#π₯ο½anime
1 messages Β· Page 217 of 1

Are those the eyes you wanted
no π
its not the eyes, its the makeup
Cute style
i need like, dark eyeshadow and stuff but its not giving me that π
(Unhinged Digital Artwork:1.3)of(Masterpiece:1.3)Miku Goth Girl Crazy clown cheerleader at Gotham high school, (by Artist H.P. Lovecraft:1.3), CGSociety, ArtStation, (Emo Art:1.3), (Dark hue:1.3), Fine neo art style, Trending on artstation, focused, well shaded, 8K, 3D unreal rtx resolution Realistic photo, holographic dark lights
i love those eyes
what about this makeup 
ok this is better
that makeup would be cute too π
i just need my AI to recognize makeup
i think i have the lora for this??
must have been one heck of a party!
Did you try increasing the weight

yeah π
:D
now she's crazy
FINALLY
i know π
YES
i have a wild card set up with about 50 eye loras
oh I have that
i hate it when it prompts those eyes 
i made my own wildcard for it 
wildcards are so fun
best part of AI
Can't you remove it
hmm where did I put them
i can yeah but i have to sift through which ones are specficially that shape 
But it should be in the prompt
and i dont want to completely yeet it either because what if it looks good to me later
its in my wildcard
But then you have to go find the lora

It you open PNG info, I believe it shows what the wildcard chose

oooo bet
found the culprit
one of them at least
temporarily yeeting it from my wildcard rn
still not it
i swear its prompted
and i put so many () around it π
Try eyeliner

bet
On pony I just do "eyeliner, makeup"

That's as far as my knowledge goes
oh
i cant run pony

my shitbox of a laptop would kill itself, i think
hmm what is pony? a model?
Should be the same ish
Pony just really wants the exact danbooru tag

Most 1.5 you can just wing it cx
hmmm
ok what is with the dress?! wtf
i still f with it
Pony diffusion on civitai, there's also autism mix (based on pony but has a more generic/stable style)
autism mix sdgcjgscsc
I still use that mix I made π

link em
this is cute, the make up is getting closer
I'm on my phone atm 
They should be easy to find they are very popular
it says XL on it π
this
ok mixing styles, with style piles stuff. wow

adds waifusion
ports to GIMP to adjust it
ok to tired
blah
dodgeball queen?

