15 hours in the past
Stability AI first launched Stable Video in November 2023, releasing the mannequin codes on GitHub and weights on HuggingFace for customers to obtain and run regionally on succesful {hardware}.
But what about everybody else who didnât have a strong GPU or the technical expertise to set all that up? Well, Stability AI has lastly launched their Stable Video web site this week, so now anybody can mess around with making AI movies.
All you’ll want to get began is a Google account and an online browser.
What is Stable Video?
Stable Video Diffusion is a strong instrument designed for a variety of video functions throughout media, leisure, training, and advertising and marketing. It permits customers to show textual content and picture inputs into vivid scenes, reworking concepts into cinematic experiences.
Stable Video Diffusion is launched within the type of two image-to-video fashions, able to producing 14 and 25 frames at customizable body charges between 3 and 30 frames per second.
This is what itâs able to:
- Video length: 2â5 seconds
- Frame fee: as much as 30 FPS (frames per second)
- Processing time: 2 minutes or much less
How does it work?
Stable Video Diffusion depends on a posh course of using diffusion fashions (DMs), classifier-free steerage, and a base mannequin structure particularly designed for video technology.
If you need to be taught extra about the way it works, try this whitepaper.
Example movies
The instance movies beneath have been generated by the neighborhood and showcased on the Stable Video web site.
Prompt: aurora borealis
Prompt: african elephant
