Luma Labs released Dream Machine for AI-generated videos

Luma Labs has released its Dream Machine for AI-generated videos. Shortly after the announcement, many users rushed to try the new image-to-video model. Initially, no videos were produced due to the heavy traffic, causing all requests to be queued. Luma Labs made their video model available for free to all users without any waitlist. Free users are allowed 30 generations per month, with paid plans available afterwards.

The quality of their model aims to compete with next-gen video models like Sora and the recently released Chinese model, which currently requires a waitlist. Additionally, Pika Labs has also released an improved model for AI-generated videos. This tool is quite advanced, generating not just motion videos to animate pictures but also attempting to create entire video scenes. While it doesn't always succeed, making videos from uploaded pictures works much better than generating them from text prompts.

The UI is very simple, featuring sections for text prompts, image uploads, and viewing your creations below, arranged in two columns. Once you enter a prompt, the video generation queues up, and you can see the estimated time remaining. After the initial load, the team has made several fixes, so now video generation takes much less time—usually around 120 seconds, though it can sometimes take a bit longer. The minimal UI is sufficient for getting the job done.

I've observed that generating videos from JPEGs works better than using other formats or text, though it sometimes requires a couple of attempts to produce something meaningful. While the videos previewed during the Sora demo were superior, there are already many impressive examples from Luma Labs showcasing advanced quality. Let's see how it performs!