OpenAI, a research group dedicated to developing artificial general intelligence (AGI), has revealed that they will be releasing Sora, an AI model that can create realistic-looking, high-quality films up to one minute in length just from text instructions, only fourteen months after the wildly popular introduction of ChatGPT.
In its introductory blog post, OpenAI released a selection of video samples with complex scenarios. The samples were produced using text prompts, and the outcomes appear strangely realistic.
The model does an excellent job of comprehending how complicated parts exist in the actual world, and the films are extremely detailed. Sora seems to be well-versed in both real-world operations and the technicalities of filming, with a keen awareness of cinematic devices like camera angles and timing.
The company claims that Sora is based on data from the DALL-E 3 and GPT models that are currently in use. In contrast to DALL-E, which is only concerned with artistic outputs, Sora is far more potent and has practical purposes. As a data-driven physics engine, the model itself is capable of simulating physical interactions and real-world situations in addition to producing images.
Moreover, Sora can generate and interact with entire virtual worlds, whether imaginary or realistic, aside from producing individual photos and movies. This will not only have scientific uses, but the technology may also aid in virtual reality and gaming creation; in fact, the timing of the release could not be more ideal.
However, OpenAI notes that the model is not yet complete and could have trouble comprehending some instances of cause and effect and accurately modeling the physics of a complex scenario. The video samples also seem to have been chosen with care and detail to highlight the best features of Sora.
To date, Sora is still going through several stages of safety testing and is not yet available to the general public, according to OpenAI. Only a group of specially assigned red testers who are actively looking for potential ways the service may be utilized have access to it.
Although we have previously seen AI video models such as Runway and Google Lumiere, Sora seems to surpass them all. One of its many abilities is its capacity to accurately simulate real-world events due to its profound comprehension of the physics and existence of the real world. However, the length of footage it can produce is what really makes it stand out.
Will Sora eventually open the door for AI to take over the film industry?
Other POP! stories that you might like:
Google faces backlash over ‘misleading’ promotional video of ‘Gemini,’ admits fabrication
‘AI boyfriends on the rise’: Young Chinese women embrace virtual companionship
Peanuts welcomes its first black character in historic debut
Available first in the PH: Brilliant visuals with 14th gen intel powered ROG strix laptops