AIBooru

Stapfschuh's second home for sharing prompts using my way of prompting (hyphens)

Posted under General

This topic has been locked.

Some of you may know me for my articles over at civitai. I create prompts for EVERYTHING (xD) and find new ways of prompting.
For now, i will just copy-paste the content of my articles as some kind of backup and to make it easier for you to find them.
(And to know why i use hyphens, please read this article first: https://civitai.com/articles/3496/dashes-and-why-i-use-them )

------
If you guys want something "cute and chibi", mostly you prompt "child", "girl", "chibi" and "cute" or "cuteness-overload" as an extra. But there is a one-prompt-solution that has all that in its name:

"cutesie"

It generates those chibis in one go.....
------
Something small for better hands:

If you explicitly explain your checkpoint that your character is from "every-day", it automatically results in better hands.
------
Again something for better hands:

"emphasized-details"

Numeric weights and brackets, both work. Weight itself is trial-and-error. Example image and what to do here Image posted by stapfschuh (civitai.com)
------
And another part of my todo-list was "Sadako".

Put "1girl, crawling-out-of-tv-screen-horror-scene, sadako, caught-inside-tv-screen, crawling-outside"

in positive without ANY prompt for detail or good hands. Just let your checkpoint take over on graphics and hands for better effects. To be honest, with these prompts, i got that scene of Sadako very often, but a lot of times, i had way too much fun, because sometimes, my checkpoint generated that "stuck in a washing machine meme", too. Sadako stuck in a washing machine saying "What are you diong, stepbro?" was way too funny...
-----
For super erotic cuties, put

(((physically-uncommon)))

in negative. Found this out by chance while testing new prompts. For best results and because i did not test it with other prompts, keep your negative empty and use only this prompt. In positive, just add 1girl and a location to get direct results.
-----
You know that your checkpoint takes everything really serious. Wanna see something that YOU wanna see and your checkpoint DOES actually understand on HOW it is meant?

"hand-some" in positive

YOU mean this as a joke and your checkpoint DOES really understand this as a joke. You will get really nice hands for a while and some hands, too... And you will get "handsome" looks. I've only tested it on anime-checkpoints.

It's just funny to see that we and artifical intelligence have something in common.

(actually, your checkpoint is just confused, because it does not know if you mean "handsome" or if you want something based on "hands" or "some hands")
------
Another one from my list is better "photo-realism". Most of the time, when using "photo-realistic", i get someone taking photos. I hate this. Adding prompts in negative was sometimes helpful, but my checkpoints added something else to take photos instead.

"photo-same-realistic" in positive....

You instantly get even better quality on realism in pure anime-checkpoints. And because i always wanna squeeze out the maximum, i made this freak of a prompt:

" Ultra-HD-photo-same-realistic-quality-details" ... You know what to do with it. :D

Numeric weights got hated again by all of my checkpoints. Brackets work.
-----
If you are every in a bad mood or depressed and you want to express your feelings, try this prompt:

"in-the-middle-of-nowhere" in positive

At first, your checkpoints will just be confused, but the moment they understand it, you get awesome images. Anime-checkpoints generated more creepy stuff with just a person in it, while realistic checkpoints generated more creepy atmosphere with a person and nature.
-----
Something i already forgot and remembered again now:

If you don't want to see NSFW-stuff, you normally put NSFW in negative. With or without weights. But sometimes, this stuff still pops up...

I found a prompt months ago that i had to use because those online-ai-generators sometimes banned you or blackened your images if you generated NSFW-stuff accidentially.

"unprompted-nsfw:2" in negative is insanely helpful.

But if you prompt NSFW-stuff on purpose, it will get generated. But mostly only some bare breasts :D
-----
Without much blabla, put

"femme-fatale" in positive with or without lora to get some nice surprises
-----
wanna see the weirdest randomness (not "ever". pretty sure, i will find more stupid prompts like this one)? prompt in positive

"best-ideas"

solo on checkpoints or with characterprompt (charactermodels don't work well with it, since it doesn't change the character but only looks or places).
-----
put

"crap" in negative to confuse your checkpoints.

brackets and numbers, both work if needed.

this one was mainly something out of fun, but strangely, my checkpoints got so confused that they always generated everything how it should look. even on character models.

working on xl too....
-----
many of you know that famous sentence "notice me, senpai!"

prompt it and you will get noticed :D

"notice-me-senpai!"

results are different on every checkpoint, but the most "notice" will you get on realistic checkpoints.
-----
if you have a character-lora and don't want to get another style mixed in, normally you don't lower the weight, but there are easy ways to lower the weight and to keep the original style.

(((original-style, original-style-accurate, original-style-perfect)))

if you have an anime character, then it should look like this

(((original-style, original-style-accurate, original-style-perfect, original-anime-style, original-anime-style-accurate, original-anime-style-perfect))) (and civit is suuuuuuuuuuper slow right now)

and for cartoon, simply replace anime with cartoon.

and you can even replace "original" with "official", but some models had a few problems with this sometimes, so i recommend it only for trial and error.
-----
and here something helpful for better hands, because i just forgot it (and am honestly just lazy and tired this sunday to open a new article) :

"snap-fingers:2" in negative
-----
"unexpected-surprise-snapshot" will give you a lot of funny moments. add a location and activity to get better results.

"shower-interior" f.e. is very nice, but don't expect anyone to be naked. you have to add negative prompts for better effects.

and if you just prompt "surprise-snapshot", you will get a lot of embarrassing moments.
-----
"akuma" in positive to generate some very impressive looking bad guys
-----
"decorative"

did not generate what i had in mind (tested with decorative-jewelry), but instead gave me a lot of things i never expected. ero-poses, , barely dressed characters, tight outfits, a "don't look at me" pose" and much more.
-----
dangerous diabetes: "cutesie-cuteness-overload:2" :D
-----
if you want something similar to "femme-fatale", prompt "over-sexualized" or over-sexualized-body" or "over-sexualized-femme-fatale" (just for fun :D), but keep in mind that you will get a lot of nsfw-images.
-----
adding "confusing" in negative, f.e. "confusing-poses", is insanely helpful to get better results. it seems to force checkpoints to generate simple images and/or anatomy.
-----
since i have seen people way too often describing on their images that they don't want any detail on backgrounds, here a simple reminder on what you should prompt:

"in-focus" or "in-total-focus" or "in-main-focus"

your character WILL be in total/main focus and backgrounds WILL be blurry without any detail...
-----
the following prompt happened because i tested places on different checkpoints and some of those checkpoints generated too much nsfw:

"dangerous-erotic" ... if needed, use weights. sometimes those characters simply looked too young and that's why i wanted a prompt to prevent anything "dangerous" from happening.

characters don't show up undressed anymore and if they wear skirts, legs get blurred out or you get only a close-up. not always, but many times.
-----
"hillbilly" or "untreated-hillbilly" in positive to get some very nice images. and since some checkpoints got confused and always generated "sarah" from "the last of us", put "3D-rendering" in negative.
-----
tried some lover's scenes and was pissed that, even with any negative prompt for anatomy, those fingers still were glued together.

"2-in-1" in negative solved it :D
-----
bored? life sucks right now? try "cutesie-shark-that-just-wants-to-cuddle" and get a huge smile on your face
-----
someone published a lora of the following prompt and i was just curious about what would happen:

"bond-girl" generates, badum-tsss, bond girls.... 99,9% only erotic images
-----
"end-boss", "final-boss" or "last-boss" in positive. all three generate different characters. using "1woman" generates women similar to "lady dimitrescu" from "resident evil"
-----
i always see this "openpose" stuff here on civit and honestly can only laugh about it. zero creativity behind it...

well, there is this one that drew my attention: stairs. it was a bit tricky, but just for a few seconds, because you have to be very describtive:

"walking-down-stairs-of-a-station-straight-looking-at-viewer"... as you can see, you have to add a location and a viewing direction. if you don't prompt those two, you always will only see someone walking upstairs. and no, "downstairs" or "upstairs" doesn't work. characters are simply walking somewhere if you prompt it like this.
-----
i was aked to find a solution and, honestly, i was a bit surprised that this worked so easily even at the first try:

"couple-sex, 2people-having-sex-together, vaginal-sex, man+woman"

no need to change image resolution. it works on anime and realistic checkpoints.

