The ‘text prompt to image generating’ AI Stable Diffusion 2.0 has released. 1.4 was the first public release back in summer 2022, and 1.5 was the previous version.
Along other new features, a shape-sensing and shape-preserving model…
It also has a few other nice features such as better in-painting and built in up-scaling. But note that the new 2.0 version is now censored, so some may consider it as effectively no longer open-source. They’ve “removed ability to copy artist styles”. This means 2.0 now ignores popular user prompts such as “in the style of [name]…”. It also refuses to create what is described as “NSFW” images, though how widely that’s defined is unknown.