OpenAI is known for ChatGPT and Dall-E and has announced the new AI tool Sora. Sora easily converts text into realistic videos up to 60 seconds.
Megan van der Wagt
OpenAI is best known for its popular AI tools such as ChatGPT and Dall-E. Now this Microsoft subsidiary has launched a new AI tool that converts text into a realistic video. Sora mainly wants to strive for high visual quality and accurate execution.
How does Sora work?
Sora, like ChatGPT, is easy to use. Via a prompt (a short script with a command) it is possible to create videos of up to 60 seconds via Sora. All you have to fill in is what you want to see in the video with as much detail as possible and then Sora tries to visualize it as best as possible.
What can Sora do?
With Sora it is possible to create both realistic and surreal videos. Scenes with animals, nature reserves and multiple characters, for example, are no problem for Sora, because he takes the background into account and understands movements. “The model understands not only what the user asked in their prompt, but also how things exist in the physical world,” OpenAI said in a blog post. The model has extensive language knowledge and can therefore generate a video with different camera angle options based on one prompt. So many visual styles are possible.
Another statement from OpenAI suggests that Sora could also create simulations of the real world, which means game design. So Sora could go beyond just creating videos of up to 60 seconds.
Complications
According to OpenAI, Sora can do a lot, but occasionally has complications. For example, the model occasionally has difficulty accurately simulating complex scenes and the program has difficulty understanding the distinction between cause and effect.
Abuse
Sora is not currently available for use. At OpenAI they are still working internally to make the behavior of the model acceptable, as there are many consequences associated with such a program. OpenAI believes that misinformation and hate have no place in the end product, which is why OpenAI is now working on tools that, among other things, detect misinformation. A kind of watermark will also appear in the videos, so that it is easy to distinguish Sora’s images from realistic images.
