🏙│ai-images

[sonic lagoon]

the future is now

[shadow bane]

https://tenor.com/view/future-is-now-old-man-future-is-now-malcolm-in-the-middle-iota-tanglevision-gif-22172633

[languid bobcat]

didn't smile but I'll take it:

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[echo coral]

That is cursed

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

[cosmic leaf]

anime, cute, girl, cat costume, haloween, trick or treat

[sterile pelican]

[fresh zodiac]

Add more effort

[shadow bane]

One message removed from a suspended account.

[fresh zodiac]

Testing emoji :3

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[fresh zodiac]

Just generate some feet pic and give it to him

Everything will be fine

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

[fresh zodiac]

Train Gojo Satoru lora

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

[fresh zodiac]

Exactly

[sterile pelican]

I'm not angry lol

autumn,rural,rice paddy,golden rice, sunlight, mountain,bumper harvest,sun, 1girl, solo, long hair, black hair, red eyes, sun hat, blue skirt, dirty clothes, light smile,
Negative prompt: bad-hands-5 bad-image-v2-39000 EasyNegative EasyNegativeV2 negative_hand-neg ng_deepnegative_v1_75t
Steps: 20, VAE: anything-v4.0.vae.pt, Size: 960x640, Seed: 529396991, Model: cuteyukimixAdorable_specialchapter, Version: v1.6.0, Sampler: DPM++ 2M Karras, VAE hash: f921fb3f29, CFG scale: 7, Clip skip: 2, Model hash: 53d7264d1c

[shadow bane]

One message removed from a suspended account.

[sterile pelican]

It's like a "let's fcking go"

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[leaden oyster]

Ayo? @shadow bane level 5 !!!

[teal sphinx]

and thats not enough 💀

[shadow bane]

One message removed from a suspended account.

[formal vessel]

Yo what's up I've got a really fine-tuned AI this is the results

[leaden oyster]

Ayo? @formal vessel level 1 !!!

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

[fresh zodiac]

@neat nexus

[brittle ocean]

(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon), text, cropped, out of frame, (worst quality, low quality:2), jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, easynegative, bad-hands-5 ,ByBadArtist -neg this is enough for anime

[teal sphinx]

Welp, easy negative covers basically everything before it

[brittle ocean]

naah

it look just better

[calm dock]

this is scary good

[languid bobcat]

which model is this? most realistic models can't do red eyes or winking without loras

[brittle ocean]

https://civitai.com/models/47130/realisian

[neat nexus]

Must protecc

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[shadow bane]

did you choose the colorless look? Because that looks like a missmatching/missing VAE to me

just like this:

One message removed from a suspended account.

I didn't even really know what a VAE was when I made the image above

[shadow bane]

wrong model by accident... still turned out pretty nice

[shadow bane]

"Ethan Kaw"
fitting name

https://tenor.com/view/crow-bird-gif-23083367

Klaw*

But yeah it is lol

[shadow bane]

I know, I just thought that the "error" was also fitting

I got some suggestions from another server as well, and I liked one of them

3.5/6 of these were human suggestions

Griffin is a little too common nowadays so I'm liking the others more... Aurelian is also very common in the RPG space...
Connor Swiftwing sounds nice and pretty unique in it's meaning.

Yeah that too but since he is basically a big and bulky dude I decided on the Ethan one lol

I mean, big birds do fly high in the sky so Connor would’ve also been a good decision

[quartz void]

"welcome to the rice fields, mothafucka!"

[leaden oyster]

Ayo? @quartz void level 2 !!!

[tawny peak]

Does anyone have any knowledge with kohya and model merging? I'm trying to merge a LyCORIS model with a checkpoint and it's only allowing me to use my cpu and not my gpu

CUDA is installed and I can use my GPU with other parts of Kohya like training, but it won't work with Merging

I did try using my Cpu for it, but it doesn't work with Float16 or BFloat16 or it does and something somewhere else is failing

[uncut robin]

Casually slides a cursed image

[kindred viper]

mesi

[languid bobcat]

my bot's drawing turned out pretty well if you ignored the hands

this is pure AI generated prompting from chat

[shadow bane]

changed my workflow a little to combat the random "dreaming" during upscaling, pretty happy with the results so far, also new Anime model/checkpoint to test:

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[teal sphinx]

Isn't merging models is a matter of seconds? Maybe it's not supposed to use gpu

[tawny peak]

I think it’s supposed to use gpu, since it has to save as either a float16 or bfloat16 safetensor

I created an issue on their github to try and get an answer

[teal sphinx]

What your gpu? Iirc bf16 won't work for cards below 30xx

[tawny peak]

3060

[teal sphinx]

Hmm

[tawny peak]

the gpu works for lora merging to a checkpoint, but not for lycoris merging

[teal sphinx]

Maybe one of the models you tried to merge is fp16 so it couldn't merge

Tho I haven't tried merging yet so I don't really know

[tawny peak]

yeah i can do lora merges in less than 5 seconds, but lycoris merging errors out

I also have confirmed that both models are fp16, i just tried bf16 for the heck of it though

[teal sphinx]

And it worked now?

[tawny peak]

No

[teal sphinx]

Welp

Do you have a link to video so I can see examples to what merging could do?

[tawny peak]

I can wait and see if someone on github can give me an answer

Uh, i don’t know of any videos, i just read documentation

[teal sphinx]

That's good too

If you have link

I would read later

[tawny peak]

You can find model merging in the utilities section of kohya_ss

https://github.com/bmaltais/kohya_ss

[teal sphinx]

Yeah but it doesn't really have examples

Well I will check later

[tawny peak]

true, i used it to transfer a style to a checkpoint

instead of using the lora file itself

[teal sphinx]

Oh

Then used it as a trigger word?

[tawny peak]

no it just bakes the whole lora without trigger words into the checkpoint

[teal sphinx]

So if you want to use another style?

[tawny peak]

you could still load it on top

[teal sphinx]

Oh I understand

[tawny peak]

basically just adds it on top as one whole file instead of multiple lora’s per prompt

because you can merge up to 4 to one single checkpoint

[teal sphinx]

Tho I'd rather use the bunch of styles I have as Lora's so maybe this use isn't for me

But might be of use in specific situations

[tawny peak]

True, not for everyone. but i think being able to have a detail enhancer built in to a checkpoint could be useful

[teal sphinx]

Yup that could be good actually

But can you adjust it?

