Skip to content

Seedance 2.5 prompt generator

Seedance 2.5 can hold one take for up to 30 seconds, or cut between linked shots inside a single clip, and it makes the soundtrack with the picture. This guide shows how to prompt it, and its generator writes for both: one continuous take, or two or three numbered shots with the seconds each one runs, the same subject carried across the cuts.

  • 11 credits a prompt (about $0.10 on Starter), with any DreamBench plan.
  • No photo needed, and no model is called.
  • A preset of the AI video prompt generator.
Go to the generator

What is a Seedance 2.5 prompt generator?

It writes a prompt for ByteDance's Seedance 2.5 from your choices: one take of up to 30 seconds, or two or three numbered, timed shots in one clip, with light, camera and sound, asking only for what Seedance can make. It is part of every DreamBench plan, from $9.99 a month.

How to write a Seedance 2.5 prompt

  1. Describe the moment

    Who or what is in frame, what they do and where. A Seedance 2.5 clip has room for a moment to open, develop and land, so describe the whole beat, not a single pose.

  2. Choose a take or shots

    One continuous take, or two or three linked shots from a wide opening to a close-up. Then the length, from 5 to 30 seconds, the frame, the light, the lens, the move and the sound.

  3. Copy or run it

    Linked shots come out numbered, with the seconds each one runs. The first time, create a free account and choose a plan, then copy the prompt into any tool that runs Seedance 2.5, or run it on DreamBench.

Start from an example

Fill in the subject and what they do, and the prompt appears here.

A prompt uses 11 credits from your plan, and the same prompt again uses none. Running it on DreamBench renders a real clip, priced on its own.

How a Seedance 2.5 prompt is built

Tap a part to see where it lands in a real prompt, or tap a sentence of the prompt to see which part wrote it.

  1. The frame and how the clip runs

    Widescreen, vertical, square, 4:3 or cinematic 21:9, then one continuous take with no cuts, or the number of linked shots, said first.

  2. The whole moment

    Who is in it, what they do and where. A clip can run 30 seconds, so describe the beat from start to finish: what happens first, what turns, how it lands.

  3. The light

    The source and its side, and what falls to shadow, the same in every shot of the clip.

  4. The shots, numbered and timed

    Linked shots come as Shot 1, Shot 2 and Shot 3, each with its seconds in whole numbers and its size, from a wide opening to a close-up as the moment lands.

  5. The same subject in every shot

    A line that keeps the subject, the place and the light across the cuts. ByteDance reports smoother transitions in 2.5, with the main subject keeping its look.

  6. The sound last

    Ambience, a spoken line, a music bed, or silence: Seedance 2.5 makes sound together with the picture, or leaves it out.

A Seedance 2.5 prompt

Widescreen 16:9, one continuous 30-second take with no cuts. Two friends in raincoats share one umbrella at a night bus stop, laughing as the bus finally pulls in, on a rainy city street, neon reflected in the puddles. Light: night, lit only by the practical lights in the scene from behind the subject, the camera shooting into the light from the shadow side, a rim of light on the edges; the exposure holds the subject; no fill from the camera side. Medium shot on a 35mm lens at a few metres: the subject and the room both readable, little distortion. The camera is locked off; nothing moves but the scene. Photoreal, shot on film: natural colour, gentle grain, true skin. Sound: the scene's own ambience, no music. No text, no captions, no watermark.

Example prompts

Each one is written by the generator on this page, from the choices its note describes.

Thirty seconds in one take

Seedance 2.5

One continuous take of 30 seconds, long enough for the moment to open, build and land, with the ambience under it.

Prompt

Widescreen 16:9, one continuous 30-second take with no cuts. Two friends in raincoats share one umbrella at a night bus stop, laughing as the bus finally pulls in, on a rainy city street, neon reflected in the puddles. Light: night, lit only by the practical lights in the scene from behind the subject, the camera shooting into the light from the shadow side, a rim of light on the edges; the exposure holds the subject; no fill from the camera side. Medium shot on a 35mm lens at a few metres: the subject and the room both readable, little distortion. The camera is locked off; nothing moves but the scene. Photoreal, shot on film: natural colour, gentle grain, true skin. Sound: the scene's own ambience, no music. No text, no captions, no watermark.

Three linked shots

Seedance 2.5

Fifteen seconds in three shots: a wide opening, a medium shot with the move, and a close-up as the moment lands, each timed in whole seconds.

Prompt

Vertical 9:16, one 15-second clip in three linked shots. A baker in a flour-dusted apron pulls a tray of loaves from the oven, breaks one open and smiles at the steam, in a small bakery before dawn. Light: night, lit only by the practical lights in the scene from camera-left, so the subject's camera-right side falls to shadow; the exposure holds the subject; no fill from the camera side. Shot 1 (0–5 s): wide shot, the place and the subject in it. Shot 2 (5–10 s): medium shot on a 50mm lens at a couple of metres: natural proportions, the background softening behind the subject. One move only: a slow push-in that ends close on the subject. Shot 3 (10–15 s): close-up on the subject as the moment lands. The same subject, place and light in every shot. Photoreal, shot on film: natural colour, gentle grain, true skin. Sound: the scene's own ambience, no music. No text, no captions, no watermark.

A line in Spanish

Seedance 2.5

A 10-second cinematic take with a line typed in the language it should be spoken in, here Spanish.

Prompt

Cinematic 21:9, one continuous 10-second take with no cuts. An old fisherman in a yellow oilskin mends a net on the harbour wall and calls to his grandson, in a small fishing port at dawn. Light: low golden-hour sun from behind the subject, the camera shooting into the light from the shadow side, a rim of light on the edges; the exposure holds the subject; no fill from the camera side. Close-up on an 85mm lens from six metres: the background compressed and soft, only the subject sharp. One move only: a slow push-in that ends close on the subject. Photoreal, shot on film: natural colour, gentle grain, true skin. Sound: ambience and one spoken line, "La marea está cambiando, ven." No text, no captions, no watermark.

Common mistakes

A pose instead of a moment

Thirty seconds of one held action look frozen. Give the clip a beginning, a turn and an end.

Shots too short to read

Three shots in five seconds leave each under two. The generator offers linked shots from 10 seconds up, so no shot runs much under three seconds.

A crowd that interacts

ByteDance says scenes where several subjects interact are not yet fully stable. Keep one clear subject and one main action.

Complex physical motion

ByteDance says complex motion can look physically off: acrobatics and juggling are where it shows. Slower, simpler action holds together.

A new subject in the last shot

A character or a place that appears only in Shot 3 breaks the link between the shots. Keep the subject, the place and the light, and let the shot size change.

A real face as a reference image

Seedance won't accept a reference image that shows a real person's face. Describe the person in words, or choose another model for that video.

Works with

What it takes
Your own words: who is in the shot, what they do and where. No photo is needed.
What you get
A prompt as plain text.
DreamBench's composer
Paste it into Video on Seedance 2.5, which sends your words to the model as written, at 480p, 720p or 1080p.
Dreamina
ByteDance's own creative app runs Seedance from the same plain-text prompt.
The BytePlus API
Send the prompt as text; its numbered shots are plain text too.

Limits

Which model it writes for, and its limits

Writes for Seedance 2.5, within what it can make on DreamBench:

Lengths
5, 10, 15, 20 or 30 seconds
Frames
16:9, 9:16, 1:1, 4:3 or 21:9
Sound
On or off: ambience, a spoken line or a music bed
Shots
One take, or two or three linked shots from 10 seconds
Dialogue
One line, said in the scene
Prompt format
Plain text

Up to 30 seconds in one clip

The length runs from 5 to 30 seconds, long enough for a full exchange or a scene that builds to its moment, and the prompt names it, so the action is paced to the clip.

Linked shots, numbered and timed

Pick two or three shots and the prompt lays them out as Shot 1, Shot 2 and Shot 3, each with its seconds and its size, from a wide opening to a close-up as the moment lands.

The same subject across the cuts

ByteDance reports smoother transitions between shots in 2.5, with the main subject keeping its look. The prompt asks for the same subject, place and light in every shot.

Sound that follows the picture

Seedance 2.5 makes dialogue, ambience and effects together with the picture. Choose ambience, a spoken line or a music bed, or silent if you only want the picture.

Where it fails

  • ByteDance says complex motion can look physically off.
  • Scenes where several subjects interact are not yet fully stable, ByteDance says.
  • Seedance won't accept a reference image that shows a real person's face.

Seedance 2.5 prompt generator FAQ

How do I write a prompt for Seedance 2.5?

