/vt/ - Virtual Youtubers

I'm running a big fat S/R on background tags I've never tried on a grugschizo prompt. So far, at 1.2 strength right after best quality:
>simple background
Looks like your generic mall photo studio vignette
>extremely detailed background
It drew a fucking black background
>zoom layer
She confused but she got the spirit. AI probably lacks context about what else it's drawing, so it just incorporated random shit that was in the prompt in the background, "graphic design is my passion" style
>heaven condition
Doesn't look like the danbooru tag implied. It drew some nice clouds and made the image brighter
>halftone background
It drew a simpler background than "simple background" - no vignette. No halftone pattern either
>sunburst background
Kinda orange faded, but not a sun/starburst pattern. Looks neat, and it added the stars (?) you see in La+ art
>abstract background
Swirly aura energy, looks cool
>"blending, no lineart, flat color"
It surprisingly worked? Half of her hair is black, and her entire coat became black and blends into the background
>drama layer
Slight cloudy outline around her - not at all as implied.
>fiery background
Chuuni
>imageboard colors
Black background, but God damn it drew some cool shit on her legs. And a lone flame behind her
>projected inset
fuckin MASSIVE horns

Unfortunately, I did not tick "Show Legend" - and I fucking hate that x/y/z grid just makes...a single row. So I made my own grid. Enjoy?

>>44681942
Holy shit, I am spoiled today. Thank you.

>>44681956
Like I said, you may find using the "fixed" one better, but you'll likely find it worse, since part of its "look" may be in how it behaves because its "broken".