OpenAI Sora 2 With A TikTok-Like Social Feed Launches Globally, All Details Here

Just a day ago, we were talking about the leaks claiming OpenAI wants its own social feed, and today we have it. Yes, Sam Altman's OpenAI has launched the brand new text-to-video model Sora 2 clubbed with

a new social media app. Sora 2 will serve as the direct competitor to Google VEO 3. OpenAI said in an official blog post, 'Sora 2 can do things that are exceptionally difficult - and in some instances outright impossible - for prior video generation models.'According to OpenAI, Sora 2 is much improved in terms of keeping up with the laws of physics, which could be dubbed one of the most important aspect in any good video generation model. To understand with an example, if a basketball player missed a shot in Sora 1, the ball may spontaneously teleport to the hoop, but in the case of Sora 2, the ball will rebound off the backboard.Furthermore, the new model is also being hailed as a better one in terms of controllability as the ChatGPT maker says it can follow instructions consisting of multiple shots while accurately persisting the world state. All in all, it is being said that Sora 2 will excel at anime, cinematic, and realistic styles.

What Else About Sora 2?

Sora 2 will be able to create background soundscapes, sound effects, and speeches, that synchronize well with the vibe of the video. What's more intriguing is that Sora 2 has the capability of adding humans or any real object directly to a created video.Also Read: Sam Altman's OpenAI Brings New Feature 'Pulse' for ChatGPT, Here's How It Works OpenAI is also planning to bring a viral trend with Sora 2, dubbed 'upload yourself', along with a social media app named Sora on iOS. Furthermore, the users will also get features like create, remix, and more, along with a cameo feature that will let them enter any scene on the app. OpenAI's excitement and concern around the model could be seen clearly as the company said in a blog post, 'The model is far from perfect and makes plenty of mistakes, but it is validation that further scaling up neural networks on video data will bring us closer to simulating reality.'