Stability AI launches stable video 3D model for 3D video rendering

Stability AI released a new 3D video rendering model called Stable Video 3D (SV3D) on Monday. The artificial intelligence (AI) model can generate videos, but unlike popular video generators like OpenAI’s Sora, Runway AI, and Pika 1.0, it doesn’t accept text input. The main focus of SV3D is to take image input and convert 2D photos into orbital 3D models. The company has made its new AI model available for commercial and non-commercial use.

The news was released by Stability AI’s official account on X (formerly Twitter) postal It reads: “Today, we released Stable Video 3D, a generative model based on Stable Video Diffusion. This new model advances the field of 3D technology, greatly improving quality and multiple views.” The announcement was released Just one month later, Artificial Intelligence Inc. Announce Stable Diffusion 3, which improves performance in multi-topic prompts.

The stabilized video 3D AI model comes in two different variants – SV3D_u and SV3D_p. The former is able to generate orbital video from a single image input but does not use camera adjustments. This means that although the objects in the 2D image will be converted to a 3D rendering, there will not be any camera movement. A more powerful variant is SV3D_p, which can accommodate single images and orbit views, which will also allow it to create fully rendered 3D videos along a specified camera path.

According to the company, the AI ​​model solves the inconsistency issues faced by older generation models such as Stable Zero123. SV3D leverages Neural Radiation Fields (NeRF) and mesh representations to improve the quality and consistency of rendered videos. “In addition, to reduce baked lighting issues, Stable Video 3D uses a disentangled lighting model that is co-optimized with 3D shapes and textures,” Stability AI said in a detailed report. blog post.

Stable Video 3D is now available for commercial and non-commercial use. For commercial use, users need to become a Stability AI member, with the professional level starting at $20 per month (approximately Rs. 1,650).For non-commercial use, users can download Model weight On the hugging face.