I generated this artist study to document what artist styles were trained in SDXL and which ones were capable of being reproduced. The images were generated with the following example parameters:

Parameter Value
Positive Prompt ‘art by {artist}’
Negative prompt ‘ugly, poor quality’
Image size 1024x1024
Steps (base model) 30
Steps (refiner model) 15
Seed 1
Hardware RTX 3090
Script Official Stable Diffusion dual model pipeline

If you would like to contribute, that’s awesome! I have set up a Google Sheet to add anything these studies are missing. Please use this link and I will be adding updates as soon as I’m able to.

SDXL Community Submissions

Special shoutout to @nerdworldorder for providing a ton of new artists!