作者:Dade Hayes
Silicon Valley startup Luma AI has launched Ray3, which it says is the first video model with the ability to reason.
Adobe and major ad firms and creative agencies Monks UK, Galeria, and Strawberry Frog are among the initial customers of Ray3, which is available on the Dream Machine platform.
The generative model is capable of critiquing itself and making on-screen notes about revisions and improvements. Production work can also be saved and picked up where it was stopped, unlike many current models. A “draft mode” also allows for more trial and error at a faster rate, with preferred results then rendered in 4K high dynamic range.
The output of a system endowed with the power to reason, Luma says, is “videos that feel more coherent, with characters that look consistent, scenes that unfold naturally over time, and physics that behave as they should.”
Luma also says Ray3 is the first model to produce video in true 10-, 12-, and 16-bit HDR ACES2065-1 EXR format, making it suitable for high-end film and advertising projects.
Luma, whose backers include Nvidia and venture capital firm Andreessen Horowitz, earlier this year opened an L.A. studio and it is also establishing a presence in New York.
“Ray3 is our first step toward building intelligence for creative work,” said Amit Jain, CEO and co-founder of Luma AI. “Creative work is one of the most intellectually challenging things humans do, yet until now, many of the AI available to creatives has lagged far behind what’s possible in coding and analysis with language models. Generative models have been more like slot machines – powerful but not intelligent. Ray3 changes that in a big way.”
Adobe and Luma AI are partnering to bring Ray3 directly into the Adobe Firefly app.