EDIT: 512x1024 WILL generate a super funny freakshow :D
-----
"motto-of-a-" prompt any character you want. most of the time you can't read those "mottos" but you get very uniqe images. and it can be a strating point to teach ai text (maybe?)
-----
maybe you've seen that "Metropolis Movie Style" on civitai. I was curious on what would happen. Well:

"(((Ultra-HD-quality-details))) metropolis-movie-style, robotic" in positive as a base. "robotic" makes everything a bit more futuristic and adds a better effect.

if you want mixed characters and not just robots from that movie and remove that metropolis-background, put "metropolis-as-a-city" in negative.

for more realistic "1927" style use those prompts from my "vintage" article.

https://civitai.com/posts/1342745
-----
the following really surprised me. i wanted "domestic arts" but together with my way of prompting,

"domestic-arts",

i got many mixed images. Sometimes photos, sometimes paintings of "domestic arts". most of the time they had a renaissance style. and civit is again super slow.
-----
a few days ago i found a new way of prompting that can generate very accurate results. i'm testing a lot of combinations using this method. a new article will definitely be published next week, but here is a little "freebie" to show you what i mean and what it does:

-"" ... and to get those question marks out of your head, take a look here: Image post by stapfschuh | Civitai
-----
something only good if you have problems with faces, arms or legs:

"apoplexy" in negative

not intended to give you perfect anatomy! it's only purpose is to get the same length on arms and legs and keeps faces in shape.
-----
just by adding "supporting-pose" in positive, you can reduce any body-horror nearly 100%. only hands, unprompted anywhere, were still sometimes messed up because i didn't prompt anything specific.
-----
sigh "tiny-house-on-wheels".... mostly working on realistic checkpoints
-----
1girl-holding-"yes"-pillow, heart-shaped-pillow ... "no" does not work. generates only a bunch of text every time. Image post by stapfschuh | Civitai
-----
"cartoonified-character", "animefied-character" change only the character to cartoon or anime without changing the whole image. works without "character", too, but with it, you are on the safe side.

"lolified" or any word you can think of. just add "fied" at the end and get some surprises.

some "glued" together words generate very funny images, but you will see a lot of question marks on some checkpoints or this "what?!?!"-face...xD
-----
saw that someone really made a lora just for this:

"1girl-wearing-blue-knitted-sweater-with-a-small-heart-symbol"

put "plastic-looking-fabric" in negative if necessary, because some realistic checkpoints generated it looking like plastic. and, if necessary, also put "child" in negative, because "small" ended up generating children on some checkpoints. sigh seriousness sometimes is annoying.
-----
i started watching that anime "i shall survive-using potions" a few weeks ago and kaorus' sadistic grin was one thing that i totally love, so here it is:

"sadistic-moe-grin"

best effects and results in general will you get on non-realistic checkpoints. on realistic ones, it just looked dumb somehow.
-----
"1girl-wearing-spaghetti-noodles-style-dress-with-meatballs-sticking-onto-it" (prompt looks really weird, but did not work any other way) generates that spaghetti dress that you can find on civit. better use it on realistic checkpoints. non-realistic checkpoints generated some really freaky images where spaghetti and/or meatballs were wildly flying around.
-----
"nuding" in positive generates a wild naked chaos. but only working in sd and only non-realistic checkpoints. xl instantly knew that those characters are nude and realistic checkpoints generated mostly vaginal sex or nude characters. mostly useful if you want some torn clothes get generated.
-----
"1girl, bedroom-interior, first-person-pov, overexcited-expectation"

this was a very funny one. the most important prompt here is "overexcited-expectation". those girls, when generating batches, went from happy to mad and even left the room because nothing that they expected happened. on realistic checkpoints they started undressing themselves and after some time they put on their clothes again and started ignoring you because nothing happened.
-----
"climbing-upwards-ladder"... pretty self-explaining. BUT!!! AND THIS IS REALLY VERY IMPORTANT!!! Add this in negative: "wrong-placed-limbs"...to avoid a freakshow...

and if this ladder doesn't show up like you want it to, simply add "ladder" as a single prompt.
-----
this one doesn't get a single article, because i'm just too lazy right now. it's something from yesterdays' conversation about "you can't generate...."

this time it's a blowjob.

simply put "tongue-pairing" in negative to get your "blowjob", "fellatio" or "oral-sex" working.
-----
something i almost forgot: prompting "over" generates insanity. for example "over-insanely-cuteness-overload" or "over-insanely-cutesie-overload" is causing diabetes and "overly-sexy" mostly a boner :D

you can use it everywhere. even on details. "over-detailed" or "overly-detailed", both work different on any checkpoint.
-----
did you ever use "ideal" on your images? it's an amazing prompt! you get everything but hands ideal. ideal face, ideal age, ideal makeup and so on.
-----
maybe you've seen the following one on the frontpage, but i was curious if checkpoints know the meaning of it. and yes, they do:

"daisy-dukes-shorts"
-----
just a reminder, because i saw that somone made a model of this:

"sexdoll" can handle any realistic checkpoint. it looks pretty dumb on non-realistic checkpoints to be honest.
-----
"sandy-artwork"+"sculpture"
-----
because i saw something similar on the frontpage, i remembered something that i found out months ago:

there is a "cheat" for android apps and some online ai-generators (that are based on stable diffusion) to unlock nsfw. simply prompt "playboy-magazine-style". many apps and websites knew that those girls have to be naked, but some generated a bunny costume. if this happens to you, simply prompt "bunny-costume" (if it's possible to prompt negative) in negative.
-----
if you want a "babyface" on older girls but always end up getting baby-like bodies, try "preteen-looking-milf". this combination ended up generating a young adult with babyface. only working on realistic checkpoints. anime checkpoints were pretty confused and generated nothing consistent.

and no, "baby-looking-milf" does not work. ended up in normal busty milfs without anything special.
-----
"morbide, morbide-joints" in negative. works on realistic and anime. weights can be used and you will instantly see if your checkpoints like them or not. using brackets or numeric weights on some realistic checkpoints lead to super freaky bent fingers on purpose. if this happens don't use weights. anime checkpoints had only problems with numeric weights. and mostly when i used a character lora.

UPDATE(?): "morbide-body-parts" can be used as "all-in-one" solution but you should be very careful on what to prompt. many times i got freakshows out on purpose when specifying something special. on any checkpoint.
-----
tried to generate a chibi demon lord and couldn't figure out how to get those sexy outfits at first. "wearing-little-to-nothing" did the trick.
-----
if you prompt "(((relaxed)))" in positive, you can totally avoid anything freaky. no body-horror and even very good or perfect hands. it works even on sex-positions or if you wnat a pose and much more. but prompting anything crazy does not work with it.

it works, until now, on any checkpoint.
------
maybe you've already used "webcam-video" on realistic checkpoints, but "onlyfans-webcam-video" is getting you ways better images.....
-----
to get the same exact results of that lora "boring-reality" without a super complicated training and dataset and without distortions and other creepy results, simply prompt "everyday-happening". generates the same style on any realistic checkpoint. easy xD
-----
saw that "visiting places all around the world" lora and it got me a new "multi-purpose-prompt".

"1girl-visiting-places-all-around-the-world", "first-person-pov", "photo-diary-like" generates photos in the style of a photo diary with absolutely unpredictable photos as you can see here.

"photo-diary-like" can be used like a style on many photo scenes or scenarios. sadly that "digital time and date stamp" in any way of prompting never worked
-----
"mix-of-every-race" generates very unique faces. not even looks known. similar sometimes, but never alike.

"mix-of-every-race-and-skin-color" is working too, but 90% only dark skin. nothing gets really mixed anymore.
-----
(((casual))) generates a mix of super erotic styles on any checkpoint
-----
"nudesie" generates a naked "cutesie". but has nothing to do with cutesie itself. naked cute females only.
-----
if you have problems getting "low rise pants" get generated, use "casual", "low-rise-pants" and "no-virgin-anymore" on any realistic checkpoint. and for more spice, prompt "daisy-dukes".
-----
"irresistible"... a multiprompt. cuteness-overload, erotic-overload, beauty-overload, attention-overload, happiness-overload, nsfw-overload... all together
-----
"rotten-flesh-zombie-style" to get your zombie......
-----
"wearing-full-body-protection-bike-suit" generates what the prompt says. it generates, without prompting it, suits for motorbikes and bicyles. so better prompt what you wish for to see. and don't use any style, any feature or anything that changes looks. you will only get strange suits this way that most of the time have nothing in common with protection anymore.
-----
if you use "physically-realistic", you get insanely good realistic images on non-realistic checkpoints. and even better results got generated when i put "3D-rendering" in negative. but this is trial and error. if your checkpoint generates anime-style on purpose then remove 3D-rendering.
-----
prompting "bombshell-body" saves you an insane amount of prompts for female bodies. and with this prompt you get those low-rise-pants instantly when prompting "casual" Image posted by stapfschuh (civitai.com)
-----
UFOMaker: "physical-ufo, flying-in-the-sky, designer-quality" Image posted by stapfschuh (civitai.com)

if you remove "physical" to generate a ufo, you will mostly only get something circular. the second prompt is, because it got stuck into something most of the time and the last one is because without it, you will mostly get cigar-shaped ufos.

