pro announcement describes the underlying model; this site's silent configuration remains the relevant limit.
Fix the edit before exporting
| Problem | Practical correction |
|---|---|
| Cuts feel late | Move the cut near the accent by a few frames and compare while listening. |
| Outfit or light changes | Replace the inconsistent shot or rebuild its brief from the continuity card. |
| Mouth movement clashes with lyrics | Use a non-mouth insert or an independently synchronized performance shot. |
| The ending chops a word | Extend the audio tail and hold the last usable frame or shot longer. |
Can I upload my song here?
No. Assemble the silent clips and your permitted audio in an external editor. The current output also includes a watermark.
Can I use my own voice in another music service?
Suno's Voices guide describes an own-voice verification workflow. Use your own voice, check eligibility, and do not treat it as permission to clone an artist.
What should I check in the final file?
Play the complete export. Check lyric boundaries, black frames, facial changes, audio clipping, crop safety, and watermark placement on the target screen.
For related shot planning, see a two-person rap scene and a simple orange-stage setting.