Luma Labs has launched a brand new video generative mannequin known as Ray2. It claims to create life like movies with pure and coherent movement. Ray2 is at present accessible on the Dream Machine platform for paid subscribers.
“We’re beginning with text-to-video. Picture-to-video, video-to-video, and enhancing capabilities will quickly comply with,” mentioned the corporate in a put up on X. It may well generate movies with a length of 5 or ten seconds in a number of facet ratios. The paid plans for Dream Machine begin at £6.99 a month and allow customers to create 1080p movies.
“At present, we’re introducing Ray2! Scaling pretraining by 10x (compared to Ray1) on a novel environment friendly structure, we’re capable of unlock the subsequent frontier in video era—quick pure coherent movement and physics,” mentioned Amit Jain, co-founder of Luma Labs, in a put up on X.
“This skyrockets the success fee of usable, production-ready generations and makes video storytelling accessible to many extra individuals,” he added.
Jain additionally revealed that the corporate is contemplating releasing an API for it quickly and that it’s ‘important to our targets’.
The corporate additionally launched an extended listing of examples of movies generated throughout themes like pure movement, physics and simulation, photorealism, cinematic scenes, individuals and expressions.
Ray 2 produces AI influencers on demand — full with broad angle vlogging and rolling shutter
pic.twitter.com/uUvgaX944u
— Bilawal Sidhu (@bilawalsidhu) January 15, 2025
That mentioned, customers did run into a couple of points accessing the mannequin on launch day. One person on X complained that the servers have been overwhelmed and that that they had been ready for half-hour for the outcome.
In the meantime, Jain famous that it was the ‘launch day rush’ and anticipated that issues would stabilise quickly.
Customers shortly examined the mannequin. MattVidPro, an AI-focused tech YouTuber, carried out a ballot on X, asking testers which mannequin is extra spectacular: Ray 2, Google’s Veo 2, or Vidu 2 from Vidu AI. The outcomes revealed that over 80% picked Google’s Veo 2 mannequin.
A person on X, Ben Nash, expressed that Veo is considerably forward of its opponents and believes it’s of superior high quality in comparison with Sora, although Sora affords a better array of options. Nevertheless, in contrast to OpenAI’s Sora and Ray 2, Google’s Veo 2 isn’t but accessible for public use.
Google’s inside testing signifies that Veo outperforms opponents (similar to China’s Kling, Meta’s Moviegen, and OpenAI’s Sora) each when it comes to high quality and immediate adherence.
The put up Luma AI Proclaims Ray2 AI Video Mannequin Pre-trained with ‘10x Compute’ appeared first on Analytics India Journal.