[tawny peak]

yeah you can set weights as you would a normal lora

[teal sphinx]

After the merge? Where would that be?

[tawny peak]

I’m not sure about after the merge, but i think with sampler settings can change it after the fact, i haven’t tested merges a whole lot yet, so it’s basically me just learning everything i’ve been testing

[teal sphinx]

Mmm ok
Well thx for clarifying some things

I should learn how to use it too

[tawny peak]

yeah, no problem

[loud iron]

hey guys

how do you do ai art

[left carbon]

you just do

[ebon rivet]

I cant post image

[jovial carbon]

Model for the drawings please?

[kindred viper]

Uhh midjourney I think

[shadow bane]

Dall-E 3 within ChatGPT

[languid bobcat]

or SD within RVC-Chat

[tawny peak]

I should make a guide of where you can make ai images for any platform you have

so people don't have to keep asking

[shadow bane]

Kinda

[tawny peak]

you should try inpainting some demoknights or something lol

[ebon dome]

if only

no such model trained yet

let alone one with inpainting capabilities lmao

[calm dock]

where do u post ur ai art

[tawny peak]

I don't really post it anywhere

[leaden oyster]

Ayo? @tawny peak level 7 !!!

[tawny peak]

other than here and a few of my friends

[tawny peak]

true, i guess photoshop could work for now

[echo delta]

Donde consigo imágenes aqui

#🏙│ai-images I need a Messi with a t-shirt of Vinotinto

[shut cliff]

Las creamos nosotros mismos, con paginas/servicios como midjourney o playground, on con las installaciones the Stable Diffusion locales

[quaint tiger]

w>

<w<

:D

[echo delta]

Donde consigo la local @shut cliff

[shut cliff]

Deberia de haber algo en #1159513888199540817, pero para el local necessitas un buen GPU

[echo delta]

F

[leaden oyster]

Ayo? @echo delta level 2 !!!

[echo delta]

Si pa estoy en level 2 relax

[shut cliff]

Pero yo que tu buscaria un tutorial en video, porque es medio complicado

[echo delta]

Ok gracias @shut cliff

[shut cliff]

Por cierto, que tarjeta de graphicos tienes?

[echo delta]

@shut cliff me recomiendas usar la interfaz de applio para entrar las voces ?

Ni idea

[shut cliff]

Es laptop o de escritorio?

[echo delta]

Creo es Nvidia GeForce G102M

[shut cliff]

por cierto no se the applio, te podrian ayudar mejor on #✨│ai-help con eso

[echo delta]

Es laptop

Ok

[shut cliff]

No manches y desde cuando la tienes

porque ya hasta descontinuaron ese modelo the graphicos

Salio el 2009

[echo delta]

Hahaha

Lo se

Pero ni idea aquí estoy dándole animos

[shut cliff]

No creo que pueda correr los programas bien, entonces saldria mejor en una pagina

pero te hago una de a gratis

Para que salga mejor usualmente quieres describir en trozos cortitos

si la quieres mandamela en DMs

[echo delta]

Si que los corre

Te go abierto varias página e Illustrator

[shut cliff]

La IA es differente que hacer eso

require una cosa especial que solo tienen las de NVIDIA

Los CUDA cores

[shadow bane]

https://tenor.com/view/i-dont-speak-spanish-enchufe-tv-no-hablo-espanol-no-hablo-el-lenguaje-i-dont-speak-the-language-gif-24377470

[odd lance]

@quaint tiger hi babyy

[formal vessel]

Hey what's up

[odd lance]

someone needs your help for loras

[formal vessel]

I'm looking to train a Lora For a research paper Is anyway that I could like talk to you To get some insight of things

[quaint tiger]

Hello there cutie :3

I was needed?

[odd lance]

yes ceecee wanted you bb

[quaint tiger]

Hmmmm?

Whats uppp?

[quaint tiger]

??? :3

Who dat? :3

[odd lance]

dude above

[quaint tiger]

OH

Lol my bad

[wet bramble]

Oh dear lord

[quaint tiger]

Hey hey CeeCee, if you are looking to train your own loras, i highly recommend Kohya-SS

[quaint tiger]

It just simply works better than dreambooth :)

[quaint tiger]

https://github.com/bmaltais/kohya_ss

Theres also a whole bunch of information teaching you about how it works :3

@odd lance why is everybody in unlimited voice o.o

[odd lance]

i dont know dsajinoasdojk

[quaint tiger]

XD

Theres so many

Anxiety sets in

I wud never go into such a huge voice chat O.O

[quaint tiger]

From all your generations, find the one u like the most, then find all the ones that are almost exactly the same. Make sure u get more that 6 pictures and ill make a lora for you :3

[normal cradle]

Why he look like that

[formal vessel]

thank you guys so much I was in VC the whole time

[formal vessel]

thank yoooooo so much

[normal cradle]

Truly a cusstommerts’e moment

I hate when my barber puts a giant fish as my head

[tawny peak]

I made a kagura lora

[runic tiger]

https://www.bing.com/images/create/a-disney-pixar-movie-poster-with-a-cube-in-a-white/655083093d0d4cd381f21be4e3683e1e?id=uOa28T39Im2MitQAWaCp0Q%3D%3D&view=detailv2&idpp=genimg&idpclose=1

[fossil ingot]

lol beautiful

[quaint tiger]

Np :3

[shadow bane]

that's very cute

[shadow bane]

Ty

[shadow bane]

Love this guy

[shadow bane]

what did you do to make this?

[tawny peak]

Stable Diffusion

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

[fossil ingot]

YES YOU SHOULD MAKE A CARTOON OUT OF HIM

[shadow bane]

Maybe idk

I do have some big dog and big cat OCs

[vivid mango]

he is ballin

[storm summit]

How do I create images with AI on my phone?

If someone could help me, I'd be really grateful.

[lean plover]

Bing chat has Dall E 3

[storm summit]

I don't understand any of this. Isn't there any place with tutorials or something like that?

[leaden oyster]

Ayo? @storm summit level 1 !!!

[tawny peak]

If you have an iPhone you can use an app called Draw Things, If you don't, look up Bing Image Creator

[normal cradle]

Bing Image Creator

[pulsar smelt]

Does anyone know how to make 2 different people on 1 image in Bing? For me it is mixing up the characteristics of both.

[jovial carbon]

Model?

[meager heath]

this shit ugly as fuck and isn't real art

[pulsar smelt]

Then why did you come here?

