With roughly a year after ChatGPT has started the trend with what AI can do, OpenAI pushes AI generation boundaries further with Sora, which their new model that can create AI-generated videos of up to a minute long. Compared to AI-generated images that the likes of Gemini (formerly Bard) can do, the idea of AI-generated videos is a very challenging thing to do given all the complexities that’s associated with any moving image.
READ MORE:
Announcing Sora — our model which creates minute-long videos from a text prompt: https://t.co/SZ3OxPnxwz pic.twitter.com/0kzXTqK9bG
— Greg Brockman (@gdb) February 15, 2024
As Marques Brownlee pointed out, ChatGPT’s implementation of Sora in generating videos is miles better from earlier iterations of AI-generated videos. While one can still spot imperfections like certain details in the AI-generated videos, Sora manages to create usable videos–all through the text prompts made. Slight imperfections aside, the AI-generated videos that OpenAI presented using Sora are good enough to be used as stock footage.
https://t.co/rmk9zI0oqO pic.twitter.com/WanFKOzdIw
— Sam Altman (@sama) February 15, 2024
As much as Sora can blow everyone’s minds with how refined is ChatGPT’s approach, it is still all hype for now: you can’t access Sora and generate videos right away, as it’s limited to a number of people–which OpenAI refers to as “red teamers”. It makes sense to limit access to Sora, as AI-generated videos are much scarier than AI-generated images especially when they fall into the wrong hands. This also puts companies selling stock footage in danger as well, with some of the sample footage OpenAI presented being convincing enough to pass as one if not for the watermark at the bottom right corner.
