Stable Video 3D

The Stability AI License offers flexibility for your generative AI needs by combining our advanced open-source models with the advantages of self-hosting. SV3D takes a single object image as input and outputs novel multi-views of that object. We can then use these new views and SV3D to generate a 3D mesh. SV3D comes in two variants tailored to different needs. SV3D_u focuses on generating orbital videos from a single image input without camera conditioning, making it ideal for simple 3D visualization. SV3D_p, on the other hand, extends this capability to include both single images and orbital views, allowing the creation of more complex 3D videos along a specified camera path. SV3D is available for commercial use with a Stability AI membership and has a wide range of applications across various industries. For non-commercial use, the model weights can be downloaded from the Hugging Face platform and are accompanied by our comprehensive research paper.

To use Stability AI models such as SV3D, you can obtain a commercial license by getting a Stability AI membership, or register to download the model weights from a platform like Hugging Face for non-commercial use. Please follow the usage instructions provided in the GitHub repository and technical report for more details.

Flexible licensing options for generative AI models
Self-hosting advantages
Access advanced open-source models like SV3D.
Commercial and non-commercial use options
Novel View Synthesis and 3D Generation from a Single Image (SV3D)
Two SV3D variants (SV3D_u and SV3D_p) serve different needs.