Meta is working on video generation for the standalone Meta AI version. Currently, in addition to text conversations, users can click the Imagine button for image generation or get inspiration from highlighted prompts on the Imagine page. In the future, it looks like users will be able to generate videos directly and combine them to create a bigger story.
Here, instead of images, you see highlights of other videos or prompts. Once you enter the video prompt or select the highlighted one, you are taken to a progress page with multiple placeholders for video generation. These videos are portrait oriented, making them suitable for Instagram Reels. This feature indicates that they may be suitable for AI video generation for the Reels platform.
A highlighted prompt may contain different scenes split into separate video files. If the prompt contains one object, this object will appear in all videos. For more complex scenes, different scenes will appear in separate videos. This approach is similar to Google's storyboard method of video generation, but Meta lets the user explain everything in the prompt and generates multiple video shots that can be combined later. Currently, to combine them into a single video, you need to download each video separately and combine them manually.
Additionally, hints within the codebase suggest that the LLama 3 405B model may be available soon, and may even be exclusive to the premium plan. The Meta AI premium offer rumored by The Information appears to be included in the code and may be available alongside the release of the LLama 3 405B model, scheduled for July 23rd.
This suggests that there may be some announcements related to Meta AI products tomorrow. Additionally, the code mentions AI Studio, which may be a separate product targeted at developers. Meta AI may offer a solution similar to OpenAI Playground or Google AI Studio, allowing users to try out Meta AI models and seamlessly integrate them into their own products.
