OpenAI’s Sora text-to-video generator will be publicly available later this year

An animated piece with a distinctive 3:2 aspect ratio. The image imbued with an uplifting and bright style, replicating the impression of a 3D animation studio. The focus of the illustration should be on a large, three-dimensional cube that symbolizes the Sora text-to-video generator by OpenAI. The cube is busy creating hyperrealistic scenes, depicted by miniature versions of a bull in a china shop and a mermaid reviewing a smartphone. On another side of the cube, a small watermark signifies the generator’s output. Looking on, a diverse group of bystanders reveal the varied target audience of visual artists, designers, and filmmakers, all expressing excitement and anticipation for the upcoming public release.

OpenAI’s Sora text-to-video generator, capable of creating hyperrealistic scenes from a text prompt, will be available to the public later this year. Initially targeted at visual artists, designers, and filmmakers, Sora is set to incorporate audio in the future to enhance realism. OpenAI aims to enable users to edit the content in the videos produced by Sora, acknowledging the need for human intervention due to AI’s occasional inaccuracies. The training data for Sora remains undisclosed, with Mira Murati mentioning that it was publicly available or licensed, and confirming the use of content from Shutterstock through a partnership. Despite concerns about generative AI tools and misinformation, Sora’s release will likely restrict the generation of public figure images and include watermarks on videos. However, the effectiveness of watermarks as a solution is questioned. The tool’s cost is a concern, but OpenAI aims to make it accessible at a similar cost to DALL-E, their AI text-to-image model. The Wall Street Journal’s report showcases examples of Sora’s video capabilities, including an animated bull in a China shop and a mermaid smartphone reviewer.

Full article

Leave a Reply