hmm overcooked?
to bed but not to sleep
ok thats good enough.. π
wholesome breakfast
anyone knows a xl dreambooth colab that works ):
namaewa? lora?
wtf is namaewa?
name name
self trained
ouhhh, what are the settings you trained?
uh i forgor i have the config tho
@bitter latch, what model are you using
Ok
oi should not have mixed the queso dip with chili with beans.. i'm regreting it
rawr, grumpy
hehe
Akira Amano
has been a fun artist to use with pony/autism/confetti
been running through this big list just for kicks
omfg finally xl dreambooth is working!!!°°
only took like 4 accounts ππππ
We Will hold a memorial for you intestines
Thanks
Why was it bugging
score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, (anime:1.3), 1girl, solo, cowboy shot, momo, from my hero academia, looking at viewer, white background, ponytail, black eyeliner, makeup, goth, eyeliner
Negative: source filmmaker, sketch, lineart, chromatic aberration, unfinished, blurry, monochrome, censored, abdominal muscles
It's the default pony model, called something like "pony diffusion v6 start with this one" in civitai
π₯Ί
Go for (child, loli : 1.5 ) neg : breasts π
With that ?
π
difference between dpo and normal models?
introducing the foxbike
2 eyebrows noooo
I lived
Hello, can you tell me any good anime model? I can't find any decent ones, and they're also quite bad, the hands turn out very disastrous in all the ones I've tried, and the eyes too.
Well done π
worth creating an account on that site?
I'd say worth since some models are locked behind account registration
Yeah the one ur using is because its marked as nsfw
umm what is that model
autismmix_confetti
π Yes i understood that part
It says mix of "Pony"? What does that mean
also why are all the preview art nsfw
the pony model is a SDXL model that was highly trained on pony/anime/furry artwork and is one of the best anime models currently
I am new in this part of AI art creation
And autismmix_confetti is just a better version?
it's also the best nsfw model, so that's what made it popular
I want to make mostly SFW art
it's better for ease of use, it's a lot easier to get good results from it
For a starter should be good then?
i make mostly sfw stuff, i use autismmix too
Which one of the 5 should i download then?
yeah, it's a really good model to start with
the confetti one is the one i'd recommend
Alrighty, downloading
i also downloaded Automatic1111 Sdui
For the generation. Have no idea if its one of the best or anything
automatic1111 is usually the best to start with
Oh, ok
And for Lora's i should use only 1 per generation?
Let's say i want ZeroTwo in her swim suit. I should use a ZeroTwo Lora and another one to complement?
depends on the loras, for a lot of stuff they're not even needed
For this example?
you'll just have to test the prompts with the model, this model should be able to produce zerotwo by itself
loras are basically to add in data that the model doesnt already know, or more refined data, so it depends on the results you want and can get from the model alone
i would start with just the model and learn how to get it to produce something close to what you want, then look into adding loras if you need them
How do you use both models?
So... About prompts, how can i learn to correctly make them
easiest way is to find images you like that have the prompt info in them, inside a1111 there is a "PNG Info" tab where you can dop an image and it will show the information if it exists
all of my images posted here usually have all the info
if they have the info in them, but most that way probably wont
Unless something changes recently
also i got a 4060 ti with 8gb of Vram, its enough right...?
you can also search on civitai, there's a lot of images there and most of them have the info
forge is good, but i don't know if i'd recommend it to start personally
it'll be slow
Do you know where to download the SDXL locally?
I have no other option...
Forge is more optimized for lower end cards
4060 ti is not lower end
It is
it is though

i live in brazil
4070 is mid range
especially for AI stuff, the more vram the better
so i should set it for midvram or lowvram
in the end, the difference is how fast it generates, you might end up running into 1+ minute generations, versus a 4090 that does the same thing in ~5 seconds
on the start prompts?
tbh, i don't know the best settings for lowend hardware π
that's if you are intending to use sdxl model on automatic1111 webui for 8gb vram
if you are intending to use it on forge like cat bot said, then you don't need any argument
imma be using this model: utismmix_confetti
no, in your commandline_args
autismmix is sdxl
added
You dare say my name

Did you look in the model description
It probably helps
yes
pony based models should have some extra tags in the prompt score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_anime
How do you do it so that it doesn't hurt your hands? They come out very badly compared to what you do, please answer me.
oh shit
i put them at the beginning, some people put them at the end
?
ppl seem to be using animagine XL for the arts
hands are always bad, regenerate until they're good or use inpainting to fix them
there are thousands of models and people use most of them, it comes down to what style you want and which model handles it best
Yeah
Can you send me the model please?
I would like an anime style that is the most used, that is, of high quality.
that doesn't mean anything
pony based models are probably some of the most used, but it's not a style
and quality is subjective at best
Look, it makes my designs very bad, have I configured something wrong?
what's bad about it? what model are you using
Tail flower 
look how well the partner in the back turns out
ok, so with sdxl the resolution should be 1024x1024, 768 at least
and like i said above, the pony based models, including autism, need those extra prompt tags
sorry, what labels?
Great it stuck in interrupting
no grabby

1024 by 1024 really takes a toll on the pc
You should try forge
It's easier to run

