Stability AI Unveils Next-Gen Stable Diffusion 3 for Faster, Higher-Quality AI Art Generation
-
Stability AI has announced Stable Diffusion 3 (SD3), its next-gen text-to-image model, touting better text generation, prompt adherence, and lack of prompt leaking.
-
Early comparisons show SD3 outperforming models like MidJourney, Dall-E 3, Google's ImageFX in image quality and prompt following.
-
SD3 uses a new diffusion transformer architecture and flow matching for faster, more stable training than GANs.
-
It can scale further and accept multimodal inputs, paving the way for video, 3D, etc.
-
SD3 will be free, open source, and available to all under a non-commercial license, though access currently requires membership.