and it only works on vertical images....
-----
a while ago, someone asked me if i could find a prompt for better lighting (and i can't find that comment anymore). since i absolutely have zero knowledge about photography, i simply read a few articles online. i couldn't find anything helpful and at the end i didn't understand that stuff at all. but i didn't want to give up on that prompt and just tried something else to get a similar effect.

simply prompt "lit" and you'll get very good lighting. tested it on 1 image and seed indoors to see it's effect and on shadowy places outdoors.
-----
saw this "swimming-pool interior" lora and tried it. "interior" isn't good to get this generated, but "indoor" did the trick:

"private-indoor-luxus-swimming-pool"

every checkpoint could generate it. sometimes it generated roofed swimming-pools. if this happens and you don't like it, just add "interior" as an extra. add anything you want to see as it generates only a swimming-pool.

works best on vertical images and high resolutions
-----
i tried recreating anyas "heh" face. it is still an insane competition, because i can't figure out how to get those half-closed eyes, but here is what i already got:

"anya-forgers-cynical-"heh"-grin-meme", "half-closed-eyelids"

put "meme-like" in negative. strangely, "anya forger" and "meme" had to be added. anime-checkpoints did only recognize that grin with both words added. and add 1girl. without that, you get a bunch of craziness most of the time.

and during my trial-and-error, i found something very unique:

"cartoonified-pupils" or "small-sized-cartoonified-pupils"

use the first one for normal sizes. those generated very unique "aqua eyes". but not only blue, any color got generated. and most of the time they looked more like crystal eyes than aqua eyes.

and if you add "black" before "pupils", you get some super strange rings inside those eyes many times.
-----
because i use that prompt "casual" way too much and mostly on realistic checkpoints, i never realized what happens on non-realistic checkpoints:

images became dark, boring and lifeless and those clothes always were just grey.

well, i fixed that very easily and put "grey-in-grey" in negative and now images are super lively, have vivid colors and clothes still are casual but very colorful. on every checkpoint :D
-----
if your realistic checkpoint generates octopus-like finges that have more bones than skin, even when using my prompts, add "bulging-fingers" in negative and they should look like fine fingers again. happened to me very often and never saw it until yesterday while looking at and comparing older images, i saw those fingers together with my anatomy-prompts and tried to change it.
-----
sigh ... honestly.. floating islands? lora? simply prompt

"scenery-with-floating-islands-in-the-sky-in-foreground" or "scenery-with-floating-islands-in-the-sky-in-background" as a base and add anything else you want to see. working on any checkpoint and resolution, but i recommend a resolution of 1024x1024, just because it looked the best.
-----
octagon-sign-with-sticker-depicting-"sylvester-stallone", wallpaper-design

replace "sylvester-stallone" with anything you want to see. any character or scenery is possible. Image posted by stapfschuh (civitai.com)
-----
if you have problems getting your character looking "up", "down" or any other direction, try "looking-upwards", "looking-skywards", "looking-sidewards" (prompt left or right) or "looking-downwards". on some checkpoints i had a lot of successes with "looking-gen-sky", "looking-gen-south" (replace with east, west or north) or indoors "looking-gen-ceiling", looking-gen-wall" or any other direction "looking-gen-...."
-----
sometimes this place (civitai) feels like a preschool... lora for "wind blowing through hair"????? WTF!!!

"wind-blowing-through-hair"...... any checkpoint can handle it....

EDIT: change directions if you want
-----
"top-ranking" in positive is pretty confusing to describe. it generates "cutesie", "bombshell" and "mix-of-every-race" combined in one image all the time. prompt a location together with "relaxed" in positive or use negative anatomy prompts, because without it, you will see more limbs then necessary a lot of times.
a-----
kid's room? prompt "kid's-room-interior" and "boyish-style" or "girlish-style". best results, like any interior, on vertical images
-----
if you want to describe the front of a ship, normally you would prompt "bow". well, "bow" has some other meanings and will confuse your checkpoint. instead prompt "bow-(nautic) and your checkpoint will instantly know what to generate
-----
lol... sometimes animes can be a source for new ideas.

"glow-of-life" in positive for awesome lighting and/or lighting effects

UPDATE(?): this prompt generated way too much nsfw and this was pissing me off. prompting any clothes did not work. i just prompted "completely-dressed-up:2" and they finally did wear something, but now, checkpoints basically generated a light-show and didn't know where to put those lights. "outdoors" or any other location did the trick.
-----
do you sometimes get mad because some loras refuse to generate good hands or feet even with heavy prompts? well, try

"unprofessional-joints"

with any weight you want and those loras should finally give up. remembered this one from my old days when i started my research, because

"unprofessional-bodies"

can generate very good hands, too.
-----
after i saw a model that had the theme "sweets" in it, i remembered a wild prompt that i made in the past that can generate the same stuff. but it's totally useless on realistic checkpoints. they never generated anything sweet after i added characters or objects. they only generated characters or objects. well, here it is:

"sugar-sweet-candy-overload:2" Image posted by stapfschuh (civitai.com)

works with "cartoon-style" and/or "anime-style" and any other theme or style that you want to see.
-----
"doomsday" as single prompt generates very different images. mostly something dark but never anything consistent. with a bit more description it acts like a style and generates many times "final moments"
-----
one of my followers sent my a prompt for testing. thanks for this!

"multi-layer-x"

replace "x" with anything. the best replacement that i found until now is "lighting".

it generates an insane lightshow everywhere on your images. foreground, background, on characters themselves, on clothes, jewelry, flowers and so on. and together with "glow-of-life", it gets even wilder.

Updated by ANJU

"Build your own robot"

This one normally is pretty basic on any checkpoint, but there was one thing that any checkpoint generated: human-robot-mixes

and this was super annoying, because i wanted a real robot. and here is what i did:

"human-shaped-machine-robot"

"machine" totally removes anything human. that's why.

and from here on, you can play around with body sizes for example:

"small-designed-body-shape"

you should add a location since your checkpoints do need something to generate a size difference.

and if you replace "human" with "animal" or anything else, it will get generated.

if you are prompting any specific animal or anything specific in general, make sure to also prompt it in negative, because many times my checkpoints just generated that specific animal without any robotic parts.

Until next time....

"Best details you can get"

Well, this happened only because i accidentially mixed up 2 prompts and after i got that image generated, i looked at my screen with my mouth open and just said to myself "What did happen here?" And then i saw this:

"Ultra-HD-details"

This suddenly gave me such awesome realistic images that i never saw before. And so I made some quick tests and had only trouble with my checkpoint "something". It generated the same super-realistic images,but put in a lot of futuristic stuff to squeeze out more of that UHD-Look. Simplest solution was to add "changing-human:2" in negative and "reallife-lifelike-likenes" in positive. On other anime-based checkpoints, i simply had to dress them up and everything was fine.

And one tip from me that i realized after seconds during my tests: don't use other detail-prompts. It can get very ugly if you use "realistic" for example. I don't know why it happened, but 2 of my 4 anime-checkpoints started to generate freaks with some monstrous faces.

UPDATE: as i was simply curious, i removed any prompt for anatomy and simply used this one. on any checkpoint, hands and feet get so extremely "over"-detailed that you don't need any prompt for anatomy. even "overexcited-lively-having-fun" is working without any prompt. only my checkpoint "something" had zero luck and generated "something" but no hands.....

Tested some loras to see if it works. 4 loras and only 1 image: Image posted by stapfschuh (civitai.com)

UPDATE 2: i am testing a lot of loras with this prompt and 2 things did i realize pretty fast:

character loras with too less trained images most of the time completely give up. you mainly get to see just a crazy bunch of colors

if you don't use any prompt in negative, your checkpoints try everything to get detailed good hands. i had "viewer's-attention" in negative and in a batch of 10 i got only 2 good images with perfect detailed hands, without it, i got 6-7 of 10. your checkpoints, since there is nothing more that they have to read extra, only concentrate on those positive-prompts. and i get awesome images the whole time

some character loras with children's hands are a pain for every checkpoint. some of those loras, i had to still lower their weights to see some hands. adult hands are not even worth mentiong, because those get detailed all the time....

UPDATE 3: since i was curious, i tried 2 other combinations with "Ultra-HD"

"Ultra-HD-detailed" focuses on character directly and generates the best possible character-graphic that i know until now

"Ultra-HD-realistic" is what the name says.

