The Next Version of Stable Diffusion Will Generate More Realistic Hands

The next version of the Stable Diffusion text-based AI image generator is capable of producing more photorealistic images with believable hands. SDXL 0.9 “significantly improves image detail and composition over its predecessor”.

The developers describe the new model as “a leap forward in the creative use of generative AI.” The samples included in the blog post have shown progress using the same hints in Stable Diffusion XL beta and SDXL 0.9.

Hands used to be an easy way to distinguish AI-generated images (at least until the launch of Midjourney v5), but progress has moved on since then, so problems with brushes are not as noticeable.

“Given the ability to work on a typical home computer, SDXL 0.9 opens up new horizons in the professional creative application of generative images. The ability to create hyper-realistic images for film, TV, music and video, as well as applications in design and industry, puts SDXL at the forefront of the practical use of AI imaging.”