OpenAI introduces Sora, its text-to-video AI model

OpenAI has just introduced Sora, its text-to-video AI model to rival the likes of Midjourney, Runway, Pika and even Google’s Lumiere figures.

According the AI company, “Sora can generate videos up to a minute long while maintaining visual quality and adherence to the user’s prompt.”

“Sora is able to generate complex scenes with multiple characters, specific types of motion, and accurate details of the subject and background. The model understands not only what the user has asked for in the prompt, but also how those things exist in the physical world”

Possessing a profound comprehension of language, Sora adeptly interprets prompts and crafts engaging characters that vividly convey a range of emotions. Additionally, it has the capability to generate multiple scenes within a single video, ensuring the faithful persistence of characters and visual style.

Sora builds on past research in DALL·E and GPT models. It uses the recaptioning technique from DALL·E 3, which involves generating highly descriptive captions for the visual training data. As a result, the model is able to follow the user’s text instructions in the generated video more faithfully.

Presently, Sora is being made accessible to “red teamers” for evaluating crucial domains for potential harms or risks. Additionally, OpenAI is providing access to a variety of visual artists, designers, and filmmakers to gather insights on enhancing the model for optimal assistance to creative professionals.

Furthermore, OpenAI is developing tools designed to identify misleading content, including a detection classifier specifically designed to recognize videos generated by Sora.

OpenAI will be releasing details about its research in the early stages to engage with and receive feedback from individuals beyond the OpenAI community. This initiative aims to provide the public with an understanding of the emerging AI capabilities on the horizon.