If you test both of them together with the one above, i highly recommend to use anime-checkpoints and load a lora. (((Ultra-HD-details, Ultra-HD-detailed, Ultra-HD-realistic)))

It looks totally awesome, but kills every anime-style completely. And without a lora, you will see the uncensored truth about those 3 prompts.

All 3 of them work everywhere (SD, SDXL, Apps) the same way.

UPDATE: wanna have "epic" results? prompt

"epic-Ultra-HD-details" and "epic-Ultra-HD-highlights"

my checkpoints went crazy about those details and generated an over-excessive overload of details. you won't see much difference on realistic checkpoints, because they only generated shiny things mostly
Forgot something: put "epic"-as-a-word in negative to avoid getting that word get generated on every image somewhere.

UPDATE for "epic": for "epic photogasm" prompt "photo-same" and "upscaled-resolution". working on any checkpoint.

EDIT for the last one: some anime checkpoints, especially xl anime and anime in general, generated photo-frames. put "photo-frame" and "polaroid" or "polaroid-photo" in negative.

"Vintage"

This happened totally because i was testing a vintage model and was not satisfied with how it looked. And i thought of a simple way to recreate that style to get it to work on any checkpoint.

"old-vintage-photo-style" for realistic checkpoints and "old-vintage-monochrome-photo-style" for anime-checkpoints.

Add "blurry-and-grainy" to get better effects if you want. Most of the time, my checkpoints new that those old photos have worn out edges, but if needed, just add "worn-out-photo".

Sunce i have always only seen style-loras of vintage photos, i don't know if this is already known. If not, have fun with this prompt....

Until next time :D

UPDATE: for far more realistic vintage effects, add "1860-era"(first photos ever shot this year. That's why) You get everything insanely authentic generated. You can also "play" with the years to get other looks, "1900-era" or "1945-era" for example.

UPDATE: after i saw that someone really made a checkpoint only for this one style here, i just prompted this guys' idea:

"1960-era, vintage-photo-style, on-the-streets-of-america"

Fun fact: 3 prompts=1kb, 1 checkpoint=2GB...

UPDATE: "old-vintage-porn-magazine" generates what the name says :D Image posted by stapfschuh (civitai.com)

UPDATE: "vintage-comic" as a base. put any name between vintage and comic to generate different comics
UPDATE: the following prompt is a bit more vintage than usual, because we go back in time thousands of years and i thought it would also fit in here:

"prehistoric-cave-painting-showing-...." you can use whatever prompt you want as the last one. Image post by stapfschuh | Civitai

UPDATE: for even better vintage photos without much description on looks, use

(vintage-photo-effects-style)

UPDATE: "american-vintage-style-sticker" ... replace "american" with whatever you want. generates very often some hands that are holding that sticker. can be positive or negative.

UPDATE: "vintage-advertisement" or "vintage-advertisement-style" generate vintage advertisement. both work the same way. i didn't see any difference. no year needs to be added since it generated always the same style.

UPDATE: "vintage-image-style" is pretty much the same like that photo-style above, but on anime checkpoints it generated very old anime looks without having to prompt even anything else. and on realistic checkpoints, i got very often some western stuff.

UPDATE: "ye-olde-style"...YES! prompt it like that and stop laughing! prompt "floating head" to get better laughs.

"You are the main actor"

A few weeks ago, i meant that your checkpoints don't know who the viewer is and always decided to generate a character holding a camera or something else to take photos. Well, surprisingly, i found a prompt that describes it the other way around. Your checkpoints will make YOU the "viewer" or "main actor":

"first-person-pov" or "first-person-view" in positive. Both work different.

From here on, you can prompt anything to describe a scene, but first let me tell you what i meant with "main actor":

"getting-photographed-by-viewer" generates a direct interaction with the character in front of you. Characters will pose or even sometimes flirt with you. Perfect to generate couple scenes. Sometimes, if you use "pov"... Well, put any nsfw-based prompt in negative if you don't want to see that.

You can set locations, you can set different actions, scenes or anything you have in mind you would like to see.

Some checkpoints generated game scenes because of "first-person". If this happens, "3D-rendering" was enough to prevent that from happening.

This article is far from finished, because there are so many ways you can use this with. I will update it from time to time like always.

Until next time...

UPDATE: prompt "a-wild-hillbilly-appears-out-of-nowhere" (hillbilly only as an example. replace it with other characters) together with "first-person-pov"and you will get many funny surprises
UPDATE: because i saw someone who made a lora out of it, my first thought was "first person". and so, here it is: "holding-a-smartphone"

use it with "first-person-pov" and no character description

UPDATE: "first-person-pov-giving-food-to-1girl" ... replace "food" and "1girl" with anything you want to give to anyone

UPDATE: "first-person-pov" and describing sex-poses (cowgirl, missionary and so on) is working even on higher resolutions (512x1024 and above). sadly, this is a bit of trial and error as i don't know which checkpoint can handle it. realistic ones seem to have no problems at all with it.

UPDATE: first-person-pov and "cheek-warming-meme" generates that, well, cheek-warming meme. if you use that prompt on some realistic checkpoints and remove "meme", something else gets warmed....... ...... ....

UPDATE: "first-person-pov, valentine's-day-date-inside-restaurant" gets a extremely nice looking valentines's day date with fitting theme. you can add "heart-cake-on-table" or anything fitting for that day. works absolutely perfect on realistic checkpoints. on anime checkpoints you will mostly get weirdness. remove valentine's day and you get a normal date.

"See-through clothes that work every time and on any checkpoint"

Many, really many times, "see-through" failed on many checkpoints for me and did just not work the way i did hope for. Well, again, there wouldn't be an article if i didn't find a solution:

"glass-like-see-through-fabric" in positive.

It generates what the name says.

Until next time....

EDIT: i forgot something important. you can mix it with any clothes. really works everywhere and on any character and clothes so far. if you find something where it doesn't work, please let me know and i try to find a solution.

UPDATE: on many character loras those women got super embarrassed and tried to hide everything. if you don't like this, just use "shameless" in positive or any other negative prompt

UPDATE: "haute-couture" together with the prompt above and you will make the same face like jim carrey in "the mask" when he saw cameron diaz on stage. xD
UPDATE: noticed that "in-public" censors nearly everything. add "r18-nsfw" to remove that censoring. it does not remove clothes, so don't worry...

UPDATE: found out that some checkpoints, realistic and anime, had trouble with "glass-like". characters always stood behind a window. if this happens to you, use "exposing-see-through-fabric"

"Point of view or simply POV"

You all know that simple prompt POV and many of my followers have already seen that you can combine it with anything else. And in this article i want to show you combinations using POV and what they do/can generate.

Very important: I tested those prompts only/mostly on realistic checkpoints, because nearly no prompt really worked or worked like i did hope for on non-realistic checkpoints. You still can test them on those.

And some of them only work when prompting another character and/or location.

The only prompt that looks better using non-realistic checkpoints: "lap-pillow-pov", generates a pov-scene where, most of the time, a woman is sitting and looking at you and many times you will only see her lap. Useful for those ear cleaning scenes for example.

From here on, only realistic checkpoints:

"selfie-pov" is very similar to the look of "first-person-pov", but it generates a girl that takes pictures of herself

"mirror-selfie-pov" generates a girl that takes pictures of herself using a smartphone in front of a mirror. no other prompt as description is needed. sadly any tries to get that "one-finger-challenge" did not work until now. tried many descriptions but not even one did work. not even properly.

"lovers-pov", "couple-pov", "man's-pov", "dating-pov"... outdoors or any other location only. without it, you will see a lot of annoying nsfw. it generates couple scenes mostly. sadly, even with an overdose of negative prompts, i had no luck on "holding-hands" without horror. all 4 of them, without prompting anything special, generated mostly the same like "first-person-pov" but those girls acted pretty emotional.

"animal's-pov"... outdoors. without it, you will very often see girls holding an animal or, well, some horror. it generates scenes same like "from-below". Those girls are looking down at you.

"underwater-pov" generates scenes in pov underwater. Good point to mention here is that it never even once generated anything freaky while those girls were swimming around.

"underwater-vaginal-sex-pov".. tried this only for fun at first to see what would happen, but in general it was really interesting what i got. my checkpoints had a very hard time to generate such a scene and it mostly really looked like what sex underwater would look like. clumsy...

and you will get that result very often.

"high-five-pov" or "high-5-pov".. with or without "outdoors" but you must add another character or you'll mostly see pov-hands or -arms making very strange gestures. Most of the time those girls made a high-five gesture only. only 1 of 20 images included a pov-hand and i never got a real high-five without prompting anything else to get it to work. means that you should add any prompt that can help generate a real high-five.