Another question, how can I resize a resolution much larger than 2000? I get this limit, people make designs with great quality
nothing usually, the pony based models were trained with a mistake and need to use them
@versed jay upscale
in the same stable? is there any option?
I imagine forge will cut that time in half
If u help me to get everything set up i will download it
@vital raptor is installing forge any different from a1111
The GitHub page should have the steps
I have to look at them
Contribute to lllyasviel/stable-diffusion-webui-forge development by creating an account on GitHub.
There is a one button install available
My hands always turn out badly, even if I reload many times, any config to make it work better or something? @vital raptor
negative prompts?
you have none
they are REALLY important
just put something like 6 fingers, extra fingers, extra legs and etc.
@nova remnant just got forge
happy?
Should be the same looking ui

If I put that in, the images don't come out with fingers, always avoid showing the hands.
Try a 1024*1024 image in it
I'm not using autism, I'm using the default pony model so negatives work differently
in autism what should they be?
6 fingers, extra fingers, extra legs
My negatives are: source filmmaker, sketch, lineart, chromatic aberration, unfinished, blurry, monochrome, censored, abdominal muscles
tf is source filmmaker
Valve source filmmaker
Stuff for 3d animations basically
(what people use to make interesting overwatch videos)
What do I do then?

how many u generated
1 at a time, and several, and none of them had their hands
On autism I just used "3d, source filmmaker" and stuff looked fine
You don't really need much, the prompt is more important
strange, im trying it here
why is it generating things that are way too realistic close to what i want, which is anime based drawing
can you put a screenshot? and see the model and config? Let's see how it turns out pls?
Wait please
If you want more close up you should do stuff like cowboy shot, full body, close up etc
tf is cowboyshot
no but like
what does it mean?
wait hold up i think im cooking
also @nova remnant it did fucking half the time it needs to generate
also what is the prompt when i want the photo from the back of the person
just as you see in the image, it does horrible things to me
Hid the hands but its just what i wanted
try leaving only extra fingers and extra legs
prompt for that one?
It doesn't even have an anime style, not to mention how bad the hands come out.
change the damn resolution
I didn't type much lmao, was just trying to get the view
(anime:1.3), 1girl, solo, cowboy shot, dragon girl, dragon horns, dragon tail, dress, white background
damn bro
And all the score stuff at the start
btw tf does the anime:1.3 does
Means anime
cto as?
With 1.3 strength
1024 width, 1024 height
what does the parenthesis do
It's not very useful on autism to increase anime weight
You'll get more of that word in the image basically
lets try with viewer from the back
what parentheses?
it doesnt generate the girl from the back godammit
()
Try (from behind)
but where?
where u want the AI to have more emphasis in
@nova remnant does sampling method change much
where would it be correct?
They are all a bit different
fking heck eyes becoming all blurry and fucked up
negative prompt for eyes and feet fingers?
The lora thing is not clear to me what it is for, can someone please explain to me? I do not know how it is used
its for what the model has no idea what it is
complement to the model if you may
and since I use it, nothing appears in the lora section, even though I have lora models
you sure u put them in?
tf is that face π
do not worry about the signature π

if you're expecting magically perfect results everytime you generate something random i have some bad news for you
your loras have to be for the same model type you're using, XL loras only work with XL models, 1.5 loras work with 1.5 models, etc
it won't show them if they're not compatible, and you might need to restart your UI for them to show
What are the / for?
escape characters, otherwise the () get used for emphasis instead of keywords
wym escape characters?
It's used for the AI to not see it as increasing the weight of words but actually () written
oh, ok
what was ur prompt and model
Grabby 
How to get it done for you with such image quality? and with such a good anime style? what you put? Can you show me the config and the model pls with a screenshot?
How do tell the AI that i want a leggings but not latex one
cloth leggings maybe
Latex in negative
imma try it
Pantyhose or stockings should work too
How to get it done for you with such image quality? and with such a good anime style? what you put? Can you show me the config and the model pls with a screenshot? @vital raptor
nah leggings with latex in negative worked
all the information is in the image, drop it in the png info tab in a1111
the only thing not really shown is that i'm generating at 768x1024 and then using img2img to upscale x2 with 0.25 denoise
I don't understand, how is the information in the image?
because it is, drop it in the png info tab
what is the name when the person is not fat but not thin?
I still don't understand, sorry for not understanding the language, where do I look at that png thing?
artist tags
pony models know a lot of artists, so if you put in the names it will change the style
midriff is the stomach area
what does midriff mean
oh
also umm how did u make so the front area is showing but still using the sweater
@vital raptor
i didn't prompt for that, but midriff with a bikini top probably influenced it
oh ok
help me with this please?
Open (any clothe)
but you could do something like open sweater or something
Should work
Chubby or thick
But AI can be like: +40kg should be fine
chubby goes to fat
Good luck x3
thick thigs goes way too thick
Try decreasing weight, AI can't really be that precise sometimes
tried with +40kg
wait im tryin
ugly eyes strike again
@nova remnant 
heeeyyyyy
'ugly'
exaaccttllyyy
the eyes are indeed different
Looks like not the main character but also no one important from an anime
Lmao
her face is literally a V
it's all about style, some anime looks that way
Loooool
the whole thing is V
Is rescaling with img a program option? I do not see it

