字节Seedance 2.0全面上线豆包和即梦，马斯克转发直呼"发展速度太快"

Core Insights - ByteDance announced the official launch of its latest video generation model, Seedance 2.0, which is now integrated into the Doubao App, desktop, and web versions, as well as the Jimeng App and its web version [1] Group 1: Product Features - Seedance 2.0 supports original sound and image synchronization, multi-camera long narratives, and multi-modal controllable generation [3] - Users can generate multi-camera video content with a complete native soundtrack by inputting a prompt and reference image, with the model automatically parsing narrative logic [3] - The web version supports four input modalities: image, video, audio, and text, allowing creators to specify visual styles, reference actions and camera movements, and convey rhythm and atmosphere through audio [3] Group 2: User Experience - Users can experience Seedance 2.0 for free after updating the Doubao App, with a demonstration showing a 10-second video generated in approximately 2 minutes from a prompt [3] - Each account has a daily limit of 10 free video generations, with 2 credits required for a 10-second video [3] - The Jimeng App also features this model but requires points, with 20 points needed to generate a 5-second video [3] Group 3: Usage Restrictions - Both Doubao and Jimeng Apps allow users to create digital avatars after verifying their identity through recording, but prohibit uploading images or videos of others as reference [4] - The platform is currently optimizing based on user feedback and has temporarily restricted the use of real human images or videos as reference [4] - ByteDance acknowledged that Seedance 2.0 is "far from perfect" and requires improvements in detail stability, multi-person matching, and text restoration accuracy [4]