Seedance 2.5 — 30秒一次性AI视频，带多模式参考

ByteDance 的下一代视频模型在一次传递中生成高质量的30秒音频 -视频剪辑 ，具有多轮扩展 ，用于更长的叙述。提供多达 30个图像、 10个视频和 10个音频剪辑 ，使用粘土渲染引用引导镜头 ，并通过时间戳级别的编辑优化结果。试用Seedance 2.5 Powered by ByteDance Seedance Team Seedance 2.5 ByteDance Seedance团队构的新一代视频创建模型。建是在Seedance 2.0统一的多模态音频-视频架构上，专注于基础生成和基于参考的创建—在长篇故事讲述、多模态参考和编辑方面取得了突破。 它一次性生成长达30秒的剪辑（ 2.0版为15秒） ，支持多分钟故事的多轮扩展，并以Seedance 2.0的统一多模态音频-视频架构为基础，专注于基础生成和基于参考的创建—在长篇故事讲述、多模态参考和编辑方面取得了突破。It generates up to 30-second clips in one pass (up from 15 seconds on 2.0), supports multi-round extensions for multi-minute stories, and accepts up to 30 images, 10 video clips, and 10 audio clips per job. Clay-render referencing helps lock composition and camera paths; timestamp-level editing, green-screen, and camera-perspective tools support film and advertising workflows. Audio and video stay jointly generated so dialogue, effects, and ambience stay locked to picture. Native 30-Second Single-Pass GenerationGenerate a complete high-quality 30-second audio-video clip in one pass — with stronger shot transitions and scene changes for continuous storytelling. Enough length for full ad beats, product demos, and multi-shot narrative arcs that shorter models struggle to deliver in a single generation. In one job, combine up to 30 images, 10 video clips, and 10 audio clips — plus stronger clay-render, motion, and creative referencing. Lock character identity, product look, style, and camera language across complex multi-subject scenes. Multi-Round ExtensionsAppend follow-up shots while keeping characters, environments, and pacing consistent. Official workflows use this to build multi-minute stories without manually splicing unrelated short clips. Precise Timestamp EditingControl narrative, camera, and rhythm for specific tim