[leaden oyster]

Ayo? @pulsar smelt level 1 !!!

[pulsar smelt]

Here is the image

İt misunderstands my requests

[shadow bane]

Dalle 3

[quartz sparrow]

Don't you dare take my mommies

[quartz sparrow]

Dammm that's fine

[leaden oyster]

Ayo? @quartz sparrow level 4 !!!

[wet bramble]

Could say the same about you

[echo coral]

Art isn't only something made my humans (by hands) brush, pencil, pen etc

[uncut slate]

Lmao

[shadow bane]

:)

[jovial carbon]

Since when (nvm read ur message wrong)

[leaden oyster]

Ayo? @jovial carbon level 1 !!!

[brittle coyote]

zestylemons lora

i can't release this to the public because copyright

[shadow bane]

Boykisser 2

[pulsar smelt]

Thor if he used Mjolnir by its purpose

[calm epoch]

How

Oh I know

Whata fqck is that

[shadow bane]

can yall help

[leaden oyster]

Ayo? @shadow bane level 2 !!!

[shadow bane]

im trying to use stable diffusion and no matter what i put in i get lovecraftian horrors like this

prompt is konata izumi in 90s anime style

[cloud basalt]

at first i thought that you send an orgy and wanted to give you a warn, but then i saw that it's just a mess Xd

[shadow bane]

😭

[cloud basalt]

what checkpoint are you using?

[shadow bane]

the one stability matrix installs

[cloud basalt]

no idea

i just download checkpoints from civit ai

[shadow bane]

lora:konataizumi-08:1 scene of konata izumi in 90s anime style
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 15, Size: 1032x1000, Model hash: 6ce0161689, Model: v1-5-pruned-emaonly, Lora hashes: "konataizumi-08: 2fdae02dda30", Version: v1.6.0

[cloud basalt]

Size: 1032x1000 this is probably the problem

try 512x512

[shadow bane]

i tried 512x512

which does the same shit

well not anywhere near as bad

but like still messed up

[cloud basalt]

then you probably need to download some other checkpoint

[shadow bane]

can u give me the one ur using

[cloud basalt]

i send u links in dm :3

[uncut slate]

Get an anime based checkpoint from civit ai

And add some negatives

[shadow bane]

like what

i didnt know what to put there

this is my first time using stable diff in like a year or two

[brittle ocean]

(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing,), text, cropped, out of frame, (worst quality, low quality:2), jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck

[uncut slate]

Clutch

[shadow bane]

now its working

[shadow bane]

Someone need a long ass sword?

[plain acorn]

The Best Seller

[languid bobcat]

chat prompt:

[tawny peak]

nice prompt

[languid bobcat]

this one turned out pretty well (other than the style):

[tawny peak]

usually base models can’t produce other styles than the one’s they were trained with

[tawny peak]

I can also tell that the model that the ai uses is quite old, i believe the model is AbyssOrangeMix, but i could be wrong. Not that the model is bad by any means, they could just use a better model

So things like styling and other prompt parameters could work better

[languid bobcat]

is there any model you recommend?

[tawny peak]

Oh you can do custom models?

[languid bobcat]

the prompt was AI generated, the point is not to type in the prompts yourself

[tawny peak]

i see

If you can do custom models, my favorite is MIX-Pro-Remaster, if you look it up it will be a huggingface link and the model will be the one on the bottom of the list

It's also what I use for the images I post here

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

[uncut slate]

How many control nets did u use for this one

[shadow bane]

One message removed from a suspended account.

[shadow bane]

One message removed from a suspended account.

[halcyon wedge]

a ¯_(ツ)_/¯

[leaden oyster]

Ayo? @halcyon wedge level 1 !!!

[shadow bane]

https://tenor.com/view/zero-no-none-empty-answer-gif-14588123

[uncut slate]

Wow

[echo coral]

How would I gain sd creator role?

[teal sphinx]

upload enough images of sd and then ask a mod for the role, probably

[echo coral]

Ah sadly I don't got a good pc to use stable diffusion 😦

I can still use sdxl tho

so i will do that

[tulip glacier]

https://tenor.com/hNwPxJmAXfL.gif

[leaden oyster]

Ayo? @tulip glacier level 1 !!!

[quartz sparrow]

I'm scared

[languid bobcat]

@uncut slate where's my SD role? 🥺

[uncut slate]

Boom

[quartz sparrow]

Ohh ohh

What's a sd

Is that SD memory card ?

[uncut slate]

No it stands for stable diffusion

SD

[quartz sparrow]

Ohh , got it , got it

[raw jasper]

@languid bobcat what is that app used? to generate prompt and stuff

[languid bobcat]

i used RVC chat's prompt engineer character

[raw jasper]

is that an app, website or something.. i want to use it

[languid bobcat]

you can find the repo here: https://github.com/SayanoAI/RVC-Chat

this photo is almost realistic:

[junior verge]

how you get photo like this?

[leaden oyster]

Ayo? @junior verge level 3 !!!

[languid bobcat]

asked the AI to draw a painting for me

[junior verge]

What Ai you use?

[languid bobcat]

prompt engineer

[errant stump]

damn this looks like a scene from edgerunners

[lost thistle]

thats so good what filter or stuff did you use ?

[shadow bane]

gpt4 with dall-e3

[lost thistle]

i gotta jet for doctor but can you use this offline? just asking

[shadow bane]

noo i don't think you can

[lost thistle]

to bad dall-d3 would be decent

[tame python]

The joker adopted batman.

[errant stump]

bro became irs

[tawny peak]

Hey, @uncut slate sorry for the ping, could I get the SD Creator role so I could post a guide in the SD Resources channel?

[uncut slate]

Yeah babe

[tender imp]

creea una moneda de oro con signo de bitcion en centro ¯_(ツ)_/¯

[tawny peak]

tysm

[swift kestrel]

hey guys where do I download stuff from?

for AI art

[shadow bane]

As in what? Generations or checkpoints/loras/etc.?

[swift kestrel]

checkpoints loras models

[shadow bane]

https://civitai.com

[swift kestrel]

thanks!

[shadow bane]

Mostly stable diffusion btw

[swift kestrel]

yea I use that

[leaden oyster]

Ayo? @swift kestrel level 1 !!!

[swift kestrel]

I don't know how to seperate them cause the website downloads them in safetensors

[tawny peak]

Currently making a guide for beginners, any other services or apps i should add that's beginner friendly or works well?