yes, use img2img tab and resize there
the ears going through the hat π
where else would they go?
@nova remnant I have no goth to show, been making questionable stuff
idk BELOW the hat

ποΈ
send in dms plz
^
im still trying on my goth girl
its just not coming out how i want
maybe because i have too many prompts/loras??
I'm just pressing generate on little prompt

im using some of my own wildcards to randomly gen hot girls rn
But what prompt do you put there to make the image bigger? That's what I didn't understand, I mean, what is the function of passing it through there? upscaled?
img2img tab
when you upscale
you enlarge the size
as well as up the quality
i think?
the hand π
but which section do you choose? general? unpaint?
the hand is not bad
it doesn't need a prompt
img2img and scroll down to here
my first that looks fine
says 'resize' right there
its horrible
she looks like she got her hand caught ina grinder π
@nova remnant did it turn good?
i think it looks pretty good 
no extra fingers or toes
no toes actually π
that doesn't appear to me
why does yours look like that 
maybe this is a problem for #π€ο½tech-support if its missing from yours??
because it should 100% be there 
you didn't like the cat ears, so i removed them
now its just a normal human π
yeah that's how that works
yeah ik
thats so cleeeaannn
idk how he does them perfectly
likely the model + negatives + highres + maybe inpainting??
Where should I give it? because no button does anything to me
hundreds of hours finding the right prompts/model/settings/artists combo, and having a fast gpu to run lots of images
plus some inpainting and upscaling at the end
fast gpu definitely helps π
i use a LOT of loras and models 
usually only inpainting for fingers and removing ears
mine is very slow tho
i dont use inpainting either cuz idk how
it always looks goofy when i do it
while i'm just random as fek
my prompts are mostly randomly generated with an extension
To start rescaling would it be here?
i mainly make anime characters but recently i decided i wanted to make original characters
i use some saved little style prompts that help me π
so i made some wildcards with my loras and im running those
i have over 2k loras so putting them in wildcards saves me from having to pick out which ones to use
yeah, wildcards are fun
Yes, but I say that to start rescaling, what is the correct button? because I don't know what everyone does
yesssss
generate
i recently found out about them and had a friend teach me me the whole process of making the text files
ive been having a lot of fun
I use an extension called Unprompted that basically does the same as wildcards but allows programming logic

wdym allows programming logic 
so like, i'll never get the 'midriff' prompt randomly added to a dress
thx
I have it randomly choose the type of clothing, full/outfit/build, 'full' clothing is something like a dress, jumpsuit, etc that covers the whole body, 'outif' is something like ninja/witch/pirate, more like a costume
:0
i need that
i cant make outfits for shit
and then the 'build' option will do a top, bottom, shoes, socks/tights, hats, jackets, etc
good?
slide what you use for outfits 
ill uhh
send you what i use for eyes

