阿里电影级视频模型万相2.6系列上线，功能比Sora2还全，人人都能当导演

Core Insights - Alibaba has launched the new Tongyi Wanshang 2.6 series model, which includes five new models that enhance capabilities in video and image generation, covering various creative processes from single-use generation to reusable creation [2][5] - The Wanshang 2.6 model is the first in China to support character role-playing in video generation, with improvements in video quality, sound effects, and adherence to instructions, achieving a maximum video length of 15 seconds [2][4] Model Features - The Wanshang 2.6 model integrates multiple innovative technologies for multi-modal joint modeling and learning, allowing it to extract and maintain consistency across visual and auditory features during video generation [7][9] - It can convert simple user prompts into multi-scene scripts, generating coherent narrative videos while maintaining consistency in key elements like subjects and scenes [9][11] User Experience - Users can upload personal videos and input prompts to quickly generate narrative videos with cinematic quality, enabling anyone to take on a director's role [9][11] - The model supports various applications, including AI comic creation, advertising design, and short video production, with over ten visual creation capabilities available [12] Image Generation Enhancements - The model has improved in style control and expression stability, allowing for better integration and transition between different artistic styles while reducing the "AI feel" in generated realistic portraits [13][15] - It can generate posters, illustrations, or infographics based on longer, structured text, enhancing the clarity of the relationship between content and visuals [15][19]