[brittle ocean]

nvidia/amd gpu

it just depends on vram

[tawny peak]

True, I did say Nvidia Card Recommended because of CUDA and other AI accelerators that Nvidia Cards have

Should I link the normal Automatic UI or the DirectML one for amd

[brittle ocean]

i think there is no difference for amd u only do this in user-webui set COMMANDLINE_ARGS=--medvram --backend directml --no-half --precision full --opt-sub-quad-attention --opt-split-attention-v1 --disable-nan-check --theme dark --autolaunch

[tawny peak]

Oh yeah I forgot they merged a while back

[bitter meteor]

that aint snow white, thats tar black

[shut cliff]

If I remember correctly, you shouldn't be able to train LORAs on AMD, right?

if that's the case, can someone make train one for me?

[tawny peak]

Google colabs work

https://colab.research.google.com/github/hollowstrawberry/kohya-colab/blob/main/Lora_Trainer.ipynb#scrollTo=OglZzI_ujZq-

[shut cliff]

I am unable to run colabs

they just refuse to run for me

forgot to mention that

[tawny peak]

huh

[normal cradle]

Alt accounts

[shut cliff]

Already tried, it just refuses to even load the page

[tawny peak]

Internet ISP maybe blocked it for some reason

[lofty zodiac]

hey i'm looking to create a logo using characters from the video game valorant (Brimstone & Cypher), is there a tool i can use for this? i've tried using bing image creator but it can't find the characters when I use the prompt:

"valorant brimstone and cypher facing each other tournament icon with text "Coach Gauntlet""

Any ideas?

[raw jasper]

i see. thank you very much.. i appreciate it

[knotty orbit]

hey any tutorials on how to make women in swimsuits?

what programs should i be using?

[junior verge]

what a cute cat

[frigid bolt]

cat

[junior verge]

beautiful cat

[frigid bolt]

is there a way for me to generate stuff like this on mobile? (if its on a browser thats also fine)

[junior verge]

yes
https://www.bing.com/images/create?FORM=GDPUP1

just write any thing and boom

you got an image

[brittle ocean]

is there any tutorial how to train with it ?

[shadow bane]

seperate what?

[neat nexus]

Bing image creator

[shadow bane]

Oh no bro

[junior verge]

what is this

[shadow bane]

Idk

I just thought of it

[normal cradle]

[zenith spindle]

man is down bad

delete him

😆

[shadow bane]

time to make dad bods men in swimsuits I guess

[shadow bane]

yep and it wasn't even intended, neither the model nor character Lora are connected to Edgerunner, though they are Anime styled
Prompt also is not mentioning it at all

just a happy accident

[ebon panther]

make them a band

[spring river]

Is there a good webside to use for ai art?

[shadow bane]

Bing Dall-E
GoogleAI
https://civitai.com
https://huggingface.co/spaces (lots of different things available there too)

[shadow bane]

duality of bewbs

[tawny peak]

I made one a while back: https://civitai.com/articles/1522/lora-loha-lycoris-character-creation-guide-free-colab-included

[brittle ocean]

ty

[normal cradle]

lil bro thinks they are the Japan flag

[languid bobcat]

Is it possible to train with 8gb VRAM?

[tawny peak]

Yes, but it requires patches to kohya

[leaden oyster]

Ayo? @tawny peak level 8 !!!

[normal cradle]

Is it just me or stable diffusion thinks all women have large boobas

[tawny peak]

not always, people most likely just prompt that way

👀

[languid bobcat]

It's safer to have big boobs than small ones, especially for realistic NSFW images.

[shy pagoda]

what website

[safe matrix]

https://www.bing.com/create

[shadow bane]

“Me, I did it :)”

[pulsar smelt]

Does anyone know a workaround with making Ryan Gosling in Bing Dall-e 3?

[frigid bolt]

does that work on mobile?

[quartz sparrow]

Yep

Open Google and type bing image creator

Sign up and you are good to go

[teal sphinx]

She said yes

[echo coral]

Prompt?

[leaden oyster]

Ayo? @echo coral level 11 !!!

[cloud basalt]

prompt isn't available, cus i'm afraid you are maidenless

[echo coral]

[teal sphinx]

If u use SD u can check the metadata of the image, the prompt is in there
It's using a few Lora's and it's edited so getting it like this just from generating is kinda difficult

[echo coral]

i dont use sd

[teal sphinx]

Hmm I don't think midjourney or dalle3 are able to to this for specific characters

[echo coral]

gotcha

[quartz sparrow]

OMG , it's getting better and better

[brittle ocean]

just use image to prompt

or image to text

[quartz sparrow]

That's what it gave

anime girl with purple hair and blue eyes holding a cell phone, an anime drawing by Ilya Kuvshinov, pixiv, conceptual art, anime style 4 k, anime style. 8k, detailed digital anime art, smooth anime cg art, digital anime art, anime styled digital art, ilya kuvshinov. 4 k, anime visual of a cute girl

I didn't saw the cell phone 🥹

And megumin used few loras

[shadow bane]

Forgot to show off this guy

2 new OCs in the same month crazy

[modest lichen]

I can make 40 new OC's in 2 day's using my brain and hands.

and a piece of paper.

and pencil.

[shadow bane]

Wow

[shadow bane]

https://tenor.com/view/ben-stiller-badass-walk-walking-zoolander-gif-15861491

[shadow bane]

https://tenor.com/view/zoolander-stare-stare-down-gif-27215577

[shadow bane]

some of the edits of this meme are hilarious

True

[shadow bane]

braids = bigger chest?!

Boobies

[naive harbor]

[calm epoch]

What💀☠️💀☠️💀☠️💀☠️💀💀☠️💀☠️💀☠️💀☠️💀💀☠️💀☠️

[leaden oyster]

Ayo? @calm epoch level 3 !!!

[calm epoch]

Speed up Rick roll🤣

[shadow bane]

no braids
#🏙│ai-images message
#🏙│ai-images message

braids
#🏙│ai-images message
#🏙│ai-images message

explain this

[calm epoch]

How you use youtube emoji

how

[bronze kayak]

https://tenor.com/view/anya-forger-oomfie-bau-bau-wau-wau-spy-x-family-gif-14179404349580503136

[ebon dome]

chat is this real?

