Bartholomew Gallacher
Well-known member
- Joined
- Sep 26, 2018
- Messages
- 5,765

- SL Rez
- 2002
The SDXL 1.0 model was released this week, and is now available for download. This is the newest, latest and greatest official model so far for Stable Diffusion.
This model works differently than previous ones: it consists of a base and a refiner model. The model is now trained on 1024x1024 image sizes, meaning when using less the image quality might suffer.
The work flow is now like this: first create an image with the base model; then use that base image with the refining model, which will sharpen it and add more details. So this model now delivers better image quality, and also according to its creators has better quality on hands now. So really finishing an image how it is intended to be now consists of two steps, if you are content enough with the base image.
The model can be used with Automatic1111 quite easily. Here's how to use it.
stable-diffusion-art.com
And now for something completely different: the annoying orange (no SDXL1.0 used here).
This model works differently than previous ones: it consists of a base and a refiner model. The model is now trained on 1024x1024 image sizes, meaning when using less the image quality might suffer.
The work flow is now like this: first create an image with the base model; then use that base image with the refining model, which will sharpen it and add more details. So this model now delivers better image quality, and also according to its creators has better quality on hands now. So really finishing an image how it is intended to be now consists of two steps, if you are content enough with the base image.
The model can be used with Automatic1111 quite easily. Here's how to use it.

Stable Diffusion XL 1.0 model - Stable Diffusion Art
Stable Diffusion XL (SDXL) is the latest AI image model that can generate realistic people, legible text, and diverse art styles with excellent image

And now for something completely different: the annoying orange (no SDXL1.0 used here).