Seedance 2.0
Seedance 2.0
DokumantasyonBaslangic
Start
StartCore Capabilities
Multimodal Performance
Seedance 2.0 ManualUse Cases

Core Capabilities

Beyond multimodality, Seedance 2.0 brings major improvements to its core fundamentals: more plausible physics, more natural and fluid motion, more accurate instruction following, and more stable style consistency. It can reliably handle difficult generation tasks like complex actions and continuous movement, while making overall videos look more realistic and smoother.

Cases:

prompt

img

vid1

Notes

A girl gracefully hangs laundry to dry. After finishing, she pulls another piece from the tub and gives it a firm shake.

Ultra-realistic

Inside the painting, the character looks guilty. Their eyes dart left and right as they lean out of the frame, quickly reach out to grab a cola, and take a sip. They then smile in satisfaction. Footsteps approach—the character hurriedly puts the cola back in place. A Western cowboy picks up the cola from the cup and walks away. Finally, the camera pushes in as the scene fades into a pure black background, lit only by a top light shining on the canned cola. Artistic subtitles appear at the bottom, with a voice-over: “Yikou Cola—You have to try it!”

The camera gently pulls back (revealing a full street view) and tracks the heroine as she walks along a 19th-century London street. The wind flutters her skirt. As she continues, a steam-powered vehicle speeds in from the street on the right and rushes past her; the gust lifts her skirt and she looks shocked, quickly pressing it down with both hands. Background audio includes footsteps, crowd noise, vehicle sounds, and more.

The camera follows a man in black fleeing at full speed, with a crowd chasing him. It switches into a side tracking shot: he panics, knocks over a roadside fruit stand, scrambles back up, and keeps running amid the crowd’s chaotic noise.

Start

Öncesi

Start

Sonraki