It's time to celebrate the incredible women leading the way in AI. Nominate your incredible leader for VentureBeat's Women in AI Awards by June 18. Learn more here
The fast-changing market for AI video generation has changed again. Luma AI, a startup backed by Silicon Valley renowned venture firm Andreessen Horowitz, today announced the free public beta of its new AI video generation model, Dream Machine, and it has already seen a surge in users.
The model promises to produce up to 120 frames in 120 seconds (2 minutes, or one frame per second), but in reality, many users find themselves waiting for hours in the digital queue on Luma Dream Machine's website for their videos to be processed. The company says this is due to the sheer volume of traffic.
“Thank you everyone so much for your enthusiasm and support!” Luma's head of product and growth, Berkeley Dai, posted a message on the company's Discord channel today. “We are currently[ly] Demand is increasing and we are working to expand capacity. Not all generations will be lost, but they will remain in the queue. Once capacity is added, we will update the status here.
A few hours later, Dai provided this update:
VB Transform 2024 registration opens
Join enterprise leaders at our flagship AI event in San Francisco July 9-11. Network with your peers, explore the opportunities and challenges of generative AI, and learn how to integrate AI applications in your industry. Register now
“The queue has gradually shortened as our team has added capacity. In the short term, we expect it will still take several hours to fully process the current backlog of pending generations. Under normal circumstances, it takes only 2-3 minutes to convert a prompt into a video. Please be patient while we accommodate the influx of people wanting to try out this innovative tool. Generation speeds will continue to improve...
These will be generated as soon as possible. We thank you for your patience and support today and ask that you continue to return for the rest of this week and beyond as we continue to improve Dream Machine.“
The little-known startup, a high-quality AI video generator previously covered by VentureBeat when it released “Genie 1.0,” a model that generates 3D assets from text, in November 2023, had raised more than $70 million as of January 2024, $43 million of which came in a Series B, according to TechCrunch.
In a smart move from a PR perspective, the company has roped in some well-known AI video creators and filmmakers early on into Dream Machine, giving them the chance to test out its ability to generate video from text prompts and still images, submitting their work throughout the day ahead of today's public beta launch.
Even those who have just started using the tool find it very powerful, with some comparing it to OpenAI's Sora, while others say it's already superior.
In VentureBeat's limited testing of Luma's Dream Machine Web app, the text-to-video conversion feature showed only sporadic accuracy when it came to depicting what the prompts asked about, though after a few minutes it produced a video with very smooth, jitter-free movement and high-resolution, detailed assets.
Clearly, the race to create compelling AI video models has entered a new phase, with OpenAI's Sora, still only available to a select few users, currently facing very stiff competition, not to mention other AI video model providers Runway, Pika, and new Chinese competitor Kling.
Source link
