Stability AI company has introduced Stable Diffusion artificial intelligence model version 3.5 for converting text to image. This new release includes several different models that run on common hardware and are available for commercial and non-commercial use under the flexible Stability AI Community license.
Developers can now customize and integrate these models without worrying about restrictive permissions. In fact, this makes them ideal for a wide range of applications. Large and Large Turbo versions are available for download on the Hugging Face website.
What’s new?
Version 3.5 of Stable Diffusion is a set of artificial intelligence models for various users, including researchers, startups, and companies, that do the work of generating or manipulating images.
The Stable Diffusion 3.5 Large model has 8 billion parameters and has excellent image quality and high accuracy in responding to prompts. This model is well suited for professional use at 1 megapixel resolution. Also, the Turbo version of this model is faster and can produce high quality images in only 4 steps.
The company has stated that the model has been optimized for optimal performance on standard consumer hardware, particularly in the Medium and Turbo versions. Also, these models produce diverse and inclusive images that well show all kinds of skin colors and different characteristics of people without the need for long and complicated explanations.
The models were trained on a subset of the LAION-5b dataset created by the DeepFloyd team. This dataset uses NSFW filter to remove inappropriate content.
This model is freely available for non-commercial use, including academic research. Startups, small businesses, medium businesses, and content creators can use this model for free for commercial purposes, as long as their annual revenue is less than $1 million. Also, users fully own the content they produce and there are no license restrictions.
Google stopped producing images with Gemini
Meanwhile, Google has announced that it has stopped the artificial intelligence image generation feature of Gemini, because this feature is not accurate enough in generating historical images.
The images produced with Gemini have been gaining attention on social media recently, causing widespread laughter and protests. Some users strongly criticized Google, saying that the company cares too much about social issues instead of focusing on truth and accuracy.
“We recently decided to stop generating images of people with Gemini to work on improving the accuracy of its responses,” Google announced.
Social media users have criticized Gemini’s artificial intelligence for producing images of historical figures, such as the founding fathers of the United States, as people of a different skin color.
RCO NEWS