Stability AI, renowned for its Stable Diffusion text-to-image generator, has introduced its latest innovation – the Stable Video Diffusion (SVD) model. Now accessible via the developer platform and API, third-party developers can seamlessly integrate this image-to-video model into their applications, websites, and services.
In a blog post, Stability AI stated, "This new addition provides programmatic access to the state-of-the-art video model designed for various sectors…Our aim with this release is to provide developers with an efficient way to seamlessly integrate advanced video generation into their products."
Despite the potential benefits for enterprises seeking AI-generated videos, concerns arise due to Stability AI's utilization of the LAION-5B dataset. This dataset, taken offline recently, was found to contain instances of child sexual abuse material.
For those interested in generative video content, Stability's SVD API plug-ins offer a quality option. Stability AI claims, "2 seconds of video, comprising of 25 generated frames and 24 frames of FILM interpolation, within an average time of 41 seconds." While not suitable for major video campaigns, it is practical for creating GIFs with specific messaging, including memes.
Competing with models from Runway and Pika Labs, Stability sets itself apart by offering its video generation model through an API. Runway and Pika Labs require users to visit their websites or apps directly, limiting external developers from building apps or incorporating them.
Although Stability plans to launch a user-facing web experience for its video generator, no release date has been provided. Interested users are encouraged to join the waitlist.
Understanding Stable Video Diffusion
Released in research preview a month ago, Stable Video Diffusion enables users to generate MP4 videos from still images, such as JPGs and PNGs. The model, while in its early stages, can produce short videos lasting up to two seconds, suitable for sectors like advertising, marketing, TV, film, and gaming.
Stability claims versatility by offering multiple layouts and resolutions, including 1024×576, 768×768, and 576×1024. Additional features such as motion strength control and seed-based control provide developers with options for repeatable or random generation.
Controversy Surrounding Stability AI
Despite the controversy surrounding Stability AI's use of the LAION-5B dataset, the company continues to expand its offerings. A Stanford Internet Observatory report revealed instances of child sexual abuse material in the dataset, prompting its removal. Earlier this year, Stability AI faced a class-action lawsuit alleging the unauthorized acquisition of copyrighted images to create Stable Diffusion.
Currently, Stability's developer platform API grants access to all company models, from the text-to-image generator to the new SVD model. The company also offers a membership allowing customers to host the models locally.
