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

  • February 16, 2024

OpenAI said it shared a preview of Sora to give the public a sense of ‘what AI capabilities are on the horizon’, but the model is still in development to deal with its weaknesses.

OpenAI has shared details on its latest AI model – Sora – that can create up to a minute of high-quality video from simple text prompts.

The company said Sora is able to create videos that maintain video quality and adherence to the user’s prompt – key issues for current examples of AI-generated videos. The company has shared various example videos on its website and the prompts that were used to create them.

OpenAI claims Sora is able to generate complex scenes with multiple characters, specific types of motion, and accurate details of the subject and background. The company said it is sharing its research to give the public “a sense of what AI capabilities are on the horizon”.

The video examples look very high quality, but OpenAI noted that the model has some weaknesses currently – such as instances of inaccurate physics, mixing up left and right in prompts and suddenly creating new characters, usually in scenes that involve multiple entities.

OpenAI is not making Sora available to the public yet. The company has made the AI model available to “red teamers” – domain experts in areas like misinformation, hateful content and bias. These red teams will be “adversarially testing the model”.

Sora is also being made available to some visual artists, designers and filmmakers, so OpenAI can get feedback on how to make the model be “most helpful for creative professionals”.

“We’re also building tools to help detect misleading content such as a detection classifier that can tell when a video was generated by Sora,” OpenAI said in a blogpost. “In addition to us developing new techniques to prepare for deployment, we’re leveraging the existing safety methods that we built for our products that use Dall-E 3.”

AI has grown into a concern for some in creative roles such as artists and film writers. For example, the use of AI was one of the issues that led to the Hollywood writers’ strike last year, as there were fears that producers may start preferring AI-generated scripts over ones written by living writers.

A deal was reached last year that ended the strike. Under these rules, AI can’t be used to write or rewrite literary material and AI-generated content cannot be considered source material.

Find out how emerging tech trends are transforming tomorrow with our new podcast, Future Human: The Series. Listen now on Spotify, on Apple or wherever you get your podcasts.

#OpenAI #unveils #Sora #texttovideo #model