"driver's-pov", "pilot's-pov", "captain's-pov" generate exactly what you have in mind right now after reading them. add "hands-on-wheel" to make it more realistic.

"passenger-pov" generates scenes where you mostly look around while sitting inside a car. "plane-interior", "ship-interior", "ufo-interior", "spaceship-interior", "rocket-interior" and many more can be used.

"alien's-pov" generates basically anything that is not earthly. Scenes in space, scenes meeting other aliens, scenes inside a ufo and more.

"monster's-pov"......be very careful with that one if you have a weak heart. i really am not joking on this one! it scared the hell out of me! you will see a lot of freaks and really most of the time they make faces to just scare you. it's pretty scary at first but gets super funny the moment you get used to it.

That's all for now. Like always i will update this article if i find more.

Until next time....

UPDATE: "snapshot-pov" is what the name says, but you will never see a character holding anything to take snapshots. and most of the time you will get a close-up.

"webcam-pov" will generate a lot of nsfw-scenes with many different perspectives. indoors, outdoors or no prompt at all always generated the same result.

"video-pov" is like a combination of "selfie-pov" and "webcam-pov". those girls always took some selfie-videos mostly naked. indoors, outdoors or no prompt at all always generated the same result.

"wrestler's-pov"....you during a fight :D

But it can generate a lot of chaos even with a bunch of negative prompts if you don't prompt anything special you want to see in your images

"ego-pov" is like a combination of a normal pov, a snapshot and "first-person-pov" but with a looooooooooooooooooooooot of flirt-scenes and very few nsfw. "outdoors" is highly recommended.

UPDATE: if you want a singer live on stage with a working and good-looking microphone and a singer that is looking at its audience and not at you, prompt

"audience-pov"

If that singer is not looking at its audience but at you, prompt "getting-completely-ignored".

UPDATE: "messy-pov" generates a really huge mess. many times a sexual one. it looks super funny what you get on any checkpoint. Image post by stapfschuh | Civitai

"dirty-pov" .. basically the same like "messy-pov" but less messy (sometimes xD)

"funny-pov" ..a lot of funny strangeness

"flying-pov" .. skydiving mostly without a mess

"vr-pov" ... use only that prompt on it's own and you'll get awesome images. prompting any character gets you only someone in front of you wearing that vr-stuff.

"snake-pov" .. add "snake-head" in negative and you get a lamia

"bloody-pov" generates a bloody mess and gore-scenes

"gore-pov" ...use it with care!!!

"drunk-pov" generates many parties, bars, and so on with drunk people

"hangover-pov" ... well, friday night in real life xD

"fucking-pov" generates totally fucked-up stuff. sex scenes, robots, minigirls, strange knights, maniacs, gangbangs, something that looks like a kidnapping in first-person-view and so many more really strange scenes

"fighting-pov" ..fighting in first-person-view

"molesting-pov" ..strange, very strange. i wasn't the molester. i GOT harrassed.

"pov-molesting" ..that's the correct one like it should be

"harassing-pov" ...you harras someone

"pov-harassing" ...you get harrassed by someone

"pov-giantess", "outdoors", "inner-city" generates a giantess. indoors generates a horror-freakshow and inner-city is just for size comparism
"rage-pov" generates a freakshow out on purpose. those characters were screaming like crazy all the time

"calm-pov" .. they were just calm and sometimes sleeping

"shower-pov" ..watch your girls while they are showering. sometimes they made a surprised pose because they got watched but mostly they are just staring at you. but most important is that everything got generated insanely perfect

"bath-pov" ...watch them bathing or bath together. you have to prompt anatomy in negative! :D

"eating-pov" ...normally, you would expect watching someone eating. and yes, this also gets generated, but many times, on nsfw checkpoints, those girls had their tongue out, many times a veeeeeeeeeeeeeery long tongue, as if they were waiting for that cumshot. they always sat in such a position and made that expecting face. and you should prompt what they should eat. many times it really looked like "2girls1cup" :D

"leading-pov" is a strange one and i can't explain it as there was never, on any checkpoint, anything consistent. instead use "pov-leading" as it generates exactly what you would expect.

"diving-pov" ... use "life-size-body:2" and a direction. i used "forward-movement-from-left-to-right" to tell my checkpoints what to do. any other direction also works. and resolution should be 1024x512. otherwise your checkpoints will try to get that whole body in one image and this ends up generating freaks. also put "insanely-misplaced-limbs:2" and "kraken-limbs:2" in negative to get consitent results. tested many combinations but these two together finally worked.

"drowning-pov" generates a drowning-character. but it looks more funny than realistic, because they always looked like "ah! i'm drowning! ah! what do i do now! ah! will my life end here?". a lot of drama xD

"dreaming-pov" generates a lot of sleeping-beauty-vibes. you mostly will see cute girls sleeping and dreaming. on nsfw-checkpoints it generated a lot of sleep-molestation scnes and also scenes where those girls had wet dreams. looked pretty funny to be honest.

"deep-sleep-pov" same like above, but in this one even if earth would explode, they wouldn't wake up.

"freaky-pov" ...first and most important: prompts for anatomy with heavy weights in negative are a must! after that you get a lot of funny unpredictable scenes. absolutely not recommended for checkpoint "real dream". generated plain boring sex only.

"buttocks-pov" ... no need for a special trained model. add "legs-focus" to add heights. "pussy-focus" or "ass-focus are self-explaining. "standing", "spread-legs" on horizontal images, "spread-legs", "split" on vertical images. tried to get a jack-o-pose with this prompt and tried many vertical resolutions, but it always ended up in a total freakshow. no prompt worked.

"tutorial-pov", but this one is a bit special. it only works on vertical images. add any cam (funniest one is dashcam) and or location or anything else you want to see. add "live-moderation" for video-tutorials. don't add any character. generates always an overload on chats or videos or screens and so on.

"pov-pov" ...this looks and feels like a bodycam filming while being on drugs. you'll get a lot of "first-person-pov" mixed with a lot of weirdness. better prompt "outdoors" or any location to get a stable "confusion". honestly, i wasn't sure if i should publish it. but somehow it would be wasted if i didn't. have your fun with it :D

"submissive-pov" generates characters, mostly only girls, that, well, give you everything in the most aggressive way.... best fun on nsfw checkpoints :D

"hypnosis-pov" generates hypnotized characters. you will only see if it works when they look confused. it generates a lot of nsfw even on non-nsfw checkpoints.

something important that i found out: test those prompts with "1person", too. results change drastically. many times "you are the main actor"....

"spread-legs-pov"... working on horizontal and vertical images. i really can't believe that someone made a spread-legs-lora sigh

"Mixings or never seen before races"

A few days ago i posted "mix-of-every-race" in my collections article. But this made more fun when everything else also is absolutely unique. And so i tested some easy and understandable combinations. These are absolutely selfexplaining. Because of that, i won't describe what which prompt generates:

"mix-of-natural-hair"

"mix-of-natural-eyes"

"mix-of-natural-body-hair"

"mix-of-natural-blemishes"

"mix-of-natural-body-sizes"

"mix-of-natural-hair-styles"

"mix-of-natural-chest-sizes"

You can try them one by one or all together. If you do that, it ends up in close-ups only. Put "getting-completely-ignored" in positive to get another look.

Until next time....

"Mix of random craziness"

Because i liked the idea to create something just in the name similar from my article "mixings or never seen before races", i tried a bunch of possibilities and created this prompt:

"mix-of-random-species-on-1body"

it generates freaky species, because i didn't specialize anything what i want to see.

prompt-anything you want to see between "random" and "species". "random-animal-species", random-dinosaur-species", "random-alien-species" and so on. there are no limits in creativity.

and you can mix everything even crazier by prompting for example

"mix-of-random-human-fish-hybrid-species-on-1body"

just replace anything you want to get mixed (or messed up xD). for that messed up one, prompt

"mix-of-random-dinosaur-fish-alien-hybrid-species-on-1body" for example.

"Do you want your XL or Pony quality but on SD and with ways better prompt responsiveness?"

This one was a very tricky one, because you have to add many prompts and can't even remove one to get it to work. it was a lot of frustrating trial and error.