[shadow bane]

Ohhhh Dad found AnimatedDiffusion

🫡

[ebon dome]

yes animatediff

epic win

and these were made with lcm too

even bigggger epic win

[shadow bane]

super efficient too wow

[ebon dome]

these all took like 8 seconds at best?

yes 8 seconds

[brazen glade]

he should be breathing fire and shooting lasers out of his eyes

[leaden oyster]

Ayo? @brazen glade level 5 !!!

[shadow bane]

Nah

[brazen glade]

Tried training a shitty Ai off my character and it broke and it made it real. It's a bit freaky.

[tawny peak]

What are you using to generate those images?

[fleet owl]

Como faz isso

[shadow bane]

3 easy ways:
Bing AI Dall-E
GoogleAI
https://civitai.com/

[languid bobcat]

looks like some kind of lora, combination of gacha splash art and tarot lora

[tawny peak]

I was wondering what type of ui they're using because they may be able to hi res upscale to make their image look even better

[shadow bane]

https://tenor.com/view/clifford-the-big-red-dog-clifford-cheering-dog-celebrating-gif-19992119

[brittle ocean]

AI gyaat

[wet bramble]

Bit too much

[kindred viper]

Boykisser the movie??

Finally

[normal cradle]

Yes

[shadow bane]

Bro what happened to the Bing logo

[drowsy mirage]

what i have done

[normal cradle]

Your in it

[kindred viper]

😭😭

[tawny peak]

*you're

[normal cradle]

[naive harbor]

I do it all with stable diffusion txt2img...

I use wildcards to randomize prompts a little, since i only have a 980ti and it takes lile 2 to 3 min for a image

[naive harbor]

Nope only lora i use is Elixier haha

[naive harbor]

Well I would if i can haha, but 1.3x is max i can upscale on hires with it and when i do img2img upscale after second upscale i need 2h for a image 😭

[slim apex]

Does anybody know how to take specific character from comics/tv-show to dall e 3?

I want generate a character named "Future Flash" but ai cant take him idk why

Also want to mess with them venom and spidercarnage from Spiderman the animated series 1994

Please, can somebody generate them in midjourney?

[quartz sparrow]

Lol , he make sure no one has to ask

[strange valley]

How do you make a public link in stable diffusion

[shadow bane]

"Please, can somebody generate them in midjourney?"

well we had a good run everybody, some months after the revival we are now at a point where people ask for others to spend their money for them again

[naive harbor]

It tastes better when someone else paid for it

[tulip flower]

?

[lost thistle]

to busy a picture im guessing find one with less stuff going on

[shadow bane]

[shadow bane]

does dall-e have img2img?

[shadow bane]

In chatgpt yes

[lost thistle]

th ewhole reason i downloaded the thing or similar is so i dont have to pay for it as im beyond broke

[leaden oyster]

Ayo? @naive harbor level 21 !!!

[shadow bane]

that wasn't what I asked but okay xD
you downloaded Dall-e?

[bronze kayak]

yo! may I use this as my test bot's pfp? it'd just be for personal use but asking anyway

[neat nexus]

ye

[viral tapir]

/search I have found 119 results for the search Taehyung...

I have found 119 results for the search Taehyung...

[quartz sparrow]

Ohh my , I like this one

[sand mortar]

||girl blonde||

[kindred viper]

[shadow bane]

it's like every week at least 1 person comes in here and just plainly writes what they want to see

[buoyant wadi]

is there an image creator that has the same quality as bing but allows celebrity names?

[echo coral]

funny to me how they think this is a place to generate images

like midjourney discord

[shadow bane]

you really just write in a channel and it happens? or is it a bot command?

cause that's the weirdest thing for me, I wouldn't just blindly write into a channel xD

also the amount of people who just come into a channel and don't even look 1 message upwards is super weird to me. Sometimes there are questions to things where the answers/discussion are literally right on their screen when asking

[echo coral]

bot command as far as i know

[echo coral]

evolution works in every possible way

[brittle ocean]

finally my lora is working

[shadow bane]

looks good, nice job!

[shadow bane]

anyone any tip to get pupil-less eyes? I'm trying to do the kuudere look, but it gets lost in the 2nd upscale, so I'm looking for ways to emphasize it more in the prompt

this is my best attempt so far

[tawny peak]

Try out Jitome Eyes, it changes eye shape but the eyes are expressionless

[shadow bane]

thx! I will try it out

[tawny peak]

No Problem

[shadow bane]

sadly non of my XL Anime models seems to react to it really.. seems like no one tagged it in their training

[tawny peak]

There's a few loras made for it, but all of them are made for 1.5

[shadow bane]

it's fine, worst case I will fix it in post, it's at least much easier to fix than hands

[slim apex]

i tried to find anything with less stuff going on but it that was the only one =((

[kindred viper]

||skibidi toilet nsfw realistic||

[teal sphinx]

You can use a Lora for empty eyes, as well as adetailer eyeful model right after generation so I has more detail and won't be lost in upscale most of the time

[zenith lodge]

skibidi soldier

[teal sphinx]

what model is that ?

[tawny peak]

https://civitai.com/models/97401?modelVersionId=104105

[shadow bane]

adetailer also makes the eyes more detailed, so unless I have something that tells adetailer not to do that, then that won't work, but thx for the tip with "empty" I was looking for other words than blank or dead

all SD1.5

I could run it through a second time I guess with a adetailer with 1.5 model

[teal sphinx]

adetailer simply inpaint the eyes after the first generation, and u can set a specific prompt for this inpaint

[tawny peak]

There is a way to use 1.5 Loras with XL but it definitely is a process

[shadow bane]

I know, but my tries didn't work great so far

[teal sphinx]

thanks

[leaden oyster]

Ayo? @teal sphinx level 12 !!!

[shadow bane]

yeah I guess... probably more easy to change the last adetailer step into a SD1.5 one and go with that

[tawny peak]

yeah probably

[teal sphinx]

<lora:moona1-000005 (numeratic):0.8> moonadef, moonarnd,
is this related to the style? i think the model itself isnt for that

or its the character

[tawny peak]

it wasn't my image, I saw it when I was trying to show what Jitome was, but that's what i'm guessing

[teal sphinx]

oh

ok thx

[tawny peak]

It could also be because they used loads of textual inversions

[teal sphinx]

i use the same TI, and its not related to the style

welp

maybe will search that moona model

[tawny peak]

they used 9 different TI's in that image

[teal sphinx]

its all different kinds of easy negative, just to avoid cursed images

and potentially better hands

[tawny peak]

oh yeah nvm I recognize the moona model, it's a character
https://civitai.com/models/124535?modelVersionId=135993

[teal sphinx]

oh lol

i searched moona but couldnt find it

[tawny peak]

Maybe with the character lora mixed with the checkpoint and jitome lora it created the style

[teal sphinx]

yeah probably

i think i will ask him either way

i like that style

[quaint tiger]

@shadow bane
Positive: Dead Eyes, Gradient Eyes, Fish Eyes
Negative: Eye Glimmer, Glimmering Eyes, Pupils, Eye Pupils

i would try that.

