Dreamina Launches Seedance 2.5 with 50-Reference Video Generation

Div_pradeep · x · 2026-07-31

Dreamina has officially launched Seedance 2.5, focusing on demanding, long-context video generation.

The new model supports up to 50 reference materials—including images, videos, and audio—in a single generation. This allows creators to define characters, environments, visual styles, and scene moods with high precision, sustaining outputs for up to 30 seconds. The author calls for rigorous testing against competitors like Omni and Kling under these complex conditions.

Related event: ByteDance Releases Seedance 2.5: Native 30s Video and Multimodal Control(60 posts)→

Original post →

More from Multimodal

Multimodal channel →