Insider Wales Sport: AI startup testing generative video

Stability AI, a well-known artificial intelligence (AI) company, has unveiled their latest product, Stable Video Diffusion, aiming to revolutionize the creation of animated videos. The new tool allows users to generate animated videos from a single image, making it easier than ever to bring graphics to life.

The Stable Video Diffusion tool is powered by two image-to-video models. Users can expect videos that range from 14 to 25 frames long, with speeds ranging from 3 to 30 frames per second. Stability AI claims that their generative AI video model surpasses leading closed models in user preference studies, putting it on par with industry frontrunners Runway and Pika Labs, both renowned in the text-to-video domain.

Despite its promising features, the tool does come with a few limitations. For instance, it can only generate short videos and falls short when it comes to perfect photorealism. Additionally, it struggles with camera motion, only being able to handle slow pans. Generating legible text or accurately rendering people and faces remains a challenge as well.

Currently, the Stable Video Diffusion tool is only available for research purposes. However, interested users can sign up for a waitlist to access an “upcoming web experience featuring a text-to-video interface.” Stability AI plans to demonstrate the potential applications of this innovative tool in sectors such as advertising, education, and entertainment.

The tool was trained on a vast dataset of millions of videos. While the company didn’t disclose the sources of their training data, they did state that publicly available video data was used solely for research purposes.

See also  AI talent raid poses challenges for regulators and VCs

Stability AI has faced legal challenges in the past, including a lawsuit filed by Getty Images for allegedly scraping its image archives. Additionally, their vice president of audio resigned due to concerns over the use of copyrighted content to train generative AI models. These challenges highlight the delicate balance between innovation and copyright protection in the realm of AI.

While generative AI has immense potential to simplify content creation, it also raises concerns regarding deepfakes and copyright violations. Stability AI has encountered difficulties in commercializing its Stable Diffusion product and has faced financial challenges as a result.

With the release of Stable Video Diffusion, Stability AI hopes to overcome these obstacles and establish itself as a leading player in the AI industry, empowering users to effortlessly create captivating animated videos from a single image.

You May Also Like

About the Author: Hanley Mallin

Internet geek. Wannabe bacon enthusiast. Web trailblazer. Music maven. Entrepreneur. Pop culture fan.

Leave a Reply

Your email address will not be published. Required fields are marked *