Artificial Intelligence: Sora from OpenAI converts text into video

OpenAI, the creator of ChatGPT and image generator DALL-E, has introduced a new tool Sora, which is able to create realistic videos that, once text is entered, can last up to a minute – a major innovation in the field of artificial intelligence.

Based on previous research on DALL-E and GPT, this new platform is still in testing, explained the California-based Microsoft startup ally, who did, however, provide some how-to videos.

Sora can create complex scenes with multiple faces, specific types of movements, and fine details

OpenAI said on its website that the software can create videos of a maximum of one minute “while maintaining visual quality and respecting user request.”

The startup says on its website that Sora can “create complex scenes with multiple faces, specific types of movements, and fine details.”

Sora also allows for creating videos from a still image, the AI ​​giant confirms, or expanding existing videos.

https://t.co/qbj02M4ng8 pic.twitter.com/EvngqF2ZIX

— Sam Altman (@sama) February 15, 2024

Experimental phase

Sam Altman, head of OpenAI, said on social media

The company hopes that Sora will be an important milestone in achieving artificial general intelligence

He also invited users to submit suggestions for creating videos, and after a few minutes he uploaded the most successful clips to the platform.

Among these videos, we see two dogs playing in the snow on a mountain. Another video clip shows an imaginary animal, half duck and half dragon, flying in front of a beautiful sunset, with a hamster wearing a flight costume on its back.

Sora serves as the foundation for “software capable of understanding and simulating the real world,” explains the startup, which it hopes will “be an important milestone in the realization of artificial general intelligence,” a highly autonomous system it is claimed will outperform humans at most effective functions. In terms of cost.

https://t.co/rPqToLo6J3 pic.twitter.com/nPPH2bP6IZ

— Sam Altman (@sama) February 15, 2024

Defects

OpenAI warned that the platform's “current model” has “flaws,” confusing left and right and showing an inability to maintain visual continuity throughout the video.

“For example, a person may bite into a cookie, but then there may be no bite marks on the cookie,” the company explains.

When unveiling this new tool, OpenAI said that the issue of security is of great importance and that simulations where users are asked to glitch or create inappropriate content will be organized to better define the boundaries of the platform.

“We will invite policymakers, educators and artists from around the world to understand their concerns and identify positive use cases for this new technology,” OpenAI said.

Meta, Google, and Runway AI, which are working on similar text-to-video applications, have already demonstrated models.

Source: Accuracy