Night train crowd presses two strangers together
- Model
- Seedance 2.5
- Task
- Text to video
- Resolution
- 720p
- Duration
- 4 s
- Aspect ratio
- 16:9
Output
This preview is blurred
This preview is marked sensitive. Confirm your age to see it.
PromptEN476 characters
A woman alone in a night train carriage, coat falling open over bare legs, stands as the train lurches and catches the overhead rail; a man steps in behind her at the stop and the crowd presses them chest to back against the door. Her head tips onto his shoulder, his hand finds her hip, neither looks at the other. Tunnel lights strobe across their faces. Slow push in through the carriage, warm light against black glass. Cinematic, shallow depth of field, heavy film grain.Written in English, exactly as we sent it.
Run this example
Checking the price…
Opens the playground with every setting filled in and shows the exact price before you run.
Same model, starting from this prompt. Other settings use their defaults.
Why this prompt works
No write-up for this one yet. The output and the settings tell the story.
Settings used
| Setting | Value | What it does |
|---|---|---|
| resolution | 720p | 480p, 720p or 1080p, default 720p, and it feeds directly into the billed quantity. Iterate on prompts at 480p or 720p and raise the tier only for the final take. The tier you pick is the tier the model renders at — there is no upscaling or enhancement step layered on top. |
| aspect_ratio | 16:9 | Defaults to `adaptive`, which lets the model choose the framing from the prompt — say "vertical short-form" and you will usually get a vertical composition. Lock a shape explicitly only when the delivery slot demands it (9:16 for feeds, 16:9 for landscape, 21:9 for widescreen); otherwise adaptive tends to compose more naturally. No effect on billing. |
| generate_audio | true | On by default. Produces a native soundtrack synchronized to the picture — voice, room tone, movement effects. Switch it off whenever the edit will carry its own music or narration: it stops the generated track from clashing with the one you add later. |
| duration_seconds | 4 | 4 to 30 seconds, default 5. The 30-second ceiling is the substantive change in this generation: a complete shot fits in one request instead of being chained together from closing frames. Length scales the billed quantity linearly, so confirm the motion and pacing at 5 seconds before going long. |
More Seedance 2.5 prompts
All Seedance 2.5 prompts
Romance on other models
Romance
Preview blurred