Core Viewpoint - Alibaba has launched the next-generation Wanxiang 2.6 model, which is the first video model in China to support character role-playing, enhancing video creation capabilities significantly [1][2]. Group 1: Model Features - Wanxiang 2.6 supports audio-visual synchronization, multi-shot generation, and sound-driven functionalities, making it the most comprehensive video generation model globally [1]. - The model has improved video quality, sound effects, and instruction adherence, achieving a maximum video length of 15 seconds, which is the highest in China [2]. - It can generate videos featuring single or multiple characters and objects, automatically performing multi-shot transitions to meet professional film-level requirements [2][3]. Group 2: Technical Innovations - The model integrates multiple innovative technologies for multi-modal joint modeling and learning, capturing emotional, postural, and visual features from input reference videos [3]. - It extracts acoustic features such as voice tone and speech rate to ensure consistency across visual and audio elements during the generation phase [3]. Group 3: User Experience - Users can convert simple prompts into multi-shot scripts, creating coherent narrative videos while maintaining consistency in key information across shots [4]. - The character role-playing feature allows ordinary users to perform in cinematic-quality visuals, enabling quick generation of narrative videos with minimal input [4]. - Wanxiang 2.6 can also be utilized for advertising design and short film production, allowing users to act as directors by inputting creative prompts [4]. Group 4: Accessibility and Applications - The model is now available for all users on the Wanxiang official website, with enterprise users able to access the model API through Alibaba Cloud [5]. - The Wanxiang model family supports over ten visual creation capabilities, including text-to-image, image editing, text-to-video, and video editing, widely applied in AI comics, advertising design, and short video creation [5].
阿里发布通义万相2.6系列视频生成模型,上线国内首个角色扮演功能 | 钛快讯