Stable Video Diffusion
Stable Video Diffusion is Stability's first open-weights image-to-video model (14 and 25 frames), released as a research preview.
Extends Stable Diffusion with temporal layers trained on a curated video dataset; produces a few seconds of motion from a still. Opened video research to the community months before closed models like Sora, but quality and length lagged.
- Date
- Tuesday, 21 November 2023
- Lab
- Stability AI
- Kind
- model
- Access
- open weights (restricted license)
Date from page metadata. Frame counts (14/25) and non-commercial research license are from memory; verify.
Sources
This record was checked against its sources on 6 October 2026. How we check