OpenAI has recently introduced its latest text-to-video generation model, Sora. Unlike other video generators, this new model has the ability to create complex scenes with multiple characters and specific types of motion, resulting in realistic imaginary scenes. It also has a deep understanding of language. However, it still has a few weaknesses, such as confusing spatial details of a prompt.
Currently, OpenAI is working with domain experts, known as red teamers, in areas like misinformation, hateful content, and bias, who will be testing the model. Besides generating a video solely from text instructions, Sora can also take an existing still image and generate a video from it, animating the image’s contents with accuracy and attention to small details.
Also, the OpenAI team working on safety according to their statement, “For example, once in an OpenAI product, our text classifier will check and reject text input prompts that are in violation of our usage policies, like those that request extreme violence, sexual content, hateful imagery, celebrity likeness, or the IP of others. We’ve also developed robust image classifiers that are used to review the frames of every video generated to help ensure that it adheres to our usage policies before it’s shown to the user.”
Sora can handle a variety of video formats, ranging from widescreen videos to vertical videos and everything in between. This allows Sora to easily create content for different devices without any aspect ratio issues. Additionally, Sora can extend videos forward or backward in time, providing more flexibility to the user.
