OpenAI is launching a new text-to-video AI model called ‘Sora,’
that is designed to produce photorealistic videos from text-based descriptions.
Recently, AI video generating models have started to assert
their presence in the market. We’ve got companies like Runway and Pika that
have developed their own text-to-video models. There is also Google’s Lumiere
figures that are one of OpenAI’s leading competitors.
According to OpenAI, Sora’s capabilities allow it to judge how objects exist in the physical world, as well as some complex phenomena like emotions. As a result, it can create complex scenes with multiple characters, specific types of motion, and accurate details of the subject and background.
Sora can perform multiple different actions with videos, including
generating videos from still images, completing missing frames on existing
videos, and extending videos. Sometimes though, as noted by OpenAI, the model might
have some difficulty correctly simulating the physics of a complex scene and might
not properly interpret certain cases of cause and effect. However, the results generated
are still relatively impressive.
Sora is still an experiment and is currently available to
“red teamers” who are reviewing the model for potential harms and risks. Simultaneously,
OpenAI is offering its access to some visual artists, designers, and filmmakers
to obtain initial feedback.