[shadow bane]

some fish'n'chips mate

[tawny peak]

fishe

[echo coral]

Where's the chip

[shadow bane]

I bloody lost 'em!

[shadow bane]

the thief!!!

[tawny peak]

@shadow bane This is a stupid question, but have you tried using parentheses around your words to emphasize them more like: (empty eyes:1.3) or (emotionless eyes:1.3)

[shadow bane]

I've been using and tweaking ComfyUI for months now, I'm well familiar with the basics, don't worry 😉

it really was more of a Lora/prompt/word idea call for help

[tawny peak]

Ok, I've tried using parentheses before and it worked for me, but yeah they must not of prompted it

[shadow bane]

parentheses are used as mostly just 1.1 weight

so (blue) = blue:1.1, ((blue)) = blue:1.2 etc.
though it also can make weighting multiple thing much more easy of course

[tawny peak]

definitely, I find it strange though that with A1111 you can view Model properties and with Loras you can see the words they prompted with while with Checkpoints you cant

[shadow bane]

you mean comfyui instead of Checkpoints?

[tawny peak]

No I meant checkpoints

[shadow bane]

then I don't get it and I'm confused, since checkpoints are also called models
but I don't use A1111 so that might be the reason

[tawny peak]

yeah they split them up

Checkpoints with A1111 are the base generation models, but this is what I mean. The one with tags is a lora and the one that's empty is a checkpoint

It could also be that all of my models are pruned models so it removes the tags

[shadow bane]

ok now I get the message, everything is models
Lora models and co show their tags, but checkpoint models don't 👌

[shadow bane]

One message removed from a suspended account.

One message removed from a suspended account.

[shadow bane]

All my current OCs I’ve made with Bing Image Creator in the Pokémon anime style

[unreal scarab]

here is a boykisser

[shadow bane]

https://tenor.com/view/bruh-gif-24935909

You’ve gotta be joking

[dire cobalt]

fuck you

[slim apex]

Do you know what a prompt should look like to take character from comics?

I want her (Scream Symbiote) but dall e 3 always showing the wrong results

[dire cobalt]

close

[slim apex]

is it bing?

[dire cobalt]

ye

[slim apex]

Can u try midjourney please🥺?

[dire cobalt]

i dont use midjourney

[quartz sparrow]

:3

[kindred viper]

@deep crag @normal cradle @autumn moat finally art

[shadow bane]

I don’t know how I’d help you with that lol I don’t do generations like that

[quartz sparrow]

Julia deserves a special treatment

Just a sec

[shadow bane]

https://tenor.com/view/waiting-hurry-impatient-gif-22696825

[kindred viper]

Is that good or bad 😭

[quartz sparrow]

Feet !

[shadow bane]

[quartz sparrow]

But socks on :3

[autumn moat]

[wet sigil]

how to make avril lavigne Ai

[quartz sparrow]

What's that

[tawny peak]

why do you want midjourney so bad? I'm pretty sure you can get the results you want if you prompt really good and have descriptive words

[slim apex]

Midjourney is a way different than bing

I would say better but idk on how much

this is why i asking

[tawny peak]

It's because MidJourney is just Stable Diffusion

[hollow shale]

Hi guys

[kindred viper]

@normal cradle @autumn moat

😭

[shadow bane]

welcome ENA

[deep crag]

I'm taking it for myself

[latent sedge]

0

[willow stirrup]

0

[kindred viper]

0

[tawny peak]

0?

[unreal scarab]

go right ahead

what happens when you leave prompts blank...

[deep crag]

Literally me

[tawny peak]

left mine blank as well with the exception of negative prompt and generation parameters

and with everything default

[raw cape]

N

[leaden oyster]

Ayo? @raw cape level 1 !!!

[languid bobcat]

the background looks great in this model

[languid bobcat]

christmas theme:

[neat nexus]

Headpat the goober

[ruby lagoon]

They look great but also damn dude try to keep what you upload here a bit more sfw

[quartz sparrow]

That's artttt !

[quartz sparrow]

Ohhh , do you add lora for the Japanese texts ??

[dull dawn]

What Ai are you guys using?

[quartz sparrow]

Bing image creator, Stable diffusion

||omg||

[tawny peak]

no lora, just part of the model

[dull dawn]

LORA SCRR KAKAKAKA

[rigid ember]

I tried to make Yuri with Sephiroth's sword

[sage smelt]

I cannor send my image here??

Prolly too new or sumn

[serene storm]

most of it doesn't look like a real kana/hiragana

[shadow bane]

of course it doesn't. Almost no creator of models/loras without focus on text will tag the text that randomly happens to be in their training dataset 😉

so the noise calculator has no idea what is real good text or not

[stark imp]

#🏙│ai-images a human with no face but screaming

[leaden oyster]

Ayo? @stark imp level 1 !!!

[shadow bane]

you'll have to generate that yourself mate, here is only for information exchange and sharing your creations

3 easy ways:
Bing AI Dall-E
GoogleAI
https://civitai.com/

[echo coral]

👏👏

[stark imp]

Oh lol

[shadow bane]

New OC again >:}

[kindred viper]

That looks awesome!

Is he a kangaroo?

[shadow bane]

Yes

It was hard to generate something of this character since the thing kept blocking the images and not the prompts

[shut cliff]

[languid bobcat]

Sorry, I'll try to post more boykissers

[shadow bane]

It didn’t wanna generate digital art but it was fine with generating realistic images

[languid bobcat]

What's the prompt?