but some outifts or clothing pieces are only used if i'm allowing it to do lewds, like no reason to randomly prompt for panties if i'm doing sfw images
i also have a flag for child/adult so certain things don't get prompted for children
living with your parents is hard if u want to create nsfw images π
i need that
im so
so tired
π
of my characters looking young
@nova remnant did mine come out pretty?
just put 'child' in the negatives or something
sometimes it happens anyway, and the model/artists will influence it a lot
does it know yeogui?
π€·ββοΈ
nsfw one
yeah but it does it no matter the model for me a lot more than id like
if it was just every once in a while itd be fine but nah its a literal struggle to force ai to make adult women π
all of my models are anime based so
yeah
thats likely my issue
i didnt think of that tbh
maybe i should get a realistic model and use loras to obtain the anime style??
a lot of realistic models can do anime too, might not need a lora depending
might be worth trying though
I merge models of anime then add to it π
any good ones that you recommend??
also that, merging can get good results
i uhhh
i use about 50 models
i switch usually between every few gens but rn im just using one
ive never merged models before though
ok making a dragon goth girl. not easy. hmmms.
what exactly does that do 
Try dark sushi
i still ahvent even made the goth girl π
i have that one 
did those yesterday
if you're using that many models instead of focusing on 1 or 2, merging probably isn't the way
yeaaaahhh, i download a lot of ai stuff π
i use most of it but like
every single model handles prompts differently
only because i downloaded it
ye
wild mutation
ALMOST the leg is a lil glitchy. How do i make the leggings even more black?
@nova remnant this one came out pretty but still no goth makeup π
What does this prompt mean?? (m (mrtarou):0.8)
@nova remnant Can you help me with that one?
im not sure tbh, maybe add more weight or use a lora for it??
@nova remnant Can you help me?
(black lipsticks, goth:1.5) 
bet
On what
.
Things like that it's hard to tell AI, it is usually trained on tags so "black stockings" is your option
ok
You can try like [leather legwear::0.2] see if it works cx
That thing stops the model from using that prompt at 20% of the steps
It might get leather dark leggings and look like dark leggings

use negatives
already has extra legs
then add uhh
3 legs
4 legs
multiple legs
getting there
a bit too much tho
i want eyeshadow more like how you, @nova remnant had yours from last night
Are you doing just 1024*1024?
What are the other settings
Autism is usually pretty good at not having bad anatomy
I can fix her
fix her 
Try using the euler sampler with 30 steps
that one doesn't fail me
noice
Tyty
animagine
Can someone help me pls? I click generate to enlarge the image, and then no enlarged image appears, what could be happening?
any errors in the console?
if you're upscaling my image, it's going to take a long time and also might not finish if you don't have enough vram
it took over 3 minutes on my machine to run out of memory and give an error
and i have a 4090 and 64gb system ram
Since your gpu probably can't handle the vram usage, you can use the SD Upscaler script at the bottom under the scripts dropdown. Set the width and height to something your card can generate, like 768x1024 and then scale factor 2. In my example, I used the full size of the image because my card can handle larger chunks, but if you have less vram you'll want to go lower
using the SD Upscaler script I can do a 4x upscale in about ~5 minutes
8K is pretty π
finally figured out how to use this thing
but hmmm....
somethings not right
ok maybe I haven't figured it out

Is that one gud?
And did you try stuff like "large/thick eyeliner"
Just thought about it
Hi, hi! Guys, who have encountered such an error in the extension Civitai Helper?
This error occurs when trying to download some Models or LoRa`s
"sunrise in a lush valley paradise" π
yawns, I should really be in bed
good thing I got today off π
Not sure where she came from but.. ok
thuds as he falls over and goes to sleep
charming
fantastic
hellol3 V1.0
- some LoRas
π
δ½ δ»¬ε₯½οΌ

Yup. but I didn't want to. it just happened that way
Belly

yes!! Kitte!

