
Turn text into motion
Describe a subject, action and camera direction. Start with a short shot to review pacing.
Create 5, 10 or 15 second videos from prompts, frames and multimodal references at 480p or 720p, with optional generated audio.
上にアイデアを入力して制作を始めましょう。
Plan the subject, motion, camera and sound, then play, download and favorite your video in Results.
Generation engine: Seedance 2 Mini
Read input and output documentation ↗
Describe a subject, action and camera direction. Start with a short shot to review pacing.

Upload the opening and ending images and describe the action or transition.

Use images for appearance, video for motion and audio for sound; describe each reference’s role.
Describe a subject, setting and action, or add frames. Upload multimodal assets in reference mode.
Choose 5, 10 or 15 seconds at 480p or 720p, and review audio settings and estimated credits.
Check status in Results, then play, download, favorite or reuse the prompt.
References, output settings and managing your work.
Generate video from text, first/last frames or visual references. Play and download the finished video from Results.
Choose 5, 10 or 15 seconds at 480p or 720p. Keep reference composition close to the selected landscape, portrait or square output ratio.
A first frame sets the opening and a last frame sets the ending. A last frame requires a first frame. Describe the motion between them; frame mode and multimodal references are separate.
Use up to 9 images, 3 videos and 3 audio files. Total reference video duration and total audio duration must each stay within 15 seconds. Audio needs a visual reference; uploads are checked for format, size and duration.
Images must be under 30 MB. MP4/MOV videos must be under 50 MB and 2–15 seconds each. MP3/WAV audio must be under 15 MB and 2–15 seconds each. Trim longer clips before uploading.
Yes. Enable Generate audio in additional settings and describe dialogue, ambience or sound effects. Disable it for silent output. Review the resulting sound and visuals.
State the subject and setting, then action order, camera motion and sound. For example: a slow push-in as a person looks up and smiles, with gentle rain ambience. Focus a short clip on one action.
Without reference videos, output costs 7.6 credits per second at 480p or 16.4 at 720p; images and frames use the same rate. With reference videos, combined input and output seconds cost 4.8 credits per second at 480p or 10 at 720p. Round each task’s total up to a whole credit. Credits are reserved on submission, charged on success and released for failures.
Filter Results to videos, then open an item to play or download it. Star it to favorite it, or reuse its prompt with the corresponding model. Add references again where needed.
Refresh history before submitting again. Check reference duration, format and dimensions. Retry failed tasks after adjusting inputs; revisit processing tasks later.