OpenAI introduced Sora, a hyper-realistic video generator that works with artificial intelligence

Company OpenAI, creator of ChatGPT and image generator DALL-E, introduces a new development that increases the risks of synthetic content. The company demonstrated the results it delivers Sora, an AI video generator with realistic aesthetics And up to one minute of text instructions.

The California-based company explained that this new platform, which was created based on previous research in DALL-E and ChatGPT, It is still in the testing phase. However, to pique the interest of its customers, on Thursday it shared some examples of hyper-realistic videos created from its text titles.

“Introducing Sora, our text-to-video model. Sora can get creative Videos up to 60 seconds long with highly detailed scenes, complex camera movement, and multiple characters with lifelike emotions. “The software can create one-minute videos, with visual quality and respecting user request,” OpenAI said on its website.

This was stated by the company's CEO, Sam Altman Currently, they are granting “access” to a limited number of “visual artists, designers and filmmakers.”For feedback on “How to develop the model to make it more useful for creative professionals.”

Through this proposal, the company seeks to “teach artificial intelligence to understand and simulate the moving physical world, with the aim of training models that help people solve problems that require interaction with the real world.”

“Sora has a deep understanding of language, allowing him to accurately interpret signals and generate compelling characters that express vivid emotions. He can also create multiple shots within a single created video, where the characters and visual style are accurately continued.”

Additionally, OpenAI said the platform allows for “creating video from a still image” and “expanding existing videos.”

This is what AI-generated videos look like

Through the X social network, both OpenAI and Altman were involved Examples of videos created through the Sora modelin addition to text instructions for each result.

“The beautiful snowy city of Tokyo is full of life. The camera moves through the busy city streets, following many people enjoying the beautiful snowy weather and shopping at the nearby stalls. Gorgeous sakura petals blowing in the wind along with snowflakes”:

“Film trailer featuring the adventures of a 30-year-old spaceman in a red wool motorcycle helmet, blue skies, and salt desert, in a cinematic style, shot on 35mm film, in vivid colors”:

“An elegant woman walks down a Tokyo street filled with warm, bright neon lights and animated city signs. She is wearing a black leather jacket, a long red dress, black shoes, and a black bag. She is wearing sunglasses and red lipstick. She walks confidently and simply. The street is damp and reflective, creating a mirror effect of the colored lights. Many pedestrians walking”:

“Bicycle race in the ocean with different animals as athletes riding bikes with drone camera view”:

“A futuristic drone race at sunset on Mars”:

Sora's flaws

Meanwhile, OpenAI has warned about thisThe “current model” of the platform has “flaws”, such as confusion between left and right, or the impossibility of maintaining visual continuity for the duration of a video. “Anyone can eat a cookie, but there may not be a bite in the cookie,” the company explained.

On the other hand, the company stated that security is a major issue in this development, and stated that it will be regulated Simulations with users who will be asked to attempt to cause errors or create inappropriate contentin order to better define the boundaries of the platform.

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