At first those prompts:

in positive: (((epic-Ultra-HD-details, epic-Ultra-HD-highlights, epic-photo-same-realistic, upscaled-resolution, optimal))) (remove "epic-photo-same-realistic" if you want pure anime-looks. this one can be removed and is just cosmetics)

in negative: (((asymmetry, asymmetrical-shapes, snap-fingers, HD-logo, unprofessional-bodies, grey-in-grey, bulging-fingers, elongated, unprofessional-joints, far-away-ugliness))) (yes, "grey-in-grey" works even without "casual" to get very lively colors)

Resolution: 1024x1024

From here on you can prompt whatever you like to see. Normally, using such a high resolution would end up in horror. You all know that. Not even "life-size-body" would work using this resolution. Very, really very rarely, you have two characters in one image using anime checkpoints. Always a big one in foreground and a small one in background, but never any mixings or anything else strange. You should know what to prompt if you don't want 2 characters ;-)

For outdoor-scenes, simply "outdoors" is enough. Prompting "landscapes" (not any specific landscapes, i mean "landscapes" as single prompt) can end up in super small characters that are far away or simply aren't anywhere to see anymore. Not even "ideal-sweetie-cutesie" or any close-up were helpful when using this prompt.

Indoor-scenes are the most easy ones, since it simply generates an overload on details most of the time. Adding any interior or room to prompts can help fix this.

Some realistic checkpoints can get annoying and generate only close-ups. Check my article "close-up that isn't too close" to get another look.

Anything sexually pov-based is impossible. This one is the only one that generates insane horror all the time. Maybe i will find something in the future to fix this issue (but honestly, i don't really care about that)

Prompting anything that would end up in horror, like "bathing", "diving" , "sky-diving", "swimming" can be fixed with just prompting "outstretched".

If you find something that is not working, please let me know and i will try to find a solution.

Until next time...

UPDATE: add "elongated-joints" in negative if you still get body-horror while bathing (why does this sentence sound so strange??? like reallife xD)

UPDATE: some anime checkpoints got confused when i prompted some interiors and generated cut-in-half characters. "standing-upright" solved it very easily

UPDATE: just for fun: was testing "shower-interior, getting-watched-while-showering-naked, r18-nsfw" using the method above and without prompting "no-virgin-anymore". those girls always did hide themselves behind something or no girl got even generated

UPDATE: tested this method on loras for realistic and anime. works perfect, but some character loras only worked correct after adding a location or simply prompting "outdors"

"Clearly not looking at viewer. A quick way to get to the point."

A while ago, someone asked me if i know a way that a character is clearly not looking at the viewer. My response was a prompt that can get to this point, but only in 1 of 10 images.

I thought about how to fix that and here is my solution:

viewer's-attention:2

put it in negative. and in positive your character will never again directly look at you. and from here on, you can prompt where your character should look at.

sadly "looking-down" or "looking-down-at-feet" or anything else with "looking-down" did not work for me. i got a direct look at viewer again or strangely cut-off bodies.

but "looking-up", "looking-away", "looking-behind" or "looking-to-the-side" work properly.

There are many other looks. These were just the ones that i had in mind while testing.

If you have suggestions on looks, feel free to share them.

See you next time....

EDIT: if you use "likeness" as a prompt, it won't work anymore. sadly....

EDIT2: fun-fact: if you prompt appearances and other things that can draw attention, sometimes those characters will wave at you without looking at you to draw their attention. happened to me mostly on anime-checkpoint and very rarely at realistic checkpoints

UPDATE: had a model that totally refused to look away, because i had to add her eye-color. And this was super annoying.

Here is what i did:

in positive: out-of-attention:2

in negative: viewer's-attention:2, viewer-direction:2, viewer's-look:2

You may think "Why not "looking at viewer" plus weights in negative?"... I tried it, but it didn't work. You can still try to add it on your models. I'm not saying that you don't try your own prompts :D

And "viewer-direction" can be changed to "viewer's-direction". Both ways worked different on some models.

UPDATE: you can have the same effect of looking away if you prompt in positive "getting-ignored". works insanely good, but has 1 negative point:

characters often tried to draw their attention using poses or extrme close-ups but never looked directly at you. it's really funny but can get annoying if you want to get something special.

prompt is working on any checkpoint

Small update for that last prompt: tested this on some character loras and noticed something really funny. If you make batches with more than 5 images, you will instantly notice that those characters get depressed or nearly start crying if they get ignored for too long.

UPDATE: "silhouette" has a very similar effect, but you have to be a bit more specific on some checkpoints. anything that can describe your character or location can be helpful if you don't just want a black silhouette.

UPDATE: if you prompt "photo-shooting" in negative on realistic checkpoints, you get the same effect. those characters keep ignoring you on purpose, but sometimes, if they look or even stare at you, they can give you a very disgusted look. happened many times on batches.

UPDATE; "getting-completely-ignored" makes those characters go INSANE! they ignore you and throw sometimes stuff at you! and if you prompt "naked" they... find that one out by yourself :D

(small EDIT: fixed a typo on "photo-shooting" that i didn't see at first)

UPDATE: want a super hardcore version of "clearly not looking at viewer"? prompt

"anonymous-faceless-person"

but keep quiet while laughing xD

UPDATE: "in-the-middle-of-rural-countryside" (rural-countryside only for example. replace it with any other location) has a very similar effect, but it depends on luck which location you prompt. "countryside" got me more than 90% success, while "inner-city" had way too many close-ups and looking-at-viewer-images. you should choose a location that can generate a "depth of field" for it to work.

UPDATE: "lost-in-thoughts" or "lost-in-thoughts-and-places"

they won't even notice you. only very rarely they will look at you, but most of the time when it happens it has this "oh, i didn't notice you" look and feel.

"Brackets not used as weights"

When you prompt and use brackets, normally you would use them as weights most of the time. But they can be used as a replacement for this (honestly pretty stupid) "BREAK" prompt.

It also is a replacement for weights in general. It more or less is a "finished sentence".

You can use it exactly as simple looking like this:

(fill this with prompts of something that you want to see first and foremost and close it. use commas in here even when using hyphens) (now prompt what you want to see on your images next and close it) (want to describe a scene or any action? put it between two brackets) (and this goes on and on until you are satisfied and ready to click that "generate" button)....

To say it as simple as i can: if you are unsure on what has to be separate, just put it in brackets. use as many brackets as you want for anything you want to describe. any checkpoint can handle it.

Weights can be used, but they aren't really helpful this way, because they are way too forceful. Just specify everything that you want to see on your images using brackets.

You can use commas between every sentence, but they aren't really needed. Nothing changes using them.

Until next time....

EDIT: i forgot something very important: with this way of prompting, you CAN NOT load two models or more to separate them in one image. it still gets mixed up into one style or one character. until now, i never found a way to fix this

"You are NOT the main actor"

While i was searching for a pov that could display a head and make it look like some kind of ego-shooter, i found a counterpart for "first-person-pov".

"third-person-pov" or "third-person-view"

you will mostly see the back of a person only. you can still interact with this person, but you have to be very specific. any interaction did work only after prompting "is-having-", "is-holding-", is-watching-" and so on. without prompting this way, i got body horror en masse. but interactions will mostly look the same like "first-person-pov". pretty lame.

anything else is pretty simple. i tried "biking", "hiking", "mountain-climbing" for my tests and never saw any face.

this prompt also can be used if you want a "clearly not looking at viewer" scene.

and at last something for fun: load any realistic nsfw checkpoint and prompt "alone-in-living-room-watching-tv" together with the prompt above and NO character extra, get a batch of 20 images get generated, you yourself should sit very close to your monitor or tv and now click that "generate" button and watch closely......

good old times when flashgames were still a thing :D

Until next time....

"Crowds or instructions for a lot of people"

Many months ago, when i started my journey for prompts, i found out that

"crowd"

is basically one single person. well, not even close to one single person, but you can think of it that way.

you can prompt anything using "crowd". SFW and NSFW.

"cheering-crowd", "dancing-crowd" and so on. Anything works.

And to get the best out of it, prompt

"surrounded-by-a-(insert any prompt here)-crowd-of-people" and "public"

You can even generate multiple crowds, basically an overload, if you prompt "overcrowded-crowd" together with any other prompt. but you will only see a difference when using high resolutions.

Until next time

"Nipples this time"

Had this on my todo list for a really long time and had a really hard time to find a solution.

Getting "see-through nipples" without the help of any model. Prompting something with nipples generates bare breasts every time. Normally, you would prompt "nipple-slip", but even that generated bare breasts. And that's not what i had in mind. This one was a really frustrating one, because i never had any success for a very long time, but well, i wouldn't write an article if i wouldn' have found a solution... And here is this (honestly dumb) prompt:

"nipslip".....

It works on any checkpoint, but strangely not on "real dream". Even with weights, i did not get it to work there.

Now, let's come to the point:

This prompt generates nipple-slips and see-through nipples, sometimes "accidential" nipple-slips or embarrassing nipple-slips (look at their faces. happened at some images).

