The short answer
A 58th Floor Seedance prompt maps two uploaded images as characters, specifies which slot leads the hook and which takes the verse, describes the teal studio setup with a hanging mic, and timestamps the four shots: two-shot hook, close-up verse, wide body shot, and face-off final. The entire prompt runs about 150 words and holds the four-camera edit together without drifting into generic.
Copy the template below, replace the bracketed sections with your details, and paste it into Seedance 2.5's custom mode. Render time: 5 to 8 minutes. Output: 12-second vertical video, 9:16, 720p or higher.
Asset mapping: who is who
@Image 1: Character A—lead rapper, hook and first verse delivery
@Image 2: Character B—response rapper, verse answer and ad-lib backing
@Audio: Optional—your own 2–15 second track (must be audio you own the rights to)If you use the AI rap option instead of a custom audio, Seedance will generate the beat and bars, and you can skip the Audio line.
Keep this order: Image 1 on the left, Image 2 on the right. The left character always leads. If you want them swapped, swap the upload order in Seedance, not the prompt text.
One-sentence brief
Use this format: [Character A name] and [Character B name] trade bars at a vintage microphone in a seamless teal studio, Anderson .Paak and Cordae style, 12-second vertical duet.
Example: James and Sarah trade bars at a vintage microphone in a seamless teal studio, Anderson .Paak and Cordae style, 12-second vertical duet.
The full prompt (copy-paste ready)
Asset mapping:
@Image 1: Character A (lead rapper)
@Image 2: Character B (response rapper)
[Optional: @Audio 1: custom track uploaded by user]
One-sentence brief:
[Character A] and [Character B] trade bars at a vintage microphone in a seamless teal studio, Anderson .Paak and Cordae style, 12-second vertical duet.
Timeline and shot order:
0–2 seconds (Hook): Waist-up two-shot of both characters facing the hanging mic. Character A delivers the opening bars while Character B nods and grins. Camera is locked medium distance, even lighting on both faces.
2–5 seconds (Verse): Cut to tight close-up of Character B's face and chest. Character B raps straight at the camera, mouth synced to every syllable. Hanging mic is visible at the edge of the left side of frame. Character A is out of frame but audible in the backing.
5–8 seconds (Wide): Cut to full-body shot of both characters from waist to head. Character A is on the left, Character B on the right. Both are bouncing slightly, shoulders moving on the beat. The teal studio wall fills the background, no other objects.
8–12 seconds (Face-off): Cut to a low two-shot where both faces are visible, close together, rapping at each other. Mouth sync on every word. Character A's final phrase ends the clip. Freeze slight smile.
Global constraints:
- Studio: seamless teal background, no shadows, no props except the hanging microphone
- Microphone: always visible in the frame, hanging from above, vintage aesthetic, centered in compositions
- Lighting: even, shadowless, warm but neutral—no heavy color cast
- Aspect ratio: 9:16 vertical, 12 seconds total duration
- Audio: Lip sync to every syllable; no subtitles
- Color: Cool teal walls, warm skin tones, no filter or heavy grading
- Continuity: Both characters keep the same outfit, hair, and expression style across all four shotsReady to use: copy everything between the triple backticks above
Paste the entire block into Seedance's custom prompt field. Replace [Character A] with the first person's name and [Character B] with the second. If you have custom audio, add the @Audio line with your filename. Leave it out if you want AI-generated bars.
Hit Create and wait 5 to 8 minutes.
Four-shot breakdown
Understanding what each shot does helps you predict what the output will look like:
Hook (0–2 seconds)
Both faces at the mic, building rapport. This shot introduces both characters and sets the studio mood. Camera should be medium distance and still—not moving, not close. Character A is leading visually as well as musically.
If Character A seems too small or Character B seems to dominate: adjust the "even lighting on both faces" to specify "Character A slightly forward, Character B slightly back" in the next render.
Verse (2–5 seconds)
Single close-up of Character B. This is where lip-sync quality matters most—every syllable has to match. The mic should be cropped into the left edge, not centered. If the mic vanishes or Character B's mouth is too wide: re-render with "tight close-up, mouth unobstructed, microphone visible at left edge" in the next revision.
Wide (5–8 seconds)
Both characters full-body, bouncing. This is your second chance to show the outfit and the studio depth. The background should be pure, shadowless teal—not a texture, not a gradient. If you see shadows or a pattern: add to global constraints "flat, no texture, no shadows" and re-render.
Face-off (8–12 seconds)
Low two-shot, faces close. The tension lives here. Both should be visible and both should be rapping. If one face is cropped or mouth movement stops: the issue is usually that one character's photo was slightly off-frame or had poor mouth definition. Re-render with a tighter headshot photo.
