#🍥|anime
1 messages · Page 79 of 1
Hell I might need those cheats fr fr
Her freckles are so dang cute omg 
Freckles is my weakness 
So next time Fruit hosts pow I need to make Miku with freckles to get 2 free votes 
Cheating abusing but smart 
Noted
What's your weakness?
Miku?
beachs and juice boxes
good thing I'm already ||modeling a beach||
I gotta make a sand castle,though
So many emotions
oh, love that beach one
Gotta prepare for Nubby's contest
suspicious
For secret society's banner
I SEE
||make an animated banner with Miku||
I need to upscale
@dull jackal You noticed this?
I noticed he was gone when I banned him 😛
I was eating my pear and opening a folder lol
also animating is fun
I don't have time for that I need to read manga and watch anime on my free time
Deforum can run in the background
I love the shadow and the shoes and forced perspective
(trying to find my folders LOL)
Changed some loras
I was thinking about making some fashion LORAs eventually
it'll fit right in with my plot
I haven't finished the shading on this yet
et
BUT look at his fabulous hair
That's one wild hair
It's actually a cropped close up of a larger sketch I was drawing
cuz he had super long hair that was blowing in the wind
Sounds familiar
It wasn't based on any anime or anythingl just drawing
It just reminded me what happens when I go outside
Rune G :O
I don't think I posted any images I posted to TouhouAI yesterday
HAHA MY HAIR, yea
it's long
YEAH RUNE G, fantasy RPG OK
Welcome to the darkest dungeon
nothing to do with ||Rune Factory|| at all
XD
Oh damn I didn't know about that game :O
more tests I WAs doing
good, good
CUZ WE'RE PLAYING RUNE G
Look at those headphones
It's funny how it interprets poses sometimes but I loved these ones
3060 time letsgooooo
your city is dust now, her work here is done
welp
Tis unfortunate. She's a bitchy Angel
I have a carefully crafted wildcard file thats just full of general and small posing details instead lol
Where’d did u get this picture of me
mostly...
stairs
wow, if I saw this somewhere there's no way I'd know it was ai generated
ah, I guess the hands still bit off
I guess someone looks a lil sussy
@dull jackal got a bot i think
you cant stop me from clicking on it and getting my free gift
so who broke imdb?
S i g h
Love those eyes
trying new ways to prompt since i'm using model that im bad at 
Almost perfect
sharp teeth
Bottom ones are fucked 
mmm nice
Do moooorrreee
Close on the teeth
that shirt is missing an o and the girl a glowup

whale tail
I love how I can almost see a knee in the first pic. It's like she's wearing a cosplay outfit, lol.
a little seed adjustment and i could probably change that lol

Also those hands are 
Sorry for the double notification. Discord wigging out on me
Yeah hands on hips poses tend to be really good with hands actually

And np
Thanks Strix!! 
@teal pivot nice v2 model

mfw I add black boots to hopefully get a full body shot
and instead it makes her sit
feels like a rare pose though
I feel I'm onto something here
Tbf that is nearly the full body... 
this is great
first pic has technopriest vibes
thank you, I'm trying something new away from gothic or dark fantasy and doing some technopunk, futurepunk, cyberpunk, dieselpunk, gearpunk(cogpunk) mix 😛
then there's the different dude stereotypes too, suave chiselled guy, boyish feminine guy, cold dark haired emo guy and .... uh....What happens when you spend time with a Tsundere for too long
Also...Fluffy, the pet dog
and for those paying attention, nightmare fuel
hmmmm this one is good
sorry for this late response lol, but it doesnt seem to be working properly, do you have any tips for it?
cute eh
very neat
would be curious about prompt on that, is hair thing a lora?
man is it possible a driver update made me avoid OOM?
I didn't used to be able to do 4x highresfix without tiled vae, now I can
i now train my first lora
re-l
Something just reminded me of back in 2019 when I first heard of and used AI for generating images for the first time--and it was for anime characters. It turns out I still have some of them on my system.
Back when everything was waifu this and waifu that, lol. Waifulabs created the image, and waifu2x upscaled it.
Terrible by today's standards, but it was revolutionary at the time.
Didn't know AI art was a thing back in 2019
thought it started around 2020~2021
Certainly nothing like it is today--I'm not sure it was even recognized as such back then. Really only for those who had a vested interest in developing such things, often with a market to match (like anime otaku.)
AI back then made like, mush eldritch horror xD
I found it earlier today and am happy to have had a perfect use for it already, lol.
Fair XD
Now that I think about it, I think I used this shit before
but not in 2019 because it looked way better when I used it
Yeah, it was pretty popular for a while.
I'm pretty sure they released a game that involved the image gen thingy
something something mancer
i2i
Big difference.
really nice
neat
Midna
oo nice emoji

