Seedance text to video

Generate video from text

How Seedance text to video works

Every active Seedance model on Seedancee (2.5, 2.0, 2.0 Fast, 2.0 Mini and 1.5 Pro) accepts a plain-language prompt as its only required input. The model interprets the subject, action, camera, lighting, style and sound you describe and renders a clip of the length and format you set. Every one of them can generate the audio track in the same pass, so a single prompt can produce a finished shot with ambience and foley.

1. 01

   Write the shot

   One paragraph: what is in frame, what happens, how the camera moves, what the light is like, what it sounds like. Concrete nouns and verbs beat adjectives.

2. 02

   Pick the model

   Seedance 2.5 for up to 30 seconds and later editing; 2.0 for up to 4K in 15 seconds; 2.0 Mini for cheap drafts; 1.5 Pro for a locked camera.

3. 03

   Set format

   Duration, aspect ratio (16:9, 9:16, 1:1, 4:3, 3:4, 21:9 or adaptive) and resolution. The estimate updates live.

4. 04

   Generate and review

   Jobs run asynchronously and usually take a few minutes. The result is saved to your library with the prompt and settings so you can remix it.

Prompt structure that Seedance follows

The most reliable prompts read like a shot description from a script, in this order: subject and setting, action, camera, light and look, sound. Seedance treats each sentence as a constraint, so short declarative sentences work better than one long list of adjectives. Sequential beats ("first... then... finally...") are followed in order, which is how you get a three-beat shot out of a single 2.5 generation.

• Subject and setting: "A grey cat on a windowsill in a sunlit Lisbon apartment."
• Action: "It stretches, then watches a tram pass below."
• Camera: "Slow push-in from across the room, eye level."
• Light and look: "Warm afternoon light, soft shadows, 35mm film grain."
• Sound: "Sound: tram bell in the distance, the cat purring."

Things to avoid: naming real people or copyrighted characters (the content policy prohibits it and the results are unreliable anyway), stacking contradictory camera moves, and asking for on-screen text longer than a word or two. If you need titles, add them in an editor afterwards.

Which model for text to video?

For most text-only shots, Seedance 2.0 Mini for the first pass and Seedance 2.0 or 2.5 for the final is the cost-effective pattern. Pick 2.5 from the start if the shot is longer than 15 seconds or if you expect to extend it. Pick 1.5 Pro when the camera must not move at all.

Example text-to-video prompts

A lone lighthouse keeper climbs a spiral iron staircase at dusk, lantern in hand, the walls streaked with salt. Camera follows from behind and below, then rises past him to reveal the lamp room glowing above. Cold blue exterior light against warm lantern light, 24fps, anamorphic. Sound: boots on iron, wind, a foghorn far away.

Vertical video. A glass bottle of cold-brew coffee stands on a wet black surface; a single drop runs down the label as the camera slowly orbits a quarter turn, hard rim light from behind, dark background. Minimal, premium, no text. Sound: a soft drip and low room tone.

Flat 2D animated style with paper-cut textures: a tiny fox pushes a shopping cart full of pumpkins across a village square at harvest time, leaves blowing, warm oranges and browns. Side-scrolling camera. Sound: cart wheels, rustling leaves, a cheerful accordion.

Locked wide shot of a mountain lake as clouds pour over the ridge and the light shifts from blue hour to full sun, wildflowers in the foreground swaying. No camera movement. Sound: wind only.

Audio in text-to-video

Seedance generates native audio from the sound sentence in your prompt. Describe layers (ambience, one or two foley events, an optional music mood) rather than a specific song. On the Seedance 2.x models audio costs nothing extra; on Seedance 1.5 Pro it doubles the per-second rate. The estimator shows the exact figure before you commit. You can switch audio off for silent drafts.

Cost and credits

Text-to-video is the cheapest workflow because it carries no reference-video surcharge. Cost scales with duration, resolution and audio. A short 480p Mini draft costs the model's minimum charge; a 30-second 1080p Seedance 2.5 shot with audio sits at the top of the range. Failed generations are refunded automatically. Generating needs a subscription: Starter (from $4.99 a week) covers Seedance 2.0 Mini, 2.0 Fast and 1.5 Pro at up to 720p and 10 seconds, and Pro adds Seedance 2.5 and 2.0, higher resolutions and each model's full length. See the pricing page for the estimator and per-model examples.