Stable Video Diffusion

What the release contained

The November 2023 announcement releases Stable Video Diffusion code and weights and distinguishes image-to-video models generating 14 and 25 frames. A waiting list for a future text-to-video web experience is separate from the research release. The presence of that list does not establish a generally available text-to-video service at that date.

Section sources: Stable Video Diffusion

From image to movement

A still image becomes the starting condition for a short movement, establishing a different form of control from frame-by-frame animation. Historical records should separate research weights, web services and later fine-tuned applications. Download availability does not itself establish permission for every commercial use.

Section sources: Stable Video Diffusion

Research paper and release dates

First submitted on 25 November 2023, the paper distinguishes image pretraining, video pretraining and high-quality video fine-tuning, including captioning and filtering. Its research configurations should not all be attributed to one initially released checkpoint; the existing release chronology is retained.

Section sources: Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets

Sources for this account

This account expands the archive within the cited scope. Historical interpretation is not presented as the original author’s statement or as a proven causal influence. The account and translation are AI-assisted, pending independent human review. Section references identify evidence without claiming independent verification of every historical statement.

Continue with a comparative question