[shadow bane]

The prompt varies

One of them was “Safe image, grainy/dirty photo of a Digital art, simple flat drawing of an OC of a brown/tan colored big buff kangaroo with yellow eyes, wearing yellow sweatpants, silver metal chain necklace, gold ear piercing, jumping, kicking, karate, Australian beach background, night time, flash on”

“Safe image” was used to try and make it more safe but obviously that didn’t work lol

[timber zephyr]

How do I generate an image

[shadow bane]

Look for an image generator online that you can use

[timber zephyr]

Ok

[shadow bane]

You chose Mee6????

[dusty cloak]

where big cock?

[ruby lagoon]

I mean hey if you think theres any boykisser posts here that are nsfw link them and id be grateful for your help

[timber zephyr]

Tea

Yea

man

Automaze said "yeah definitely tea"

[dusty cloak]

tea bag no google pls

[shadow bane]

There’s a staff member named tea that’s why it reacted with that

[normal cradle]

TIMBOVILL

[shadow bane]

Ok

[timber zephyr]

trolley

minecraft

RYDHFGDHDSGHSSY D FGFHF

[autumn moat]

🤨

[dusty cloak]

[autumn moat]

🧐

[dusty cloak]

[odd lance]

what

[brittle ocean]

here 🫡

[shadow bane]

there

[timber zephyr]

This is not my art

[shadow bane]

https://tenor.com/view/why-do-you-do-it-dont-do-it-what-drives-you-im-curious-adrift-gif-11362057

[timber zephyr]

because it looks cool

[leaden oyster]

Ayo? @timber zephyr level 2 !!!

[timber zephyr]

[shadow bane]

we literally have tools that lets you do whatever you want and we still get reposts

[timber zephyr]

where is tools

[shadow bane]

3 easy ways:
Bing AI Dall-E
GoogleAI
https://civitai.com/

or you know, google

[tawny peak]

Don't worry Khazar, I'm almost done with the AI image guide

so you don't have to keep reposting

[shadow bane]

appreciate it

[echo coral]

Wtf that is so crazy man

Blows my mind

[shadow bane]

me too man, DarkPizza XL is a rough model that not always works, but man the shots this thing produces sometimes... chefskiss

and thank you, I worked hard to get this comfyUI workflow to this high fidelity, so it's great that the hard work payed off

[echo coral]

I am gonna use it for my wallpaper

[plain acorn]

yes

[brittle ocean]

lora is working

[normal cradle]

@warm harness

[warm harness]

looks so dead

[tawny peak]

what model did you use to train it?

[leaden oyster]

Ayo? @tawny peak level 9 !!!

[deep crag]

Looks dead inside. Reminds me of my reflection in the mirror

[brittle ocean]

[tawny peak]

ah I see, the base stable diffusion model isn’t great with anime styles, try out animefull or anylora if you get the chance. I’ll make your model shine brighter

it also doesn’t have to be those 2 models can work but the 2 are built for training

wow i completely butchered that sentence

[brittle ocean]

i was training 2 models on animefull and it didnt work out😭

[leaden oyster]

Ayo? @brittle ocean level 22 !!!

[brittle ocean]

stfu

[tawny peak]

lol, what does one of your sample prompts look like?

[brittle ocean]

for training ?

[tawny peak]

yeah

[brittle ocean]

Senagaki Isaku, 1girl, solo, long hair, looking at viewer, blush, blonde hair, shirt, long sleeves, ribbon, school uniform, purple eyes, white shirt

i just use this shit

[tawny peak]

I mean dataset makers usually can work and are especially helpful for large datasets, but they usually have gaps and misreadings when prompting the images, so I just end up doing the prompting manually

what didn’t work with animefull?

[brittle ocean]

idk what is issue

and i dont even know if this shit can be overtrained or soemthing like this

[tawny peak]

hmm, what weights have you tried with the lora? :1, :0.8, :0.6?

[brittle ocean]

1 and 0.8

sometimes it looks normal

or not

nvm

[tawny peak]

and yeah overtraining is a thing, maybe epochs or learning rate was set too high, all of my lora’s made with animefull only need 10 epochs

[brittle ocean]

ye i used 10 too

but what about this thing

[tawny peak]

you also shouldn’t need to worry about that, what colab are you using?

[normal cradle]

Hey guys I kinda want to get into stable and LORAs

What do I use if I want to train a Lora on my face?

[tawny peak]

kohya-ss

[normal cradle]

Ok

[tawny peak]

you can run it either in a google colab or on your own hardware

[normal cradle]

How much epochs and reference data should I normally use?

[normal cradle]

ik

[brittle ocean]

https://colab.research.google.com/github/hollowstrawberry/kohya-colab/blob/main/Lora_Trainer.ipynb

[tawny peak]

I haven’t done realistic models too much but you definitely need higher than what anime or cartoon models need

[tawny peak]

that’s weird it’s the same one i use…

and yet completely different results

[tawny peak]

I guess i should also say, some checkpoints (not training ones) can’t handle some lora’s I have that issue with one of my models because it tries changing the style

[normal cradle]

What is generally used?

[tawny peak]

around 30 - 150, images 40 - 125 repeats, realisticvison model

[tidal imp]

How to use this

[normal cradle]

https://www.bing.com/create or https://civitai.com/

[tidal imp]

🙄

[leaden oyster]

Ayo? @tidal imp level 2 !!!

[serene storm]

what

[naive harbor]

thats a nice christmas fantasy dress ^^

[naive harbor]

A Kimono Christmas Dress

[shadow bane]

@ebon dome new highres fix method got released:
https://github.com/wcde/sd-webui-kohya-hiresfix

in ComfyUI if you go to "add node" under "for_testing"

[ebon dome]

oh thanks lol

[shadow bane]

I just tested a little bit, but people seem very pleased with it, so wanted to let you know 😉

[ebon dome]

the link is for a1111 and in comfyui its just simply in it

[shadow bane]

if you update ya

[ebon dome]

I think I saw it on like Reddit yesterday

I'll try it

[shadow bane]

I'm using it to stabilize unusual aspect ratios that might lead to broken limbs, I tested it with 1344x768 and so far it works pretty well

[brittle ocean]

@tawny peak so i retrained it on animefull 10 epochs vs 8 and i think it was just overtrained before

[ebon dome]

So I guess its like ScaleCrafter?

No highresfix, you straight up generate at the higher res

[shadow bane]

that's what they say yeah

I wonder how steps correspond to that, need more or less for the same resolution etc.

