Stable Video Diffusion is the latest AI video generator. The tool creates a video from existing images provided by the user. AI video creation is a field in which there has been more development recently, which results in various offerings which are most of the time more impressive.
The publisher behind Stable Video Diffusion is Stability AI. This provider of open-source AI models already released the image generator Stable Diffusion earlier. The video generator is a follow-up to this model and is meant for creating videos. Stability AI says in the announcement it is now releasing the research preview. The code behind the AI video generator is, therefore, released on the company’s GitHub page.
Stable Video Diffusion is actually an umbrella term for two models. SVD is the general model that can generate fourteen images for video. SVD-XT is the more advanced model that extends the length of the video by ten frames. The frame rate, or the number of frames shown per second, ranges from three to 30 frames.
Meta experiments with text-to-video
The field of AI video generators has seen more activity lately. Meta released its own variant called Emu Video less than a week ago. The social media company is already conducting research on creating videos from text prompts. A combination of text and images is also possible. However, Meta has been experimenting with these capabilities since 2022 but can now create more realistic videos.