Dream Machine could level up your GIF game
(Image credit: Luma AI / Future)
The surge in text-to-video AI technology has gained significant momentum recently, yet eagerly awaited platforms like OpenAI Sora are still inaccessible for trial. If you’re itching to explore, a fresh contender named Dream Machine has just launched, offering immediate hands-on experience.
THE DETAILS
Dream Machine, developed by Luma AI, known for its iPhone app facilitating 3D photo captures, has now ventured into generative video. Its free tier allows usage with a Google account, though with some limitations. Paid plans start from $29.99 per month.
At present, Dream Machine appears to be grappling with high demand, as indicated on its website with notices stating that processing times for video generations may be up to 120 seconds due to queues. During testing, our text prompt took over 20 minutes to process, but the resulting outputs were notably impressive.
Dream Machine’s outputs are brief at five seconds each, delivered in 1360×752 resolution. Simply input your prompt into the interface, wait for processing, and download a watermarked clip directly to your account.
While it currently lags behind in length and resolution compared to advanced competitors like OpenAI’s Sora and Kling AI, Dream Machine provides a promising glimpse into the capabilities of such services. Despite longer wait times initially, our test prompt – ‘a close-up of a dog in sunglasses driving a car through Las Vegas at night’ – yielded a clip closely resembling our vision.
Previewing the Future of AI Videos
Like other generative AI video tools, questions linger about the training data utilized by Luma AI, potentially limiting its applications beyond personal use or enhancing GIF creation. Moreover, Dream Machine isn’t the pioneer in free text-to-video tools, with Runway’s Gen 2 model exiting beta just last year.
The Dream Machine website acknowledges its current technical limitations in handling text and motion, making trial and error a significant part of the user experience. However, as an introductory glimpse into the more advanced and likely pricier AI video generators on the horizon, it presents an enjoyable tool for experimentation.
This is especially pertinent given the existing alternatives, such as Google Veo, which have lengthy waitlists. Meanwhile, high-powered models like OpenAI’s Sora, capable of generating 60-second videos, are slated for release later this year, with Kling AI currently restricted to China.
As text-to-video technology continues to gain popularity, these dynamics are expected to evolve. In the meantime, Dream Machine offers a valuable platform for practice, albeit with patience required due to processing times.