OpenAI launches Sora, a revolutionary video tool powered by artificial intelligence | Technology

While the fascination with ChatGPT and generative artificial intelligence language models has yet to pass, OpenAI has just introduced a dazzling and revolutionary video creation tool called Sora. With it, simply give a description of what you want to see on the screen and there you have the clip, created by artificial intelligence. Some are more accomplished than others, they sometimes have this video game style which differentiates them from reality, but they are all surprising.

Sam Altman, CEO of OpenAI, announced the launch on the social network, which was flooded with new creations in no time. Realistic, futuristic, crazy videos, cartoons… Videos include all kinds of automatic creations created with generative artificial intelligence. Sora is capable of producing entire videos in one go or extending generated videos to make them longer.

In the tests shown by Altman and your business, you see complex scenes with multiple characters, specific types of movements, and precise details of the subject and background. The model understands not only what the user asked for in the query, but also how those things exist in the physical world, according to OpenAI. The model has a deep understanding of language, allowing it to accurately interpret signals and generate compelling characters that express vibrant emotions, the company says.

“This is Sora, our video generation model”, Altman wrote. “We offer access to a limited number of creators,” he added, before asking his subscribers for suggestions for creating new videos, in addition to the samples he had already offered on his website.

Instructions may be more or less detailed. One of the examples offered by OpenAI fits the following description: “An elegant woman walks down a Tokyo street full of warm, bright neon lights and bustling urban signage. She wears a black leather jacket, a long red dress, black boots and a black bag. She wears sunglasses and lipstick. Walk confidently and carefreely. The street is wet and reflective, creating a mirror effect of the colored lights. Many pedestrians are walking. And the result is surprising (above).

Another notes: “Trailer for a film about the adventures of the 30-year-old astronaut wearing a motorcycle helmet knitted from red wool, blue sky, salt desert, cinematic style, shot on 35mm film, bright colors .” (below).

In addition to being able to generate video purely from text instructions, the model is able to take an existing still image and generate a video from it, animating the image content with precision and attention to small details. The template can also take an existing video and enlarge it or fill in missing frames.

You can ask for content, style and give all kinds of instructions. Altman posts new videos requested by tweeters, proving that results are immediate. Sora can also create multiple shots in a single generated video while accurately maintaining characters and visual style.

“We teach AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems requiring real-world interaction.” explains OpenAI during the presentation of the new text-to-video conversion tool. “Sora can generate videos up to one minute long while maintaining visual quality and fidelity to user instructions,” he adds.

For the moment, the tool is made available to so-called red teams. Members of these teams attempt to question a product or service, push it to its limits, put it to the test, and find its faults as if they were enemies of the company. Here, they have the specific mission of assessing critical areas for potential damage or risks. Among them are experts in areas such as misinformation, hateful content and bias.

Open AI also provides access to a range of visual artists, designers and filmmakers to provide feedback on how to improve the model to make it more useful to creative professionals.

“We’re sharing our research progress in advance to begin working with people outside of OpenAI and receive their feedback, as well as to give the public a sense of the AI ​​capabilities that are on the horizon.” explains the company.

Defects to be polished

The artificial intelligence company itself acknowledges that Sora still has very obvious flaws. You might struggle to accurately simulate the physics of a complex scene and not understand specific cases of cause and effect. He gives the example that a person can bite into a cookie, but the cookie may then not bear the bite mark.

The model may also confuse the spatial details of a signal, for example by confusing left and right, and may have problems with precise descriptions of events that unfold over time, such as following a path of specific camera.

Before making the tool available to the public, OpenAI promises to take some precautions. Among them is taking into account the instructions of the red teams. Additionally, it creates tools to help detect misleading content, with detectors capable of knowing when a video was generated by Sora. It has also developed powerful image classifiers that are used to review the images of all generated videos and ensure that they comply with its usage policies before displaying them to the user.

Additionally, you will reuse the security methods you created for your products that use DALL-E 3. For example, the text classifier will check and reject text entry requests that violate your usage policies, such as those that require extreme violence, sexual image content. , images inciting hatred, images of celebrities or the intellectual property of third parties.

“We will reach out to policymakers, educators and artists around the world to hear their concerns and identify positive use cases for this new technology. Despite extensive research and testing, we cannot predict all the beneficial ways people will use our technology, nor all the ways they will abuse it. This is why we believe that learning from real-world use is a fundamental part of creating and launching increasingly safe AI systems over time,” OpenAI concludes.

You can follow EL PAÍS Technology In Facebook And X or sign up here to receive our weekly newsletter.