国泰海通｜计算机：Sora 2与Sora APP重磅发布，AI视频的“GPT-3.5时刻”

Core Insights - OpenAI's release of Sora App and Sora2 model marks a significant shift in AI technology from text and images to high-quality video generation and social applications [1][2] - Sora2 is described as a "GPT-3.5 moment for the video field," indicating its potential to revolutionize video content creation [2][3] Sora App Overview - Sora App focuses on "pure AI-generated content" and has quickly risen to the third position in the free app rankings in the U.S. App Store, indicating strong market interest [2] - The app features a TikTok-style interface but eliminates real human footage, relying solely on AI-generated materials [2] - There is a notable phenomenon of invitation code reselling in the domestic market, reflecting high demand [2] Sora2 Model Features - Sora2 has made significant advancements in accurately simulating physical movements and character performances, addressing limitations of previous AI video models [3] - It can realistically replicate complex movements, maintain scene continuity across multiple camera angles, and generate synchronized audio effects [3] - Current limitations include a fixed video length of 10 seconds, a resolution of only 360P, and challenges with complex scenes leading to facial blurriness and text generation errors [3] Social Interaction and User Engagement - Sora App aims to transform users from content consumers to creators through features like "Cameo" and "Remix," allowing for collaborative and creative engagement [4] - The "Cameo" feature enables users to create digital avatars for virtual performances, while "Remix" allows for secondary creation based on existing materials [4] - The app employs a non-addictive recommendation algorithm and includes parental controls to mitigate issues seen in traditional short video platforms [4]