top of page
Voltaire Staff

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



OpenAI has unveiled Sora, an AI model designed to transform text prompts into engaging minute-long videos.


The Microsoft-backed firm announced the service in a blog post on its website on Thursday, "Introducing Sora, our text-to-video model. Sora can generate videos up to a minute long while maintaining visual quality and adherence to the user’s prompt."


Sora is currently accessible for purposes that help identify flaws within the AI system.


It is also open for use by visual artists, designers, and filmmakers seeking to solicit feedback on the model's performance to best serve the needs of creative professionals.


In 2022, the company launched its DALL.E, a text-to-image generator. It is now in its third variant, DALL.E 3, launched last October.


"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," OpenAI said in the statement.


In addition to making videos from text, Sora can bring still images to life.

Moreover, Sora has the ability to craft several scenes in one video, maintaining consistency in characters and visual style throughout, the company claimed.


Meanwhile, Meta has also enhanced its image generation model, Emu, to add two AI features for editing and generating videos from text prompts.

Sora is still in development, and the company notes that the model might struggle with understanding the spatial details of a prompt and following a specific camera path.


"The current model has weaknesses. It may struggle with accurately simulating the physics of a complex scene, and may not understand specific instances of cause and effect. For example, a person might take a bite out of a cookie, but afterward, the cookie may not have a bite mark," the company said.


It added, "The model may also confuse spatial details of a prompt, for example, mixing up left and right, and may struggle with precise descriptions of events that take place over time, like following a specific camera trajectory."


OpenAI also said that it is prioritising several critical safety measures prior to integrating Sora into its products. Collaboration with red teamers, specialists in fields like misinformation, hateful content, and bias, is underway to test the model under adversarial conditions.

 

 

 

Comments


bottom of page