A scene with time to develop
Use up to 30 seconds for an entrance, a product reveal, or a short conversation. Give each action a clear place in the sequence instead of asking for every movement at once.
Turn a photo or a written scene into a video. Choose 4–30 seconds, up to 4K, and sound.
Keep the perfume bottle still and slowly move the camera closer. Preserve the product shape as studio light moves across the glass.
Build a longer shot, animate a starting image, or direct a scene entirely with words.
Use up to 30 seconds for an entrance, a product reveal, or a short conversation. Give each action a clear place in the sequence instead of asking for every movement at once.
Start from a product photo, portrait, or storyboard frame. Describe what moves and what should stay steady. Seedance 2.5 image-to-video follows the proportions of your uploaded image.
Specify dialogue, ambient sound, or music in your prompt and enable audio. Turn audio off when you plan to add your own soundtrack. Review speech and timing before publishing.
Watch actual model outputs. Each clip identifies its model and source; results vary with the input.
5 seconds · 480p · Audio on · 16:9
A bicycle courier rides along a coastal road at sunrise. Track the rider from the side, with moving waves and grass in the background.
View source and full prompt · Seedancer2 ↗5 seconds · 480p · Audio on · Original image ratio
Keep the perfume bottle still and slowly move the camera closer. Preserve the product shape as studio light moves across the glass.
View the original product image ↗
View source and full prompt · Seedancer2 ↗Use image-to-video when the subject already exists in a photo. Choose text-to-video when you want the model to build the whole scene.
Name the subject, then its action, camera movement, and sound. For a longer video, divide the description into timed beats.
Start with 5 seconds at 720p to check the idea. Increase duration or resolution after the composition and movement work for your project.
0–3s: A ceramic teapot rests beside a window; steam rises from the spout. 3–7s: The camera moves closer to the glaze while morning light crosses the table. 7–10s: Hold on the teapot. Quiet room tone, no music, no text.
Back to the generator ↑Seedance 2.5 is ByteDance’s audio-video generation model. Use a written scene or a starting image to create a video with camera movement, subject action, and optional sound.
Yes. Select a whole-number duration from 4 to 30 seconds. For a 30-second scene, describe the order of events and avoid repeating the same camera instruction in every sentence.
Choose 480p, 720p, 1080p, or 4K on Seedancer2. Higher resolution costs more credits. Use the resolution selector above to see the exact cost before submitting.
A 5-second video costs 21 credits at 480p, 41 at 720p, 102 at 1080p, or 203 at 4K. The same rates apply to text and single-image input. Disabling audio does not reduce the 2.5 price.
Seedance 2.5 image-to-video follows the input image’s proportions. Crop your image to 9:16 before uploading for a vertical result. In text-to-video, choose 9:16 directly.
ByteDance describes editing and extension capabilities in the model family. This tool currently exposes text and single-image generation; video upload, editing, and extension are not available in this form.