Stability AI Releases Stable Video Diffusion

Stability AI Releases Stable Video Diffusion
Image Credit: Stability AI

Stability AI has unveiled its newest generative AI creation, Stable Video Diffusion, which the company states represents a major leap forward in synthesizing realistic videos from text prompts. Released in an initial research preview, this new video generation model builds on the capabilities of Stable AI's widely used Stable Diffusion image creation system.

Stability is open-sourcing Stable Video Diffusion and an image-to-video model, for research purposes:

  • SVD: This model was trained to generate 14 frames at resolution 576x1024 given a context frame of the same size.
  • SVD-XT: Same architecture as SVD but finetuned for 25 frame generation.

Boasting the ability to produce high-quality videos up to 25 frames from text inputs, Stable Video Diffusion expands the horizons of what is possible with leading-edge generative adversarial networks. Through finetuning, the model can also generate multi-view videos from a single input image.

Stable Video Diffusion is capable of generating 14 and 25 frames at customizable frame rates between 3 and 30 frames per second.

In its current form, the system surpasses comparable proprietary models from competitors in user testing. This positions Stable Video Diffusion as a top-tier offering in a rapidly accelerating field.

While not yet ready for real-world deployment, Stable AI sees the research preview as a starting point for refining this technology responsibly, with public input playing a key role. A waitlist is open for those eager to experiment with video generation through a forthcoming web interface.

Stable Video Diffusion supplements Stable AI's extensive selection of open-source AI models spanning modalities from images to 3D shapes. It represents the company's enduring commitment to democratizing access to artificial intelligence's most transformational capabilities.

With this launch, Stable AI continues staking its ground as an innovative force in synthetic media. If initial reception mirrors that of Stable Diffusion, Stable Video One day, this technology may soon become an indispensable tool for creative professionals and AI enthusiasts alike. Yet, the company cautions, much work remains to ensure this technology matures safely before reaching mainstream adoption.

Chris McKay is the founder and chief editor of Maginative. His thought leadership in AI literacy and strategic AI adoption has been recognized by top academic institutions, media, and global brands.

Let’s stay in touch. Get the latest AI news from Maginative in your inbox.