About this model
Seedance 2.5 US R2V is the reference-to-video ("R2V") route of ByteDance's flagship Seedance 2.5 video family, deployed on US infrastructure. Where text-to-video starts from a prompt alone, R2V lets creators supply mixed reference assets — images, video clips and audio — and refer to them in the prompt to drive character consistency, style and motion transfer, camera blocking and lighting. It sits alongside the text-to-video Seedance 2.5 US and image-to-video Seedance 2.5 US endpoints, and mirrors the non-US Seedance 2.5 R2V.
ByteDance describes Seedance 2.5 as a next-generation audio-video joint generation model built for 30-second storytelling with flexible referencing and editing capability. Relative to the earlier Seedance 2.0 R2V line, the launch materials emphasise longer single-take generation, broader multimodal referencing, and consistency-preserving editing and extension of shots already produced. ByteDance also reports work on the "artificial" look of generated footage — object textures, skin and eye detail, lighting physics and colour — plus steadier long takes with continuous camera movement.
Note that ByteDance has not published a technical report for Seedance 2.5, so the above reflects the provider's own launch presentation rather than independent measurement. Within the catalogue, this endpoint is the full Seedance 2.5 tier, distinct from the separately named Seedance 2.0 Mini R2V and Seedance 2.0 Fast R2V variants.
This About section is AI-generated from public sources (Claude Opus 5), with no human editing. It may contain inaccuracies — verify critical details against the sources listed above.
Data sources: Venice API · HuggingFace · Wikipedia — enrichment updated 2d ago