Stability AI's new AI model can turn 2D images into 3D videos

Stability AI has released a Stable Video Diffusion-based generative AI model called Stable Video 3D (SV3D) to simplify the creation of 3D videos. The SV3D has two components that help users generate 3D videos from 2D images: SV3D-u and SV3D-p.

“Today we are releasing Stable Video 3D (SV3D), a generative model based on Stable Video Diffusion, advancing the field of 3D technology and delivering greatly improved quality and view-consistency,” the company said in a blog post.

As highlighted by Stability AI, SV3D-u generates orbital videos with static images as input with the viewpoint remaining unchanged. On the other hand, SV3D-p will help users generate 3D videos with camera motion. And it can do so using both single images and orbital views. The company, however, isn’t new to 3D object generation. It introduced Stable Zero 123 last December for that specific purpose, but the latest SV3D is more powerful and can outperform other open-source alternatives like Zero123-XL.

“By adapting our Stable Video Diffusion image-to-video diffusion model with the addition of camera path conditioning, Stable Video 3D is able to generate multi-view videos of an object. The use of video diffusion models, in contrast to image diffusion models as used in Stable Zero123, provides major benefits in generalization and view-consistency of generated outputs,” explained the company.

Stability AI’s Stable Video 3D is available for commercial usage to those who have Stability AI Membership, which starts at $20 a month. For non-commercial usage, the model weights can be downloaded on Hugging Face.