A OpenAI revealed this Thursday (15) a tool that can generate videos from text instructions.
The new model, dubbed ‘Sora’ after the Japanese word for sky, can produce realistic images up to a minute long that follow the user’s instructions on both subject and style. According to a post on the company’s blog, the model is also able to create a video based on a still image or augment existing footage with new material.
We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction, the blog post says.
A video included among several early examples from the company was based on the suggestion: A movie trailer featuring the adventures of the 30-year-old spaceman wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on film 35mm, vivid colors.
The company announced that it has opened access to Sora to some researchers and video creators. Experts would red-team the product, testing it for susceptibility to circumventing OpenAI’s terms of service, which prohibit extreme violence, sexual content, hateful images, celebrity likenesses or third-party IP, according to the blog post. from the company. OpenAI only allows limited access to researchers, visual artists and filmmakers, although CEO Sam Altman responded to user requests on Twitter following the announcement with video clips that he said were made by Sora. The videos have a watermark to show that they were made by AI.
The company launched the Dall-E still image generator in 2021 and the generative AI chatbot ChatGPT in November 2022, which quickly accumulated 100 million users. Other AI companies have released video generation tools, although these models have only been able to produce a few seconds of footage that often have little relationship to their prompts. Google and Meta have said they are developing generative video tools, although they have not released them to the public. On Wednesday, it announced an experiment to add deeper memory to ChatGPT so it could remember more of its users’ chats.
OpenAI did not disclose how many images were used to train Sora or where the training videos may have originated, other than telling the New York Times that the corpus contained videos that were publicly available and licensed by copyright owners. The company has been sued multiple times for alleged copyright infringement in training its generative AI tools, which digest massive amounts of material scraped from the Internet and imitate the images or text contained in those datasets.
*With information from The Guardian/ Cover photo: Disclosure/ OpenAI
Follow Adnews on Instagrame LinkedIn. #WhereTransformationHappens
