Stability AI's video diffusion models (SVD) generate short video clips from still images using latent diffusion techniques, producing consistent and high-quality motion sequences. The models are open-weight and available for research and commercial use via the Stability AI platform.