Stability AI Introduces Video-Generation Tool 'Stable Video Diffusion'

Stability AI has unveiled its latest foray into the world of artificial intelligence with the introduction of "Stable Video Diffusion," an AI model designed to generate videos by animating existing images.

Building upon the success of Stability's previous text-to-image model, "Stable Diffusion," this new venture marks a significant milestone in the company's mission to democratize generative AI video models.

"Now available in research preview, this state-of-the-art generative AI video model represents a significant step in our journey toward creating models for everyone of every type," the company announced in a press statement.

Stable Video Diffusion

Stability AI has described the Stable Video Diffusion as a state-of-the-art generative AI video model. This model is a natural extension of the existing Stable Diffusion text-to-image model, demonstrating the company's commitment to expanding its capabilities across various modalities.

As part of the research preview, Stability AI has made the code for Stable Video Diffusion openly available on its GitHub repository. Interested users can access the model's weights required for local execution on Stability AI's Hugging Face page.

The company has also provided comprehensive technical details about the model's capabilities in a dedicated research paper. Stable Video Diffusion is presented in the form of two image-to-video models, each capable of generating 14 and 25 frames.

What sets these models apart is the ability to customize frame rates between 3 and 30 frames per second, providing users with flexibility in their creative endeavors.

During the foundational release, external evaluations indicated that these models outperformed leading closed models in user preference studies, according to Stability AI.

While Stability AI is excited about the potential applications of Stable Video Diffusion, the company emphasizes that, at this stage, the model is not intended for real-world or commercial applications. Instead, Stability AI actively seeks user feedback to refine and enhance the model's safety and quality aspects.

Stable 3D

Stability AI's exploration of generative AI models doesn't stop with Stable Video Diffusion. The company has ventured into 3D model creation by introducing "Stable 3D," which was announced earlier this month.

"Stability AI is pleased to introduce a private preview of Stable 3D, an automatic process to generate concept-quality textured 3D objects that eliminates much of that complexity and allows a non-expert to generate a draft-quality 3D model in minutes, by selecting an image or illustration, or writing a text prompt," the company said in a blog post.

This new tool, currently available in private preview, aims to simplify the complex process of generating concept-quality textured 3D objects. Targeting graphic designers, digital artists, and game developers, Stable 3D allows users to create draft-quality 3D models swiftly by selecting an image or providing a text prompt.