The Stable Diffusion update removes the ability to copy artist styles or NSFW work

Stable Diffusion, the artificial intelligence that can create images from text in a stunningly realistic way, has been updated with a host of new features. However, many users are not happy, complaining that the new program can no longer create images in the styles of specific artists or create NSFW artwork, the edge It has been reported.

Version 2 introduces a number of new features. Key among these is a new text encoder called OpenCLIP that “dramatically improves the quality of generated images compared to previous V1 versions,” according to Stability AI, the company behind Stable Diffusion. Also includes a new NSFW filter from LAION designed to remove adult content.
Other features include a depth-to-image diffusion model that allows one to create transformations that “look radically different from the original image but still maintain coherence and depth from the image,” according to Stability AI. In other words, if you create a new copy of the image, the objects will still appear correctly in front of or behind other objects. Finally, the text-oriented interior drawing template makes it easy to swap parts of the image, keeping a cat’s face while changing its body, for example.

Stable deployment release 2
AI stability

However, modernization now makes it difficult to create certain types of images such as realistic images of celebrities, nude and pornographic output, and images that match the style of some artists. Users said that Stable Diffusion version 2’s prompt to create images in the style of Greg Rutkowski — an artist often copied for AI images — no longer works as well as it used to. “They emasculated the paradigm,” said one Reddit user.

Stable deployment has been particularly popular in creating AI art because it is open source and buildable, while competitors such as DALL-E are closed models. For example, YouTube VFX Corridor Crew It offered an add-on called Dreambooth that allowed them to create images based on their own selfies.

Stable Diffusion can copy artists like Rutkowski by rehearsing their work, examining images and looking for patterns. It’s probably legal to do so (albeit in a gray area), as we detailed in our explainer earlier this year. However, the Stable Diffusion license agreement prohibits people from using the form in a way that breaks any laws.

Despite this, Rutkowski and other artists objected to the use. “I probably won’t be able to find my job there because I… [the internet] Rutkowski said: MIT Technology Review. “This is worrying.”

All products recommended by Engadget are selected by our editorial team, independently of the parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission. All prices are correct at the time of publication.

Leave a Reply

%d bloggers like this: