Seedance 2.5 on CapCut: 30s Videos, Lip Sync & Vital Prompting Tricks

minchoi · x · 2026-08-07

The author tested Seedance 2.5 integrated into CapCut and shared its core capabilities and prompting tutorials. The new model supports generating up to 30-second scenes in a single pass, accepts up to 50 references, and can generate real dialogue with accurate lip sync.

Using the viral "miniature city flooded in a water tank" video as an example, the author breaks down the workflow:

The article includes 5 generated video examples and their exact prompts for users to replicate.

Related event: Seedance 2.5 Global Launch: 30s Multi-Reference Video Generation(62 posts)→

Original post →

More from Multimodal

Multimodal channel →