Seedance 2.5 Test: Generates Video with Native Audio from a Single Reference Image

aziz4ai · x · 2026-08-11

A creator demonstrated the powerful capabilities of the AI video generation tool Seedance 2.5. They first generated a reference image using Midjourney, then fed it into PixVerse's Seedance 2.5, producing a high-quality video using only direct prompting.

Impressively, the music in the video is native Seedance audio, showcasing the model's potential in audio-visual multimodal generation.

Related event: Seedance 2.5 Tested: Impressive Long Video and Native Audio Generation(7 posts)→

Original post →

More from Multimodal

Multimodal channel →