Seedance 2.5 Image to Video creates a continuous video shot between the first and last frames of a selected Generate Region span.
Kepler renders both boundary frames from the timeline and sends them as the required starting image and optional ending image. The model derives aspect ratio from those references. Resolution offers Match Canvas, which chooses the nearest supported 480p, 720p, or 1080p output for the project canvas. Duration can be set from 4 to 30 seconds.
With Native Audio enabled, Seedance generates ambience, sound effects, and spoken dialogue together with the video at no extra model charge. For lip-synced dialogue, put the spoken line in quotation marks and identify the speaker clearly. Lip movement, timing, and exact continuity can still vary between generations.
Write one prompt describing what changes between the boundary frames: subject motion, camera movement, sound, and the intended final state. Seedance does not accept a Negative Prompt, Prompt Strength, reference Elements, or an input Seed on this endpoint, so those controls are not shown. High Bitrate requests a larger, higher-quality output encode.
The generated result is stored as new media and placed on a generated track inside the selected region. Existing timeline media is never overwritten.