OpenAI’s text-to-video Sora

A new company supported by Microsoft is impressing people on social media by making incredibly realistic videos using just text prompts.

OpenAI, the company behind ChatGPT, has revealed a groundbreaking AI that generates lifelike videos from text prompts, causing a stir of surprise of the quality of the videos across the internet.

OpenAI announced in a blog post on Thursday that their text-to-video model, called Sora, possesses "a profound grasp of language" and has the ability to create "captivating characters that convey vivid emotions."

"The model comprehends not just the user's prompt but also the context of how those elements manifest in the real world."

During a recent event, OpenAI CEO Sam Altman invited users to submit prompts for Sora. The results showcased realistic videos, such as two golden retrievers podcasting atop a mountain, a grandmother preparing gnocchi, and marine animals participating in a bicycle race above the ocean.

In its blog post, OpenAI mentioned that it would be implementing several crucial safety measures before making Sora available to the public.

"We're also developing tools to identify misleading content, including a detection classifier capable of determining whether a video was generated by Sora."

OpenAI also recognized that Sora has its limitations, such as struggling with maintaining continuity and differentiating between left and right.

"For instance, a person might be shown taking a bite out of a cookie, but afterwards, the cookie may not accurately show the bite mark," stated the San Francisco-based startup.