Home » Stable Diffusion update removes ability to copy artist styles or make NSFW works

Stable Diffusion update removes ability to copy artist styles or make NSFW works

by admin

Stable Diffusion is an AI that can generate images from text. amazingly real way, being Updated with lots of new featuresHowever, many users have complained that the new software is no longer able to generate photos in a particular artist’s style or generate NSFW artwork. The Barge Reported.

Version 2 introduces many new features. Key among them is a new text encoder called OpenCLIP, which according to Stability AI, the company behind Stable Diffusion, “has significantly improved the quality of the images it produces compared to his previous V1 release. It will improve.” Also included is LAION’s new NSFW filter designed to remove adult content.

Other features include a depth-to-image diffusion model that can create transformations that “look radically different from the original image, but maintain image consistency and depth,” according to Stability AI. increase. This means that objects will appear correctly in front of or behind other objects, even if you create a new version of the image. Finally, the text-guided inpainting model makes it easy to switch parts of the image. For example, you can change the body while keeping the cat’s face.

Stability AI

However, this update made it more difficult to create certain types of images, such as photorealistic images of celebrities, nude and pornographic output, and images that match the style of a particular artist. said he commissioned Stable Diffusion Version 2 to generate images in the style of artist Greg Rutkowski. Often copied for AI images — no longer works like it used to. ‘They undermined the model,’ said one Reddit user.

Competitors like DALL-E are closed models, whereas Stable Diffusion is open source and can be built on top of it, making it particularly popular for AI art generation.For example, the YouTube VFX site corridor crew Showed off an addon called dream booth This allowed them to generate images based on their own personal photos.

Stable Diffusion can be trained on the work of artists like Rutkowski and copy them by examining images and looking for patterns. It’s probably legal to do this (albeit a gray area). explainer Earlier this year. However, the Stable Diffusion license agreement prohibits using the model in any way that violates the law.

Nonetheless, Rutkowski and other artists oppose its use. “I probably won’t be able to find a job for myself, because [the internet] There will be a flood of AI art,” Rutkowski said. MIT Technology Review. “That bothers me.”

All products recommended by Engadget are selected by an editorial team independent of the parent company. Some stories contain affiliate links. When you purchase something through one of these links, we may earn an affiliate commission. All prices are correct at the time of publication.

You may also like

Leave a Comment