How to use Seedance 2.5 for long videos (version 2)?
The first note showed the six-part formula for one short clip: a subject doing one thing in one scene. But many videos are a process — a florist packs an order, a cook prepares a dish, a parcel travels from shelf to doorstep. Write all of that as one long paragraph and the model loses track: people change clothes halfway, props jump around, hands hold things that were put down.
Seedance 2.5 has a way to keep a longer video coherent: split the prompt into stages. Each stage says where things stand at the start, the one main thing that happens, and where things stand at the end. The end of one stage becomes the start of the next.
What you need before you start
- An account on Jimeng (free to start).
- The six-part prompt formula from version 1 — the stages build on it.
- One idea for a process with several steps — packing an order, making a dish, setting a table.
The long-video template
This is ByteDance’s official long-video template.
[Generation Goal]
Generate a <video type>. The central subject is <subject>, and the
primary event is <story summary>.
[Stage 1]
Initial state: <initial state of characters, props, and scene>.
Primary event: <one primary action or event>.
End state: <character positions, prop ownership, or visible scene state>.
[Stage 2]
Continue from the previous stage: <state that must remain unchanged>.
Primary event: <one primary action or event>.
End state: <observable state>.
[Stage 3]
Primary event: <closing event>.
End state: <final visible state>.
[Maintain Consistency]
Keep <character identity, number of characters, clothing, prop ownership,
spatial direction, and audio relationships> consistent.
The four building blocks
- Generation Goal — one or two sentences for the whole video: what kind of video it is, who it is about, and what happens overall.
- Stage — one step of the process, with three lines: the initial state (who is where, what is on the table), the primary event (one action only — not two or three), and the end state (what the camera should see when the step is done).
- Continue from the previous stage — from Stage 2 onwards, say what must still be true: same people, same clothes, the bouquet still in the florist’s hand. This is how the stages connect.
- Maintain Consistency — a final checklist of everything that must not change across the video: identity, clothing, where people face, who owns which prop, and how the sounds relate.
The habit to build: end states matter more than actions. If the end of each stage is written clearly (“the scissors are back on the right side of the workbench”), the next stage starts from solid ground.
A worked example
A flower shop packs a delivery order, in three stages. It is the worked example from ByteDance’s own Seedance 2.5 materials.
[Generation Goal]
Generate an instructional video showing a flower shop's order-packing
process. <Florist> and <Store Assistant> arrange, wrap, and hand off
a bouquet together.
[Stage 1]
Initial state: <Florist> stands behind the workbench. Loose flower stems,
scissors, and wrapping paper lie on the tabletop.
Primary event: <Florist> arranges the stems and trims them to length.
End state: <Florist> holds the bouquet in the left hand, and the scissors
are back on the right side of the workbench.
[Stage 2]
Continue from the previous stage: both characters retain the same identities
and clothing, and <Florist> still holds the bouquet.
Primary event: <Store Assistant> unfolds the wrapping paper. <Florist>
places the bouquet inside and ties it with a green ribbon.
End state: the wrapped bouquet lies flat in the center of the workbench, with
the ribbon bow facing the camera.
[Stage 3]
Primary event: <Store Assistant> picks up the bouquet and places it on
the pickup shelf.
End state: the bouquet is centered on the pickup shelf, and both characters
stand behind the workbench inspecting the finished order.
[Maintain Consistency]
Keep <Florist> and <Store Assistant>'s identities, clothing,
workbench orientation, scissors position, and bouquet ownership consistent.
Notice how small the details are — which hand holds the bouquet, which side the scissors rest on, which way the ribbon faces. These are exactly the details an AI gets wrong when you do not say them.
Watch the demo video
Open in Google Drive (no account needed).
The video shows the staged prompt from above turning into one coherent clip: the florist arranges and trims the pink roses at the workbench, the assistant wraps the bouquet and ties it with a green ribbon, and the wrapped bouquet ends up centred with the bow facing the camera while both stand behind the bench — stage by stage, with the same people, clothes and props throughout. The “AI生成” (“AI-generated”) mark in the corner is added by the platform.
Good to know
- Three stages is a good length to start with; add more only when the process genuinely has more steps.
- Naming the people in angle brackets (<Florist>, <Store Assistant>) makes it easy to refer to the same person again and again without confusion.
- To connect videos you already have instead of writing one long prompt, see How to use Seedance 2.5 for video-to-video (version 3)?
Further reading
- Seedance 2.5 prompt guide — ByteDance’s official prompting guide.
- Seedance 2.5 prompt examples — more example prompts from ByteDance.
- Jimeng — the website where you use Seedance.