OpenAI, one of the leaders in generative artificial intelligence, has launched a new tool to continue dominating the world of AI. This is Sora, a model that allows you to take text instructions and transform them into realistic video scenes with multiple characters and specific movements.
According to the company, it was founded in 2015 and is led by Sam Altmanthat has developed ChatGPT and the still image generator DALL E“Sora” is capable of creating “complex scenes with details of the main component and background.” The company also stated that its model can do this “Understanding how objects exist in the physical world” and has the ability to “correctly interpret instructions and Generate convincing characters that express vivid emotions.
“Sora” is the Japanese term for “Treasure” and the company has not yet made it available to the public as this is still the case analyze it to understand the dangers of the system, according to the New York Times. Instead, OpenAI has shared the technology with a small group of academics and other outside researchers whose mission is to do so Identify ways Sora could be used for malicious purposes.
Wired Magazine’s Steven Levy described his experience when OpenAi allowed him to view, but not operate, the “Sora” model in response to a text instruction to create a Tokyo city scene captured by a moving camera. Follow some people on the streets on a snowy day. “The result is a view convinced of what it is, without a doubt, “Tokyo at this moment when snowflakes and cherry blossoms coexist,” Levy wrote.
“The virtual camera, placed on a flying robot (drone), follows a couple walking through the landscape.” The company states that “Sora” allows the creation of many worlds, with realistic or fantastical, three-dimensional scenes, with fluid movements , and even in scenes that do not exist in the real world, the model applies the correct physics rules. In this way, OpenAI, which has revolutionized the emerging regenerative AI services sector, becomes a A range of services ranging from copywriting to static images and videos.