Artist Wildcards for NoobAI (NAI-XL)
v2 update: I have sanitised the prompts, i.e. removed underscores and added \( and \) where parentheses are needed. They actually look a lot better now when generating. Thx to Daru_22 for pointing this out.
This wildcard is intended and recommended to be used with NoobAI (NAI-XL) only.
I was inspired by this comment thread and started to make wildcards for it. Hope you enjoy.
Data extracted from https://huggingface.co/datasets/Laxhar/noob-wiki/blob/main/danbooru_artist_webui.csv
How to use
Install dynamic-prompts if you haven't done so already: https://github.com/adieyal/sd-dynamic-prompts
Unzip the file into
*\webui\extensions\sd-dynamic-prompts\wildcards(The * is the location where you installed your webui in)Go to the Wildcard Manager tab. You should see your wildcard on the left-hand side. If you don't see it, press
Collection actionsand thenRefresh wildcards.Copy the trigger prompts on the right, then paste it into your prompt as shown below:
1girl/1boy/1other, <character>, <copyright>, artist:PASTE WILDCARD PROMPT HERE, <special>, <other>What's in the .zip file
NoobAI-danbooru-artists - 419,781 danbooru artists scraped from the dataset. I recommend using this one if you want to accomplish a more traditional style like the ones you see on the NoobAI-XL gallery.
NoobAI-e621-artists - 205,479 e621 artists scraped from the dataset. Has more western artists so you might get a western style.
NoobAI-combined-artists - A combination of the above, totaling up to 625,260 artists.
Description
Initial release
FAQ
Comments (5)
You need to sanitize these, remove the underscore and replace it with a space.
But more importantly the escape '\' before the ( ).
Right now I had a mess using this because the prompts were seeing stuff like asuna_(evangelion:1.1) basically.
It's a quick fix
Thanks for the heads up. I just released a v2 with your advice now.
gj thanks!
@Daru_22 Is there a specific reason to avoid underscores? I always prompt danbooru tags with underscores and it felt more like a subjective difference.
@pholtur depends on how the model is trained! But in their examples they don't use them, so I take it as the good practice