# Directing a 30-second take instead of a single moving frame
Many AI video tools are built around a brief clip: a face turns, a landscape drifts, and the shot is over. Seedance 2.5 is framed around a longer take. One generation can run from four to thirty seconds, which is enough time to set up a shot, follow the action, and land an ending.
Contents
The site’s own example is a clockwork bird on a bench. In four seconds the bird might only cross the wood. In a longer take you can show the winding key, the release, and the flight as connected beats rather than as separate clips that have to be stitched later.
If the story still needs more room, the workspace can extend a clip from the last action and give the next shot its own direction. That is closer to directing a short scene than to rolling the dice on one isolated motion.
The product overview on Seedance 2-5 describes this 30-second window and the option to carry several cuts in one generation.