To write a prompt for Seedance 2.5, open with the frame and how the clip runs, one continuous take or two or three numbered shots, then who is in it, what they do and where, the light, the camera and the sound. A clip can run 30 seconds, so describe the whole moment: what happens first, what turns, how it lands. The generator writes it in that order.

Can Seedance 2.5 put several shots in one clip?

Yes. One clip can cut between linked shots, and ByteDance reports smoother transitions with the main subject keeping its look across the cuts. Choose two or three shots and the prompt numbers them, gives each its seconds and moves from a wide opening to a close-up, the same subject and light in each.

How long can a Seedance 2.5 clip be?

Up to 30 seconds in one generation, ByteDance says. The generator offers 5, 10, 15, 20 or 30, and linked shots from 10 seconds up, so no shot runs much under three seconds. Of the three models here, Seedance is the one that runs 30 seconds and cuts between shots in one clip; a DreamBench video can be longer still, because the composer joins clips.

How do I write a multi-shot prompt for Seedance?

A Seedance multi-shot prompt numbers the shots and gives each its span in seconds: Shot 1 (0–5 s) wide, Shot 2 (5–10 s) medium with the one camera move, Shot 3 (10–15 s) close-up as the moment lands. Keep one subject, one place and one light across the cuts. ByteDance's own examples time their segments as 0–5s and 6–10s; this generator writes whole seconds so no shot runs under three.

How do I make a character speak in Seedance?

Choose a spoken line and type it: the generator puts it in quotes after the person who says it, so the voice belongs to someone on screen, and Seedance 2.5 makes the voice with the picture. Keep it short enough to say in the clip, and type it in the language it should be spoken in.

Can I use reference images with a Seedance prompt?

This generator writes text only. Seedance 2.5 takes reference images in ByteDance's own apps and in the DreamBench composer, where a scene's references, a product, a place, a character sheet, go to the model with the prompt; it refuses any image that shows a real person's face, so describe a person in words instead. The prompt then says only what moves.

Does Seedance take a negative prompt?

This generator writes none: it closes with one line, no text, no captions, no watermark, and puts everything else as a positive direction, which video models follow more reliably than a list of exclusions. If the tool you paste into offers a separate negative field, keep it to a few short nouns and say the rest positively.

How long can a Seedance prompt be?

60 to 150 words for one take covers the frame, the moment, the light, the camera and the sound without repeating itself; a three-shot prompt runs longer, because each shot carries its seconds and its size. That is what this generator writes, and it pastes as plain text into Dreamina, the BytePlus API and the DreamBench composer.

How do I write a Seedance image to video prompt?

Describe only what moves and what the camera does; the picture already sets the subject, the light and the look, so a sentence about the motion and one about the sound is enough. This generator writes text-to-video prompts; on DreamBench the product video maker runs Seedance 2.0 from one product photo, with the labels asked to stay as photographed.

Does a Seedance 2.0 prompt work on Seedance 2.5?

Yes: the structure carries over. Seedance 2.5 runs longer, up to 30 seconds against 15 on 2.0, so a 2.0 prompt often describes less than a 2.5 clip has room for. Add the turn and the landing, or split it into linked shots. DreamBench runs both models; this generator writes for 2.5.

Can the dialogue be in another language?

Type the line in the language it should be spoken in; the generator writes its own sentences in English and passes your line on as typed. ByteDance's Seedance 2.5 pages don't list the spoken languages, so try a short line first and listen before you build a longer clip on it.

What does Seedance 2.5 find hard, and does this generator need a plan?

ByteDance is candid about it: complex motion can look physically off, and scenes where several subjects interact are not yet fully stable, so the generator keeps to one clear subject and one main action. The generator itself needs a DreamBench plan: you create an account on the way, and the button shows the price in credits before you press it.

How much does the Seedance 2.5 prompt generator cost?

Each prompt uses 11 credits from your DreamBench plan, and the same prompt again uses none. Starter, $9.99 a month, includes 1,100 credits a month, about 100 prompts, plus AI video and the image tools.

Do I need an account to use the Seedance 2.5 prompt generator?

Yes: a free account and a DreamBench plan. The tools are part of every plan, from Starter at $9.99 a month.

Copy checked against the engine on .

More tools

Every tool

Learn

Guides, comparisons and terms

Then make it move

Every result can go into a video.

Run your prompt in the composer, and DreamBench writes the scenes and renders them on the model you pick, from Seedance 2.5 to Veo 3.1 and Kling 3.0. Rendering uses credits, which come with a plan.

See the plans