Seedance 2.5 Flash text to video

Quick answer: Seedance 2.5 Flash can make a clip from text alone: describe the shot, the action, the lines and the sound, and it returns up to 30 seconds at 480p or 720p. The catch for a story is that text alone does not fix a face, so the same character drifts between clips. SceneMixer keeps the text but adds images: it draws each character and place once and attaches them to every Flash segment.
Still from “The Glass Floor”, an AI short drama generated end-to-end by SceneMixer (5 characters, 1 episode)
SampleThe Glass FloorDigital HD1 episodeProduced end-to-end by SceneMixer from the scriptWatch the episodes

Writing a text-to-video shot

  • Framing and move: “Medium shot, slow push-in”.
  • One action: “she folds the letter and looks up”.
  • The line, if any, with who says it and how.
  • The room’s sound: “Ambient: rain on the window”.

Describe the person precisely if there is no image: age, hair, clothing, one distinguishing detail. Keep lettering out of the frame. Seedance 2.5 Flash prompt guide

Where text alone breaks

Two clips from the same description give two slightly different people: the haircut moves, the coat changes colour. That is fine for a one-off clip and fatal for an episode. References fix it: ByteDance allows up to ten images per Flash request for exactly this reason.

Text first, then images, in SceneMixer

You still start from text — a script or an idea — and SceneMixer turns it into reference sheets in step 2, before any video. Every segment’s shot text then names the characters with @, and the sheets go to Flash with it. The words say what happens; the images say who.

Pure text when you want it

For an establishing shot or an insert with no named people — a skyline, a phone on a table — the shot text carries it on its own. Text to video in SceneMixer

FAQ

Can Seedance 2.5 Flash do text to video?

Yes. Text alone produces a clip of up to 30 seconds with sound.

Why do my characters change between clips?

Text does not fix a face; add reference images, which SceneMixer does for you.

Does text to video cost less than image to video?

No. ByteDance prices both the same per second.

How long should a text prompt be?

One framing, one action, the line and a sound cue per shot.

Can I paste a whole story?

In SceneMixer, yes: it reads the script and writes the shots.

Does it understand dialogue?

Yes. Write the line with who says it; Flash generates the voice and lips.

Turn your text into an episode

Paste a scene; SceneMixer draws the cast and renders on Flash.

Start a project

Updated November 16, 2026