Which AI Art Prompts do you use to make Reference Pictures?

[gaylolis]

So lately I've been using AI Art to also generate reference pictures while I'm drawing.

Any ideas on what prompts are good so it can generate pictures from various positions and poses? NSFW ones are also welcome since I'm also practicing that as well.

Plus the prompt to make the image simpler and easier to reference is also welcome.

 

[BenJepheneT]

[unrelated] If you're proficient with tags on Danbooru sites, you'll never need AI prompts to get your references, ever.

 

[Corty]

@gaylolis

white background

This one needs a good amount of strength so use your chosen AI engine/method's way to give it ample strength in creation.

1girl / 1boy

Once again, you need ample strength but less then white bg

solo

For good measure but leave it without strength modifiers

masterpiece character design reference sheet

Believe it or not, this works. At least, it worked for me. You can try using it as "reference sheet" or "design sheet," "character sheet," depending on the AI you are using. One of them should work. It needs ample strength, preferably only one below you gave to the white background.

full body

medium shot

These two at equal strength, raise or lower it, depending on the images that your AI makes.

Sharp focus

Same as above.

best quality

masterpiece

highres

extremely detailed girl

Some extra prompts for better imagery

negative prompts, pick and chose or just copy. But don't forget to change girls and boy around, depending on what you are making:

lowres, polar lowres, bad anatomy, bad face, bad hands, bad body, bad feet, bad proportions, bad leg, more legs, worst quality, low quality, normal quality, gross proportions, blurry, poorly drawn, text,error, missing fingers, missing arms, missing legs, short legs, extra digit, 2girls, long skirt, indoors, out of frame, only_upper body, only_lower body, low background, distorted perspective,black background, colored background

That is the basics I think. Hope this helps a little.

Edit:

If you have your own SD set up, it is much easier as you can use control net, poses and other, handy-dandy extensions. But if you work it through one of the popular sites, the above mentioned things should help out.

 

[Cipiteca396]

Use underscores(_) in place of spaces in some models.

Strong (in my experience)

(artstyle)
looking at viewer
from behind
upper body
arm raised
standing
waving
sitting
kissing
hugging
spread legs ()
head tilt
pointing
jacket on shoulders
wide sleeves
hand on own chest
clenched hand
hairstyles (ponytail, twintails, short, long, ringlets, etc)
color_(thing: skin, hair, eyes, nails, clothes, background, print, etc- but the more colors, the more confused the AI will get- use specific words like blond where possible!)
bare_(bodypart) (Or just use bodypart)
blush
bright pupils
glowing
aura
attention (A pose)
mature female
pointy ears

Iffy:

expression (smile, angry, crying, pout, grin, mischievous, etc)
running
hair between eyes
looking over shoulder
turning
laying down (laying on _?)
holding item
two-tone background
puffy pants
bridal gauntlets
holding hands
closed mouth
casting magic
gradient

Negative:

highleg_leotard
tongue
bodypart
no_panties
censored
nsfw
nude
barefoot
interlocked fingers

There's a lot more, but I'm having some trouble with it.