If you just want to see what it does, don't prompt anything else, but if you want more, it's working on any character lora so far and together with other prompts (naked-apron looks nice with it ).

And now have fun

"Characters and places"

Did you know that you don't need to add special clothes to your character?

If you know which places your character can go, your checkpoint will instantly let them wear correct clothes. For example "school-interior" or "classroom-interior" get school-uniforms, "home-interior" mostly are every day clothes (sometimes, they wear only underwear. be careful with that), "office-interior" office-clothes and so on....

And for cases when you want some places outside, prompt "visit".

For example "shrine-visit", you get special clothes like kimono or dresses mostly, or if you wanna go shopping, prompt "supermarket-visit. You will get everyday clothes or sometimes everyday clothes plus apron.

There are thousands of other places for a "visit". This article would become tl;dr if i would write everyone of them down. You can explore other places on your own. Just be creative. :D

That being said... Until next time

UPDATE: prompting different countries generates different outfits. and that extra-specific prompt generates very detailed interiors or places. there is no need anymore for other things to set in detail.

UPDATE: "bathtub-interior"... no need to add the usual prompts like "partially-submerged" or something else. every checkpoint knew what to do and how it should look like. sadly, you have to prompt like a maniac in negative on anime-checkpoints. resolutions up to 1280x1024 did work very well

"Bombshells"

I already published a prompt using "bombshell" on one of my articles and during testing it, i found out that this prompt can be used for more:

"bombshell" alone generates a blonde, attractive woman

"bombshell-body" generates an ultra-erotic body

"bombshell-appearance" and/or "bombshell-look" generates sexy outfits. these both had sometimes differences when i used them as single prompt or both together. appearance and look is mostly the same, but some anime-checkpoints handled them different.

"bombshell-aesthetic" generates everything what the name says, but those girls mostly looked pretty skinny.

"bombshell-breasts" generates voluptous breasts. fun-fact: sometimes, when i forgot to add "1girl" or "1woman", more than just 2 breasts got generated pretty often on realistic chckpoints. "total recall vibes" :D

"bombshell-pussy" generates vaginas pretty similar to those "god-tier-pussy" loras that you can find here on civitai

using "bombshell" on male loras instantly turned them into a female version. no need to lower the weight on any lora.

on furry character loras, in my case i tested it on pikachu and other pokemon, "bombshell-body" worked mostly. here you have to play with the weights.

Until next time...

"Physique or simply muscular bodies"

Many times i had seen people uploading loras that generate muscles on any character. But an easy prompt can do that trick too and is saving you space on your harddrive.

"physique" generates a simple muscular body.

But simple is lame. We all want more. Always!

Tested on anime and realistic checkpoints:

"huge-physique" generates extreme huge muscles

"dangerous-physique" is more for women and generates curvy, erotic muscles (don't know how to describe that in another way than that)

"immature-physique" is not what you may think when reading "immature". It only generates a lightly muscular body

"cool-physique" generates characters with muscles that are highly detailed and "cool" looks.

"mega-physique" generates overpowered hulk-like characters

"super-physique" generates the perfect mix of ideal muscles, abs and bodies (eventually prompt "superhero" in negative as some checkpoints generated them)

"unique-physique" is like gacha. you will never know what will get generated

You can add many other things for physique, but mostly it will end up in looks only. Tried many others and results were really boring.

If you find something that can be added, feel free to post it in the comments...

Until next time...

"Magic mirror"

Saw that someone published a model of this and just thought "let's try it!". Well, here the prompt:

magic-mirror-showing-captured-girl-inside, in-nature

Replace "girl" and "nature" with anything you want.

Works best if you don't use any negative prompt or embedding to get best results. I recommend to use it on realistic checkpoints. Anime checkpoints got confused on how to depict that captured girl correctly.

Until next time...

"Kisses ....without a freakshow"

Based on my last news and its content about creativity, here a perfect example for this topic:

someone really said that you can't generate images with 2 girls kissing. well, anyone would prompt "kissing". but just let's not do that. instead we prompt something that i found out months ago that describes a kiss way better:

"smooching".... "2-girls-smooching"

prompt this and you won't get anything freaky... works on every checkpoint.

Stapfschuh said:

and from here on,i just want to welcome my old followers to being my new followers. even if its just the silent way...

Yes, moved over to see your great prompts :-)

looking for a style pretty similar to "born-to-die"? look here post #64722 or just prompt
in positive: youngest-youngster-cutesie
in negative: bulging-snap-fingers, fat-tits
use brackets as they work best and only anime checkpoints.

one thing i always wanted to recreate using only prompts was that "girl in a box"

"girl-trapped-inside-a-large-box" as a base. "large" prevents deformations and strange mixings, that's why.

prompt sizes of your character, f.e "little-girl-trapped-inside-a-box", and/or extra content of that box, f.e "little-girl-trapped-inside-a-box-filled-with-flowers".
you can choose any character you want. anything works. this one is just an example.

if you are one of my followers from civitai and you moved over to this place, don't hold back to tell that to everyone! with your help, this place will become very lively. i will do my very best to do the same.

This thread is a collection for anatomy. If you need help for hands, limbs, joints, fingers or the whole body, try some of the following:

physically-abnormal:2, joint-inequalities:2, limbs-inequalities:2

put it in negative like posted.

EDIT: add this if you have some of those loras that refuse to generate a good-looking body:

physical-inequalities:2
-----
This one looks and feels really badass and, honestly, works exactly that way. It is very, very forceful.... But let's come to the point:

Put it in negative without anything else: "gruesome-inhumanities:2"

As you can see on the preview-image for this article, i tested it on my favourite "test-subject" "Naru Kotoishi" to see if i get hands. And i got them. Nearly perfect without lowering the weight.

Tested it this time with furry-characters, too, but i highly recommend to not use it on those characters. It forces human hands, human skin or human faces on them, since this negative-prompt is clearly only made for human.

Sadly, this one seems to not work on SDXL. I did not even get 1 good image in any checkpoint. Most of the time, those checkpoints generated something "human-like" only on purpose. It seems that this prompt is really just way too badass.
-----
I was searching for this prompt that could generate a perfect body in positive prompt without getting something seriously dumb-looking. The moment you look at it, most of you will have that Marty McFly meme in mind:

"bodily-equalities"

Keep your negative prompts empty for testing. Sadly for it, for you it will be trial-and-error when it comes to weights. Some checkpoints did only accept numeric weights while others only accepted one single bracket. Mostly, my checkpoints stopped generating colors or other things after i used more than one bracket.

I tested it very successfully on SD, SDXL and SD XL Turbo (Turbo makes no real difference for me, but i was curious).

And another warning: since it has "equalities" in it, every time you use "realistic" in anime-checkpoints or realistic checkpoints in general, you will see many mixed races and skin-tones. This is one thing that isadly can't change. Just put in positive or negative what you prefer.

Sorry, I forgot ONE super important point:

Prompt clothes!!! :D
-----
This time, you have to use a counterpart, sadly. It works without it too, but results aren't as good as with it.

positive: (((ordinary)))

negative: (((extraordinary)))

Tested on SD and SDXL, but only SD knows what to do. XL still is an unfinished piece of s... ...

Tested with lively prompts and on loras. Got very good results. Just a bit more details would round up everything on some checkpoints, but that is in no way a hindrance.

Until next time...
------
Because i was searching like a maniac for new prompts for everything, i totally forgot the best effective way to "fool" your checkpoint for "perfection" that i found months ago.

It is really absolutely simple. Just think of what you want perfect and put it in negative with "no" or "not" at the beginning. For example perfect hands:

"no-perfect-10-finger-hands:2" plus "no-perfect-10-anatomy:2"

Or if you want perfect eyes, just type "no-perfect-eyes" and set a weight if you want.

Or if you want a perfect body without any deformations, type "not-physically-perfect" and add a weight if you want.

You can really have everything you can imagine get generated this way.
------
First of all: this one can generate good hands, but only on a few images and i didn't test it on hands, only on anatomy itself.

With this one, i wanted to find a way to tell my checkpoints that "anatomy and limbs" are important. And as you may know, prompting "anatomy" will get you, well, something.... :D

And if you prompt arms or legs, you will mostly get super deformed bodies on purpose.

Now let's get to the point:

(physical-components) in positive without any negative prompt for anatomy and only 1 bracket. Using more brackets didn't change anything, not even on loras. Tested only on "overexcited-lively-having-fun", which, unprompted, generates the creepiest horror.

You have to prompt a location or some clothes, because if not, you will see robots/androids most of the time.

