SDXL came out for public use last week and I spent most of the time testing and playing around with this new model locally. I then stress tested by GPU by running close to 4000 prompts back to back generating images that will try to emulate the Artist styles that SDXL would recognise. I did a similar study with Midjourney as well which you can see here, it was focused on version 4.0 model of Midjourney.
But running my own instance gives me a lot of flexibility and using Automatic1111 with SDXL made is quite easy to produce these images. The prompt was straight forward “art by <artist>” which would get the SDXL mode to emulate the style and creations of that artists.
Prompt: art by <artist> Negative prompt: blurry, soft, low quality Steps: 25, Sampler: Euler a, CFG scale: 7.5, Seed: <random>, Size: 1024x1024, Model hash: 31e35c80fc, Model: sd_xl_base_1.0
Warning: Some nudity may be found with some artist’s image
Below is the entire resulting gallery for your exploration or use the search box below to find the artist you are looking for:
Comments are the bottom of these images use the button to jump to the comments. Let me know if this was of any use to you, or if you want an artist to be added to the list.
Offline version of this Study
Many have asked over on other social platforms and here that, they would like to be able to get the images offline so they can explore at their leisure. So I have found a way to make this feasible as a content creator and website. You can now download the offline version of the entire catalog (1.3GB) at a very nominal fee.
NOTICE: The online gallery has been limited due to the sheer size of it which was causing my servers to go into meltdown. You can access the entire collection with all the images as offline download using the link above.
NOTICE: The online gallery has been limited due to the sheer size of it which was causing my servers to go into meltdown. You can access the entire collection with all the images as offline download using the link above.
If you'd like to support our site please consider buying us a Ko-fi, grab a product or subscribe. Need a faster GPU, get access to fastest GPUs for less than $1 per hour with RunPod.io
Thank you so much for the gallery. It would also be really cool if you could download it as an cheat-sheet offline version of the page! Thanks again!
Offline version of the entire collection is available at a nominal fee. It will help support the website!!
Thanks for the gallery! How did you end up choosing all these artists?
There is public list/spreadsheet shared by some AI artists. This was used to do an Artist study using Disco Diffusion model. It was not my work but I worked with the creators to give it a home on my site. https://weirdwonderfulai.art/resources/disco-diffusion-70-plus-artist-studies/
Wow….this guide is awesome! Just what I’ve been wishing for! I”m learning about so many new artists here. Yes, I think I’d like to buy an offline copy if it was searchable by style as well as artist. Also, how can I send you a tip?
Thank you for the kind words. That’s something I will work on for the future. You can support via this link: https://ko-fi.com/weirdwonderfulaiart
I am a bit confused. You offer two versions one for SDXL 1.0 and an older one (which seems even richer in Artists) for v.5.2. So are all artists in both offline versions also here listed on this website or are there more in those sets we can buy? Because I guess this site is updated too but what about the ones we can buy? are the updated too?
v5.2 is using Midjourney model. So two different representations of models both commonly used in image generation. Depending upon which one you prefer (or both) you can download the complete list online.
Midjourney one is a preview as 4000 image gallery does put some stress on the site. You can buy either you prefer or both.
Question for you: How did you decide if the artist name had an impact? If the seed was random for each generation, how would you know how much influence was form the artist name versus a random generation from the program?
When the artist is not known you generally get some sort of grey image with pencil like drawing. Also as I stated, I don’t make reference of each image to confirm if it matches. I simply provide a resource which has all the artist names I could source, it is up to the viewer to decide if they agree with what SDXL infers in the resulting image.
Awesome, thank you for taking the time to reply and clarify 🙂