据直播介绍，Sora可以生成1080p、最长20秒的视频，支持横屏（16:9）、正方形（1:1）和竖屏（9:16）多种常见比例，并且可以对生成的视频进行Remix（重混）：替换、删除或重构视频中的元素，Re-cut（重新切割）：找到最佳的一帧再进行延展，循环剪辑，混合：将两个视频进行无缝合并剪辑，故事板剪辑：精准指定每个帧的输入，风格预设：预设创建视频的风格等操作。The domestic video large model is launched, and the film and television industry is accelerating its landing pace.
On December 9 local time, OpenAI CEO Sam Altman had a live broadcast for about 20 minutes with Sora team leader Bill Peebles and Aditya Ramesh.
The difference this time is that the Sora Turbo version is being released. OpenAI stated that since the release of Sora in February this year, development of Sora Turbo has been underway, which will be much faster than the previous version.
According to the live broadcast introduction, Sora can generate 1080p videos up to 20 seconds long, supporting various common aspects such as landscape (16:9), square (1:1), and portrait (9:16), and can Remix the generated videos: replace, delete, or reconstruct elements in the video, Re-cut: find the best frame and extend it, loop editing, blending: seamlessly merge two videos, storyboard editing: precisely specify the input for each frame, style presets: create video styles through presets, and other operations.
According to the official introduction, Sora also supports exploring community features, where user-generated videos can be shared to the community, and users can view, search, or recreate videos shared by others in the community (users can also choose to disable this feature).
OpenAI CEO Sam Altman revealed on social media: 'During the next 12 working days, there will be a live broadcast every day. The company will showcase some new products or samples.' Following this announcement, AI concept stocks rose significantly in many global markets.
According to Zhito Finance APP, this year, video generation has become one of the hottest sub-sectors in the field of domestic large models.
So far, companies like Tencent, Kuaishou, and SenseTime have successively launched products related to video generation.
Listed companies in the film and television industry are embracing AIGC to empower their main business.
Industry insiders believe that the multimodal video large models have gone through iterations and have gradually improved their enabling effects on content production, and the commercial space for AI in the film and television sector is expected to open up in the future.
Hong Kong stock companies related to text-to-video:
Kuaishou-W (01024): According to Kuaishou, all these short films were generated using Kecai AI, with film directors completely relying on the large video generation model, and the community deeply participating in movie-level content creation, which is the first of its kind in China. In July this year, Douyin and Kuaishou successively launched two AI micro-short dramas "Sanxingdui: The Future Revelation" and "Shanhaijing: The Cutting Waves", both of which had no live-action shooting or real human performances, yet gained significant attention.
Tencent (00700): Tencent's Hunyuan large model officially launched video generation capability and has open-sourced the video generation large model with a parameter count of 13 billion, making it the largest open-source video model currently available.
Alibaba-SW (09988): Alibaba launched the Tongyi Wuxiang video generation function in September, supporting the generation of up to 5 seconds of high-definition film-level video.
Meitu (01357): Meitu upgraded the video generation capability of its Qixiang large model in September.
Weimob Inc (02013): The company's relevant technical leader stated that the marketing field has been confirmed as one of the best application scenarios for AI large model technology in the past year. Sora will reshape the short video marketing industry, and this will become one of the most important focal points for Weimob in 2024.