What to change if the output drifts
The most common failures and quick fixes:
One face looks small or far away In the one-sentence brief, change "medium distance" to "closer framing" and specify which shot (hook or wide) needs adjustment.
The teal studio looks textured or shadowed Global constraints need "flat teal background, zero shadows, zero texture, shadowless fill light" instead of just "teal."
Mouth does not sync to audio Either the photo quality was poor (lips not clearly visible) or the audio is heavily processed. Try re-rendering first. If it persists: in the timeline, specify "every syllable lip-synced" and add "no autotune or heavy processing on reference audio."
One shot is blurry or low quality Seedance sometimes renders one shot lower quality to save compute. Specify "all four shots, 720p minimum" in global constraints and re-render.
Background changes between shots Teal studio should stay identical. Add to global constraints: "Teal studio background is identical and unchanged in all four shots."
Changing the style: swap the studio
The prompt above locks you into the teal 58th Floor studio. To use the same four-shot structure in a different room, swap one line:
For Hotel Lobby (orange booth):
Replace the studio line in global constraints with:
- Studio: seamless orange background, flat walls, no shadows, vintage vibes—Hotel Lobby COLORS setThen replace the one-sentence brief with:
[Character A] and [Character B] trade bars at a vintage microphone in a flat orange booth, call-and-response style, 12-second vertical duet.Read the full Hotel Lobby guide for more details on this set.
For Green Booth (Luxury Lobby):
Replace the studio line with:
- Studio: seamless sage green background, modern minimalist aesthetic, ambient lighting—Luxury Lobby styleAnd the brief:
[Character A] and [Character B] trade bars at a vintage microphone in a modern sage green studio, contemporary call-and-response style, 12-second vertical duet.Read about Luxury Lobby and other variations in the main guide.
For Street Cypher (outdoor):
Replace the studio line with:
- Studio: outdoor street corner, weathered brick wall, daylight, street-jam aesthetic, zero studio lightsAnd the brief:
[Character A] and [Character B] freestyle at a street cypher in daylight, back-and-forth freestyle style, 12-second vertical duet.Each set changes the vibe but keeps the four-shot structure: hook two-shot, verse close-up, wide body shot, face-off final.
Why prompt structure matters
Seedance 2.5 treats your prompt as a visual script. If you skip asset mapping, the model has to guess which photo is which character and might assign them backwards. If you skip timestamps, all four shots render at the same length instead of letting the hook breathe and pacing the verse close-up correctly. If you skip global constraints, the teal color drifts, shadows creep in, and continuity breaks between shots.
The prompt above is structured to survive variations—change the character names, swap the audio, flip between studios—without losing the four-shot identity.
Before you paste into Seedance
- Upload your two images in order. First image (left character) uploads first. Second image (right character) uploads second. Order matters.
- Optional: upload your audio. If you're using custom audio, make sure it's MP3 or WAV, 2–15 seconds, and audio you own the rights to.
- Paste the entire prompt block. Do not edit the timestamps or the shot descriptions—those are what tell Seedance when and how to cut between shots.
- Fill in the names. Replace all
[Character A]with the real name and[Character B]with the real name. - Hit Create. Seedance will take 5 to 8 minutes. You can close the tab and come back.
- Download and post. The output is an MP4, no watermark, ready for TikTok or Reels.
Frequently asked questions
What if I want to change the names but keep the same prompt?
Replace [Character A] and [Character B] with any names you want. The prompt structure stays the same.
Can I add a topic for the AI bars?
The prompt above doesn't include AI bar generation—it uses either custom audio or a pre-written beat. To add a topic prompt, you would need to generate the bars separately and then upload the result as custom audio. Seedance's custom prompt mode does not natively support topic-driven bar generation; that's a feature of the template interface instead.
What if one character's mouth is not syncing?
The most common cause is that one photo had the mouth slightly blurred or partially open. Try with a different photo. If the photo is sharp, re-render and specify "extreme mouth definition on both characters, every syllable crystal clear" in the timeline.
How do I know if my prompt is working?
Render a test version. If all four shots appear in the right order and the studio background is consistent, the prompt is working. If you see black frames, missing shots, or the faces are swapped, adjust the asset mapping and timeline order and try again.
Can I use this prompt outside Seedance?
Seedance 2.5 is structured for this prompt style. Other models (Runway, Kling, Pika) use different prompt syntax. This prompt is built for Seedance and won't work as-is elsewhere without translation.
Start rendering
Paste the prompt above into Seedance 2.5's custom generation mode, fill in your two photos and optionally your audio, and hit Create. The four-shot 58th Floor duet will render in about 5 to 8 minutes, ready to post straight to TikTok, Reels, or Shorts with no editing needed.
Want step-by-step guidance without prompts? Read the full 58th Floor how-to instead.