Stability AI Announces 'Stable Video 4D,' an AI Model That Generates Eight Different Angles and Views from a Single Video
Stability AI has announced ' Stable Video 4D ,' an AI model for generating multi-angle video.
Stable Video 4D: The latest AI model for dynamic multi-angle video generation — Stability AI Japan
https://ja.stability.ai/blog/stable-video-4d
SV4D: Dynamic 3D Content Generation with Multi-Frame and Multi-View Consistency
https://sv4d.github.io/
Stable Video 4D is Stability AI's first video-to-video generation model, which can generate videos of different angles and views from a video. Users can upload one video and output eight different angles. Users can specify the camera angle of the generated video and can also tailor the output to suit their specific creative needs.
We are pleased to announce the availability of Stable Video 4D, our very first video-to-video generation model that allows users to upload a single video and receive dynamic novel-view videos of eight new angles, delivering a new level of versatility and creativity.
— Stability AI (@StabilityAI) July 24, 2024
In… pic.twitter.com/1YbI2W514K
Stable Video 4D is an AI model developed based on Stable Video Diffusion and Stable Video 3D , which are AI models that can output videos from images. It generates videos from different angles by generating multiple new view videos (4D image matrix) of objects appearing in the input video.
Users simply upload a video and specify the desired 3D camera pose. Stable Video 4D will then generate eight new viewpoint videos according to the specified camera viewpoints, providing a comprehensive, multi-angle view of the subject. The generated videos can be used by Stability AI to 'efficiently optimize the dynamic 3D representation of the subject in the video,' he explained.
Stable Video 4D can generate five frames of video across eight viewpoints in about 40 seconds, and the entire 4D optimization takes about 20 to 25 minutes. Regarding the uses of Stable Video 4D, Stability AI wrote, 'We envision future applications of Stable Video 4D in fields such as game development, video editing, and VR content generation. Experts in these fields will greatly benefit from the ability to visualize objects from multiple viewpoints, enhancing the realism and immersion of their products.'
The training dataset for Stable Video 4D uses Objaverse , which is distributed under the Open Data Commons Attribution License .
Stable Video 4D is available through Hugging Face, however, individuals or organizations with annual revenues exceeding $1 million must obtain an enterprise commercial license directly from Stability AI before using Stable Video 4D commercially.
stabilityai/sv4d · Hugging Face
https://huggingface.co/stabilityai/sv4d
Related Posts: