OpenAI, the creator of ChatGPT and DALL-E, continues to be a step ahead in the development of artificial intelligence solutions. This is demonstrated with the experimental launch of Sora, a new video generation tool from simple text instructions.
Sora is an innovation in the field of generative artificial intelligence. It is a text-to-video model that can create deeply realistic clips that respond to prompts. It is also capable of processing static images into dynamic presentations, as well as enlarging or reducing videos for editing purposes, always under the specifications given in a text instruction.
Sora can accurately understand user requests, regardless of their level of complexity, given his high understanding of human language. However, according to OpenAI, what sets Sora apart from other text-to-video models is its understanding of the physics of objects. From there, it is able to recreate real or imaginary situations with great cohesion and precise details of characters and backgrounds.
However, this high level of detail in Sora’s results has once again brought into debate the possibilities that AI models offer to generate false content. Thus, at first, OpenAI is only launching this product to “red team members,” a community of experts in areas such as disinformation and hate speech who will evaluate the model under adverse conditions. At the same time, they are working on developing fake content detection tools before releasing Sora to the general public.
Where can you see content created with OpenAI’s Sora?
To appreciate the amazing videos created with Sora you just have to go to the page that OpenAI maintains on the Internet. Its web address is https://openai.com/sora.
There you can also access the technical report of the product and additional details.
Image: OpenAI