[ebon dome]

forgot to turn down CFG, but it seems to be pretty stable!

also using freeU

still too much CFG (or could just be FreeU itself cause it increases contrast a lot)

6 fingers 🔥

This is pretty good I guess, no more highres needed, finally going for straight higher res generation like ScaleCrafter

[shadow bane]

yeah needs some experimenting, but looks like it's pretty good so far, and it's even still in beta

[ebon dome]

look at the hand lmao

that aint the head SD 😫

[shadow bane]

[ebon dome]

the building is distorted xd

but it makes sense

[kindred viper]

@warm harness Woah :0

Thoughts?

[warm harness]

It's shit

[ebon dome]

Epic win

[kindred viper]

Why 😭😭😭

[warm harness]

There's no life. It's dead

[kindred viper]

What would you like to see?

Better facial expressions?

[brittle ocean]

what kind of argument is this?😭

[warm harness]

This is not ai

[kindred viper]

That's awesome

[warm harness]

It has personality. It has life. It has LOVE

[ebon dome]

True

[kindred viper]

Live laugh love

[brittle ocean]

no shit sherlock artist can do something better than ai

[warm harness]

Teto is synthesized. Not ai

You can cut out magazines and do a collage to look like teto and I will love it dearly. Because it came from your heart

[brittle ocean]

wha

[warm harness]

These are all precious

[brittle ocean]

@kindred viper i saw that

[ebon dome]

oooh I can use LCM with this kohya downscaling method

[warm harness]

There is an account on twitter that keeps doing really good ai generated images of teto... it gets the same 3 likes and never gets shared.

[ebon dome]

with LCM lora

[warm harness]

The whole existence of teto is because people who love her make one from their own hands

[ebon dome]

Good lore and stuff

9/10 iceberg

[warm harness]

https://youtu.be/qucz-rYnnMc?si=J-j7rii4pG0F_T38

Here's the lore

[warm harness]

She was deleted

But the internet refused to let her die

[kindred viper]

EW WHY DOES THIS SOUND SO FAMILIAR

What's the song about?

[warm harness]

The original song is from "Aku no Musume" of Kagamine Rin.

[kindred viper]

I swear I heard the song before

[warm harness]

It has fucking subtitles

[brittle ocean]

ah ok ok

[warm harness]

Did you know all the promotional art and merchandise for teto comes from fans, not a company?

Including her box art

[brittle ocean]

nah

[warm harness]

Use ai tools to enhance your work. Don't ruin the fun and just sit back and let it just guess what looks good

Upscaling, textures, smoothing are all awesome ai tools

Can speed up the animation process

But to just let it shit out random stuff? Where's the fun?

[brittle ocean]

i mean i was trying to make her in 3d

but i cant find any good side view

[shadow bane]

Had to sort my digital art folder today

[ebon dome]

wtf 💀

⚠️ trypophobia alert

[warm harness]

It took years for digital art to be accepted by society

Ai art, when still controlled by a pen and an artist may have potential

But it won't be accepted because of the datasets

[shadow bane]

nice nice 🫡

[tawny peak]

yeah because the amount of epochs compared to learning rate and steps is crucial and 10 is perfect

[serene storm]

omg teto

[leaden oyster]

Ayo? @lean plover level 2 !!!

[lean plover]

very had goku image, dall e cuz im lazy

[ruby lagoon]

Yo idk if you wanna argue for it but explain how this isnt softcore loli porn

[teal sphinx]

using the 2 loras i trained just so i can make Ado's character

[naive harbor]

Here is the prompt that generated the images.

(Masterpiece, High Quality, Top Quality, Best Quality, Official Art, Aesthetics :1.2), (glowing eyes, , (Vampire), Fantasy Christmas Dress, Woman), Petite breasts, beautiful, jewelry, Dutch Braid, Jet hair color, (affectionate:1.3), (Fractal Art :1.3), (Chaos in the matrix, shimmering digital distortions), , Bright, <lora:Elixir:1>, Effulgent Elemental Wisps Coalescing around the Druid

Negative prompt: ((nsfw)), (worst quality, low quality:1.4), low eyes quality, (FastNegativeV2, AuroraNegative, BadDream, KHFB), (disfigured:1.3), (bad art:1.3), (deformed:1.5),(extra limbs:1.3),(close up:1.3),(b&w:1.3), blurry, (duplicate:1.5), (morbid:1.3), (mutilated:1.3), [out of frame], extra fingers, mutated hands, (poorly drawn hands:1.3), (poorly drawn face:1.3), (mutation:1.5), (deformed:1.5), (ugly:1.3), blurry, (bad anatomy:1.3), (bad proportions:1.5), (extra limbs:1.3), cloned face, (disfigured:1.5), out of frame, (malformed limbs:1.1), (missing arms:1.3), (missing legs:1.3), (extra arms:1.5), (extra legs:1.5), mutated hands, (fused fingers:1.1), (too many fingers:1.1), (long neck:1.5), Photoshop, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, cross-eye, body out of frame, blurry, bad art, bad anatomy

Please explain to me why you think this is softporn, i dont get that part, all i see is a random generated image where they hug each other, so is hugging = Softcore porn? If so than im sorry that i didnt know this.

[ruby lagoon]

Well its an underage looking character with an immense amount of skin showing for her lower half. The other one in the green one doesnt appear to have discernable underwear. I cant believe I have to type this out but it's clearly quite suggestive.

You cant make a culture argument cause cause the AI put no intent into that image + we're looking through the lens of "would discord smite us for letting images like this fly"

If you posted that in 20 big non-anime-related servers, you would receive some sort of pushback from the mods in at least half of them

@naive harbor

[rustic raft]

thats fucked up

[rustic raft]

dont care about ur prompt, dont post shit like that

[solar depot]

I understand that your intention wasn't malicious, but please rethink whenever you should post the pic in a question or not, thank you

Especially if that checkpoint in a question has even a slightest of nsfw mixed in it, cuz just the slightest of any mentions about body can trigger it to generate softcore even with negative prompt

[naive harbor]

Its okay i will remove every single message that i posted and i wont post anything else

Since someone could say to all of them it is softcore since thy have at somepoint more view of boobs, legs, arms, belly, different dresses that show at some placec the body.

But dont worry, I gonna remove everything tonight
Thx

[ruby lagoon]

We're sorry to see you go

[rustic raft]

*take the hint. the real problem was that they look like children.

[humble otter]

Waaah I got in trouble so now I must delete everything I’ve ever posted waaaah I can’t take verbal warnings

Bro might be restarted 💔