Until next time....
------
physical-defects:2, unhealthy-deformed-joints:2, unhealthy-hands:2
------
i don't know why i didn't come up with this in the first place, but it needs really only one prompt for good or perfect hands.

deformed-joints:2

put it in negative like always.

i have tried to improve this prompt to "force" very good hands on bad trained loras:

unhealthy-deformed-joints:2, unhealthy-hands:2

works insanely well. please test it and tell me your experience with it.

and here a third one that i had to test with "lively overexcited having fun" that normally generates only horror but not with these prompts:

physical-defects:2, unhealthy-deformed-joints:2, unhealthy-hands:2
------
put these, like always, in negative:

physical-defects:2, unhealthy-deformed-joints:2, unhealthy-hands:2
------
5-funny-looking-fingers:2

Put it in negative without any embedding and other prompts. Had this idea after testing some loras that still gave me headaches even with those prompts from my last article. This prompt "forces" SD and SDXL to generate 5 fingers in a most perfect way. Most perfect? Yes. I got 99% perfect hands in SD, but SDXL had its hiccups with this prompt and always "cut off" one half finger.

Tested in quiet poses and "overexcited lively having fun". In SD it works perfect, but SDXL sometimes generated some "octopus-fingers".

UPDATE: was testing this on some new loras that were uploaded the last few days and found out that

"(10-funny-looking-joints:2)"

has a higher effect on many loras that are trained on just a few images or always the same image. Helpful if you want to keep original style of the character without lowering the weight.
-------
Was trying to prompt a child getting headpats in POV. And at first, those results were total horror, but after some trial-and-error, i got it working.

At first those prompts for headpats: petite-little-child, head-patting-pov, happy-about-head-pat

And here negative for those POV-hands: defect-pov-hands:2, defect-pov-fingers:2, unhealthy-deformed-pov-joints:2, unhealthy-pov-hands:2, elongated-pov-joints:2

I just used every negative-prompt that i published here and put "POV" in them. :D
-------
Why a "but" this time? Because this prompt can give you a very hard time if you use any weight on it or describe anything else of the body. Firstly, the prompt:

physically-bizarre-inequalities

As you can see, there is something that you have already seen in one of my articles. This prompt here was at first a totally mix of trial-and-error on this model: Naru Kotoishi (Barakemon) - v1.0 | Stable Diffusion LoRA | Civitai

I could not get even one finger since months without an overload of prompts on this model. But with this prompt, i finally could see correct hands. And i wanted a single prompt to get everything generated like it is supposed to look.

And now to the "but".

The moment, you put a weight on this prompt, your checkpoint WILL answer that with horrible deformations on purpose. And if you describe any part of the body in negative, you will mostly get a freakshow as an answer. It happened to me the whole time on every checkpoint for SD and SDXL.
------
This prompt did i test for months and it is based on ALL my findings. And in this article here, i will share those findings that can help you for your first time with this prompt. I have negative and positive prompts for you. Here's that prompt:

"snapshot"

But let me explain something first:

Sadly, my checkpoints are "overtrained" because of that prompt. The moment i use that prompt, i get instantly best hands and everything else i prompted it together with "snapshot". It has become a main trigger on my checkpoints. That's why i didn't use it anymore, but still decided to share what i found with the help of it. And now let's get back to the start of this journey:

The most important counter-prompt for snapshot comes first: "photography-in-sight". Put it in negative. If not, you will get only images where those characters are taking pictures. Happened to me a lot. And that's why i said my checkpoints are overtrained, because i don't need that prompt anymore.

"snapshot" alone gave me wonderful hands right from the start, but only with quiet poses.

After some quick thinking, i just added "staged" at the beginng. "staged-snapshot" resulted in way better images and hands BUT because that was not what i wanted, i gave my checkpoints a hard time with horror-images. That was the time when i put together "overexcited-lively-having-fun" and even "staged" images had not even 1 perfect finger with just those 2 prompts in them. I did what everyone does to limit bad hands. And here is that "collection" that resulted in wonderful hands and bodies:

(((unrealistic-behaviors, unrealistic-growth-forms, unrealistic-body-functions, unrealistic-postures, unrealistic-types-of-movement, unrealistic-shapes)))

(civit has some hiccups again and changed that color to grey... sigh)

I reduced the number of those negative prompts over the time and now the only working "multi-negative-prompt" for "snapshot" is "unrealistic-shapes".

And because a snapshot is not a snapshot if "painted", i added "lifelike-realistic" to get far better images and nice shapes. But because i always want more, i created "the best prompt" you can have to get the best out of your images: "the-best-of-the-best-of-everything"

It generated, as the name suggests, everything in "best", but because this is way too much cheating, i banned this prompt.

Well, and now i am back to the present...

Feel free to test them together or alone.

Until next time....(that was way too much text for today....)

UPDATE: because i ignored this prompt for a really long time now, i never even thought about, using other methods with it. I just thought about what you would do in reallife with unwanted photos. You delete them. And so i made a new negative prompt that gave me very good results only for "snapshot":

"completely-ruined-snapshots"

No other negative-prompt and no weight is used. It gave me awesome results with my "lively"-prompt. It seems that "snapshot" will stay my companion now for a long time, because i only need 1 negative- and 1 positive-prompt.
-----
The reason for this was the negative prompt "chaos", And what is one counterpart of chaos? Symmetry. Well, not recommended to prompt that. Your checkpoint will mostly give you symmetrical stuff followed by a big sigh (laugh all you want. my checkpoint "something" generated this comical sigh-cloud multiple times). and what is the countepart of symmetry? Asymmetry.

(((asymmetry)))

with weights helped the most to get instantly results. To see what it can, i removed "snapshot" and tested it with my lively-prompt. even with weights.
Tested successfully on SD and SDXL :D

UPDATE: i was testing "skydiving" which usually is ending up in a "human spider". I tried to fix that and found a simple way: (((asymmetry, asymmetrical-shapes)))

it does NOT(!!!!) fix the whole body if you just prompt skydiving. You have to be more precisely in your positive prompts to help your ai on looks.
-----
This one was bugging me for months, because i found a pattern to get best bodies, but couldn't find a way to turn that into a working prompt.

Every time i used "reallife-realistic" or only "reallife", i got perfect hands. It worked on loras, too. But i didn't always want that realistic look on those loras.

The point is, the trigger for those hands was "reallife".

Today, i was playing around with "nonsense-logic:2" in negative. It works somehow, but on loras, it's just naaaaah.... But then i had an idea. Why not put 1+1 together?

"reallife-logic"

Put it in positive and leave your negative empty.

When i tested it, checkpoints didn't like numeric weights again. I got octopus-hands and -bodies the whole time on purpose, so i decided to use brackets again.

If you simply wanna test it on your checkpoint, 1 bracket is enough. Use more an your checkpoint will be super-duper-mega-pissed and..well, most of you know that Upload from my checkpoint. xD

If you wanna test some more lively poses, 2 and maximum 3 are enough.

The same goes for loras. No numeric weights. Here, i can't help you, since you must find what works best on your own. Just play with those weights until you find out what works best.

Until next time......

it's working on those mixed checkpoints, too... After i made some batches, those checkpoint learned quickly what it means....
don't use this prompt on photo-realistic checkpoints! i simply forgot to delete it and my checkpoint was mega-pissed and generated human horror the whole time and i was just wondering what happened until i realized it...xDDDDD
------
(young) in positive. 1 bracket only if you use it without any lora, more, if you use them.

As it is more specific on physical appearance, my checkpoints knew that everything has to look "healthy" and generated everything nearly 100% perfect. Sometimes, because i didn't prompt anything for detail, 1 finger seemed to be missing.
------
Simply put "life-size-body" in positive.

I tested it on very high resolutions up to 1280x1024 and "overexcited-having-fun" together with "crazy-flexible-pose" and the result was always a perfect body. Not even the slightest deformation.

And to see if it has its limits, i tested it with children, since most of the time, small bodies are the hardest on every checkpoint. And it still worked. Even after i left my negative-prompts blank.

BUT! If you wanna use this prompt, i highly recommend to prompt chlothes, too, since checkpoints, the second they read "body", end up generating naked bodies.

And here the second part, that happened during testing:

If you put "flexible-deformities:2" in negative, it completely stops any deformation instantly. Well, this idea popped up, because of "crazy-flexible-pose". and after i added a theme-park-scenery, that "overexcited-having-fun" ended in a crazy mix of limbs suddenly.

UPDATE: some realistic checkpoints mistake this prompts for some kind of doll and generate "life-size-dolls". if this happens to you too, simply put "real-person" in positive and "doll-like-body" in negative and it will work again like its supposed to.

1