On October 1st, Beijing time is early this morningOpenAI The next generation was releasedVideo Generation Model Sora 2.
This new model is significantly enhanced in the visualization, andAdd Audio GenerationI don't know. It inherits and expands early image generation techniques, and in the new version of App launched during the same period, users can complete their identification by recording their own video and voice in one-off, and then “passenger” themselves or others in the resulting video。
Sora applications are similar to existing social media, provides an algorithm-recommended stream of information, promotes individualized content based on interactive objects and interests, and introduces a `regulated sorting' function to allow users to further customize what they see。
Sora 2 continued the original model released in February 2024. It's more rigid than the first generation when it comes to basketball, etcIt's more natural in physical behavior.
Unlike the previous generation, Sora 2 also produces voice. OpenAI says that although the model is far from perfect and there are still a lot of mistakes, it validates that it is possible through a larger video trainingIt's a simulation of the real world.
Sora application currently on iOS, but with an invitation system, userEligibility required through applicationI don't know. OpenAI indicated that it would take the lead in gradually opening up in the United States and Canada and giving users a lighter amount of creativity at an early stage. However, because video generation is extremely difficult to calculate, companies limit its use to ensure that services are stable。
OpenAI, frankly, the only business plan at this time is when there's not enough time to figure it outOptions for generating additional video at a fee for users in need.
In response to security concerns, OpenAI simultaneously published a blog note. According to information from the IT House, the company emphasizes that all Sora videos will carry watermarks and metadata markers and that portraits of users can only be used with their consent and can be revoked at any time. Adolescent user accounts are subject to parental control and time constraints。
At the production stage, systematic protection measures attempt to shield unsafe content, including pornography, terrorist propaganda and self-harm related content, and intercept through multiple frames and audio tests. OpenAI also plans to expand the manual clearance team to focus on videos that may involve bullying。
OpenAI describes Sora 2 as an important step towards a stronger AI system. The company said: “Video models are evolving at a fast pace. Universal World Simulator and robotic intelligence will reshape society and accelerate human development.”