Stable Diffusion, an AI software that creates stunningly realistic images from text, has just released an update that includes new features such as an upscaler, depth maps, and more. However, users are unhappy that the new software can no longer generate NSFW artworks or copy the styles of certain artists, according to The Verge.
One of the most significant additions to Stable Diffusion Version 2 is the new text encoder called OpenCLIP, which improves the quality of the generated images. The update also includes a new NSFW filter from LAION that can remove adult content. Meanwhile, the depth-to-image diffusion model allows users to create new versions of an image while preserving the coherence and depth from the original image.
The update makes it harder to create certain types of images like photorealistic images of celebrities, nude and pornographic output, and images that match the style of certain artists. Users have complained that they can no longer generate images in the style of some artists, who is often copied for AI images.
Stable Diffusion is an open-source AI that allows users to build upon it, making it popular among AI art enthusiasts. The software can copy the styles of certain artists by training on their works. However, the license agreement prohibits users from using the model in any way that breaks the law.
Despite some objections from artists, Stable Diffusion remains a popular choice for generating AI art. The software’s versatility enables users to produce unique and compelling images that reflect their creativity.
