A blockout (previs) is a rough 3D clip of grey figures that fixes the essentials: where people stand, where they go and how the camera flies. From text alone a video AI moves the camera however it likes; with a blockout it repeats your camera move frame by frame.
The blockout itself sets the camera: you describe the move in words or edit a version, and the AI gets a ready camera path to shoot over.
This is a real scene AI built from one sentence about a waltz. Press “Control camera”: fly around the hall, drop points, and the camera flies through them, just like in your own blockout.
Any shot where how the camera moves matters as much as what's in frame.
Upload the blockout clip to Seedance 2.5 as a video reference and describe who is in the shot and where it happens. The model takes the camera move and blocking from the blockout, and the look from your text and photos.
From text alone the AI decides how to move the camera, and does it differently every time. A blockout fixes the move in advance, so you get the right motion in one or two tries.
No. AI builds the scene, people and camera from your description. You edit a version with a plain sentence too: “remove the waiters”, “make the ending low”.
The blockout only handles motion and blocking. The AI takes looks, clothes and light from your prompt and photos, so grey figures don't get in the way.
From 4 to 30 seconds, 16:9, 9:16 or 1:1. The clip downloads as mp4 and works as a video reference right away.
Yes. Open a finished version and press “Edit camera”: fly around the scene and drop the points the camera will pass. The scene and people stay the same, no second AI run needed.
The price is shown on the start button before you press it. A new scene and an edit cost the same.