hey is there someone can make this handsome goat into ai version
never played with furry :/
can make it yourself with a loraπ
made dreambooth
tried dreambooth
failed dreambooth
still has colab runtime so decided to just fuck around
Ok
But he look soo delicious eat
Ok btw who is lora
Did i talk to you
Did you?
lora is a slightly sexy mf
Well i go talk anyone that can make this anime goat have a ai version
Geezz
yes plenty of people can do it,but will they do it for free just for you?
Yay π
nope
hi @bitter latch can you please make this anime goat into ai version
blud's gonna start mentioning everybody on the server π
Anything for that goat
blud boutta ping the entire world
he shall dox every single person on this sphere ball of dirt
mhm
Just one person and not all
Hey guys! Can you suggest any 1.5/XL model with thick outlines, please? I wasn't able to get out of Anything 3/XL and some other models.
HMM .. maybe
I mean wait isn't it anime already?
there should be a lora that gens tick outlines
like this?
switches it up :D
π
@nova remnant its getting there π
It's nice seeing perfect hands for once lol
omg fortnite
Finally I found way how I don't suck in shooting games 
The same thing happens to me, nothing appears on the right, and it doesn't give any error message or anything.
i upscaled this image with the same method , looks good thanks π
isn`t "extras" working the same way ?
a lora?
The anime is called beastars and the goat is from season 2 and that goat character named Pina
Like this
thought so
Soo if the this ai generate can make Pina perfect from beastars character from the season 2 π€ but the wolf from beastars season 1 is perfect from e6ai Sfw btw
look realistic anime protagonist character from beastars
thats from game and not anime
yesh
Stop sending gifs and gen more thighs
π
dragon maid > normal maid
i got thighs for ya
pick a number 1-8 cat

Black and white
In pony you can do "themes" it's pretty cool
Like black or white theme
π
i still cant figure out pony
i got a model for it
but when i use it its like rainbow grain
(blacked /theme):2.5)
Can you generate something with it?
That's surprising
You probably need a vae
yeah, i can generate rainbow fuzz

pag = pog, improves image quality for sure
(built in into latest release of comfy ui)
indoors horse 
home carousel with horse
man what happened to her eyes?
furry day 
but if that Director Fury, then where his blindfold
well I did try to make it Anime-ish. yeah the patch didnt' show up. hmmm
don't really look anime
but look like furry
really look anime?
yep
ππππππππππ YYYYY
cute, i like this mesh of colors
How to describe lop ears?
lop ears seems to work fine
floppy ears?
@steady grail @vital raptor as an option. but so far I canβt βpersuadβ the model
treaten it!
if adding weight to it doesn't work then the model doesn't know what it is
or you are not persuasive enough
it know.
(lop:1.1) ears was enough for autismmix to do it consistently
could also try lop-eared or floppy ears
crap! what an intractable model!
charming
responded to you prompting a punch in the face? XD
The fact of the matter is that it doesnβt. these ears come out 1:25
eheheh
yep. it really sad!

alien
Hi hi hi hi
that come with upgraded alien
I'm tryna get baseline down rn

mf imediatly tryna ask personal info
dunno why you delete the "I'm new" part but that's a lil too late

Inside your walls
Not even in tinder people ask that that fast



she's cutting tomato
for dinner

oh wait you were referring to other guy
wait did he do smt in other channel that I missed?
he got unlucky that him stalking you got seen by someone
π€£
He really wants to know
Looks like some WWE mask

not sure where that cushion came from tho.. shrugs

How do I save the config in the stable diffusion? every time I open it the configuration goes away.
on the settings tab, scroll down to the bottom where it says 'Defaults'
How can I download the SD 1.5 model? I want to download lora for shampoo from the anime ranma, but I can't find the sd 1.5 model on civitai.com
@vital raptor
most models on civitai are 1.5, there's not just one
the lora probably recommends a model it works best with
I want to use this lora, but it doesn't appear with the animagine model or with autism, how can I use it? https://civitai.com/images/1029666


