How to write prompts for Seedance 2.5
Seedance 2.5 turns a sentence or a start picture into a short clip with sound. The first two sections put the provider’s own prompt guide in our words, and we did not test them. The last rests on seven sample clips made with this model that we watched ourselves.
Written from the provider’s documentation and from results we checked ourselves. Last checked November 8, 2026.
Start with one sentence, then the shots
From the provider’s documentation
- The guide asks for a one-sentence summary first: who or what, where, what happens, the style and the camera. Then the detailed plot, written shot by shot with what is seen, how the camera moves, the action, any dialogue and the sounds.
- Shots can be numbered (Shot 1, Shot 2) or given whole-second timestamps. The guide warns that too little in a time range lets the model improvise, too much brings extra cuts or lost beats, and timestamps should not control fast repeated actions.
- Describe what you want. Negatives are supported for subtitles and for sound, such as no subtitles or no background music, and the guide says subtitles can still appear now and then.
Camera, action and sound
From the provider’s documentation
- Camera moves and shot sizes can be written as plain terms such as push in, pull out, pan, track, orbit, low angle or first-person view. A rare term should be written as the term plus a plain explanation. For a transition, say when it happens and how.
- Describe actions in general, readable terms and write expressions as sentences rather than idioms. The guide also says intense emotion words can make the eyes glow, so describe the feeling through the face and body instead.
- Text in the picture can be misspelled. The guide’s workaround is to write a short word letter by letter so that the letters appear one after another.
- The provider’s guide also covers reference clips, editing and extension. Here this model takes text, a start picture or reference pictures, without reference clips or sounds, so we leave those out.
What our sample clips showed
From results we checked ourselves
- A style named at the start of the sentence held through the clip: the Tokyo wave stayed in flat woodblock colours from its first frame to its last, and the girl in the sunflower field stayed in an anime look.
- The prompts for the sunflower clip and the Chang’an clip named no camera move, yet the camera moved on its own. The sunflower clip opens close on a flower and pulls back to the girl. The Chang’an clip pulls back and up from street level until the whole avenue is in view. Write “static camera” if you want it still.
- The hanbok prompt put its actions in order, and the clip shows them in that order: she walks away, turns and looks back over her shoulder, with lanterns and falling snow, ending close on her face.
- The Mount Fuji clip with drifting maple leaves, the red panda eating bamboo and the pancakes with syrup each matched a one-sentence prompt. We saw nothing odd in them, but we watched five-second clips only and did not judge the sound. None of these prompts used shot numbers or timestamps, so we cannot say how well those work here.
Real results and the prompts behind them
Results made with this model. The prompts are shown exactly as written, in English.
Settings of this model
- Kuanzia
- ByteDance
- Aina
- AI video model
- Jinsi ya kuunda
- Maandishi kuwa video · Picha kuwa video · Picha ya mwongozo kuwa video
- Urefu
- 4–30s
- Ubora
- 480p / 720p / 1080p
- Aspect ratios
- 16:9 · 9:16 · 1:1 · 4:3 · 3:4
- Sauti
- Pamoja na sauti
- Bei
- from 174 / sec
Maswali
Does Seedance 2.5 follow timestamps?
The provider’s guide says Seedance 2.5 understands whole-second timestamps as well as shot numbers, but they should not control fast repeated actions. We have not tried timestamps in our own samples.
Why did the camera move when I did not ask?
In our samples two clips moved the camera although the prompt named no move. Name the move you want, or write that the camera stays still, as the provider’s own still-shot example does.
Can I stop subtitles and background music?
The guide says to write no subtitles or no background music, and also says they can still appear sometimes. We have not tested this ourselves, so look at and listen to the whole clip.
Provider documentation we read
We read these pages and wrote the points that matter in our own words. The provider may change them.
- Dreamina Seedance 2.5 prompt guide (BytePlus ModelArk) (docs.byteplus.com)
More prompting guides
- How to write prompts for GPT Image 2.6 FlareAI image model
- How to write prompts for GPT Image 2.6 SunburstAI image model
- How to write prompts for Nano Banana 2.1AI image model
- How to write prompts for MOSS ImageAI image model
- How to write prompts for Grok Imagine ProAI image model
- How to write prompts for Nano Banana ProAI image model
- How to write prompts for Nano Banana 2AI image model
- How to write prompts for GPT Image 2.5 FlareAI image model
- How to write prompts for IndigoAI image model
- How to write prompts for SaffronAI image model
- How to write prompts for CeladonAI image model
- How to write prompts for LapisAI image model
- How to write edit requests for UmberKihariri cha picha cha AI
- How to write prompts for Kling Video 4.0AI video model
- How to write prompts for Seedance 2.0 MiniAI video model
- How to write prompts for Kling V3AI video model
- How to write prompts for Kling O3 OmniAI video model
- How to write prompts for Veo 3.1AI video model
- How to write prompts for Veo 3.1 FastAI video model
- How to write prompts for SilkAI video model
- How to write prompts for LacquerAI video model
- How to write prompts for SatinAI video model
Majina ya modeli, kampuni na bidhaa kwenye tovuti hii ni alama za biashara za wamiliki wao; Pickshot haihusiani nao, haifadhiliwi nao wala haiidhinishwi nao.