

The model also does a better job of generating complex objects, such as people and animals. The images are sharper and more realistic, and they have better color and lighting. SDXL 0.9 produces images with significantly improved detail and composition. (Left - SDXL Beta, Right - SDXL 0.9) Credit: Stability AI What is new with SDXL 0.9?

The second stage model of the pipeline is used to add finer details to the generated output of the first stage," the blog post said. "SDXL 0.9 has one of the largest parameter counts of any open source image model, boasting a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline (the final output is created by running on two models and aggregating the results). Up to the March introduction of Midjourney v5, a competing platform based on Discord, hands were a simple "tell" to identify AI-generated art. The newest model appears to produce images with higher resolution and more lifelike hands, including aliens, wolves, and a human holding a coffee cup. The new model, according to Stability AI, offers "a leap in creative use cases for generative AI imagery." The blog post's example photos showed improvements when the same prompts were used with SDXL 0.9 and Stable Diffusion XL beta.