Older moth girl gen i did
yeee
awesome pose
Also my moth girl model https://civitai.com/models/23972?modelVersionId=28648
omg so cute
I can't even run lycoris yet lmao
gonna have to upgrade one day for this
then maybe try to retrain some of mine loras with the new tech
your Rauru lora is also very cool
Thank you!!
How I use
@lime walrus
Currently, there is a public bot on the server that generates images available as a research beta for SDXL, you can find the current status of the bot in #1047610792226340935. There are plenty of ways to use Stable Diffusion such as the official https://dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware - check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝|tech-support for any problems you encounter while installing it locally!
Mikoooo 
Can there be better compliment than this
Try changing the model, it definitely works well on some and not others
Oh no! I'm so sorry
I can't say anything else than ....
I haven't picked a name yet, I got it
But, I haven't changed
xD
Yeah... I'm pretty sure yours was gone in a heartbeat. 
Yeah, so Imma stay like this for a while xD
.l.u.n.a.

._.l._.u._.n._.a._.
XD
Quink before someone takes your new identety
do ___luna

Tru

luluna
lüna
Luuuuuuuna
Aw, I think it's cute even if it wasn't what you'd wanted originally
I think it's nice
Maybe at some point the person who has just "luna" will change and you can steal it back lol
I mean, I had Lunabelle as my name for a while before I change it to just Luna
Besides, you can always change it to luuuuuuna later.
That'd be neat xD
Lol oh no
but im almost certain that such nice name will be taken by dev of discord
Oop that's a good point, yeah
I just realized this is going to be a thing now
This is what happens when Discord loses its roots
yup, it's already happening, a lot of people from other discord was doing stuff like that
agreed

LOL
do it
XD
So beachwear means wearing the beach itself?
Looks like Violet Parr grew up.

its kobeni
Hmm. Haven't seen Chainsaw Man. The idea of a protagonist that has a chainsaw for a head wasn't appealing to me.
WAT
You need to see it
You have to see it, otherwise I'll find you and steal your E, A, and B keys from your keyboard 
Specifically those 
And thats fine, the anime is dope af, is like one of the best animes I've ever seen
Did you enjoy FLCL?
I've never heard of that, I'm a youngin
we're boomers wydm
Am not a boomer, I think >:o
^
How old are you then 
im definitely not 27
same
DEFINITELY
i am DEFINITELY not dying of old age at 27


I'm close ^
And good morning to ya.
Past the 30 hill myself, feelin 40 everyday.
Goodmorning 
O, good morning nubbers 
You aint 

I also have a bad back which doesn't help me feel any younger.
Can relate. I've had tons of health issues for the past 20 years. I've always felt much older than I look.
Love the eyes.
same, and the outfit looks like from the webtoon Return of the Mount Hua Sect
Oh, I love that manhwa.
📆 
My Google search suggestions... one of these things is not like the other.
I don't know why I'm trying to get a SFW leotard side shot
that's a very specific kink but well
I'm trying to make it sfw to post it 
post the not so sfw one at you know where
well, you definitely can't post those here 
Yeee, I told you xD
I'm handing bonks out so line up.
Pog
i recently hurt myself in January and im just more or less recovered from it, still abit weird and since then i feel like ive aged 30 yrs physically. it suckssss lol
Sorry to hear that! Yeah, it's not fun. Physical therapy is helpful for some things like that, but what they don't often tell you is that it needs to continue after the fact on your own. Stretching and moving is just as important to help get what you had before back. I had hip surgery when I was in my teens (being so young helped a ton) and never really got back my full range of motion until years later.
Sometimes you get some very interesting things when you don't explicitly ask for a school uniform.
very very true on the stretching bit. j stopped for a few weeks and i definitely felt it and had start back up luckily i got vr for cardio to help
but fr couldnt get out of bed first 2 months. it sucks
hopefully you are feeling better now. sending virtual hugs 
been there done that 
I guess I need my black text revoked, lol. Only made it one day.

Tbf there's so darn much to remember with SD. LoRAs, extensions, embeddings, hypernetworks, upscaling, inpainting, Adetailer, Multidiffusion, VAEs, varied models and their rules and guidelines... not to mention all of the different settings that can be changed within SD and/or have been modified or added by extensions.
true, and when you get into the habit of prompting a certain style, it's hard to unlearn that and learn different style
Between the recent UI upgrade (on Cozy Nest now) and switching between anime and real to test new models and LoRAs, it takes me a hot second to think through the process. And most of the old styles I have saved are too old to be relevant anymore.
i'm now settled into a very comfortable state, where I'm happy with my generation, too lazy to learn new stuff, and thus I'm staying stagnant. Unless something big happens, then I'll most probably stay like this 
tbf the stuff you gen is really nice as-is so no rush
thank you
well, it's just a very specific style, so not everyone will like it

Could you share the parameters to this?
here you go ```(detailed background, best quality, masterpiece:1.2), cinematic lighting, extremely detailed, hyper realistic,
1girl, yellow hair, bob cut, hime cut, detailed glowing red eyes, low contrast, flat color, upper body, high-waist shorts, closed mouth, light smile, galaxy, neon,
Negative prompt: (negative_hand-neg:1.2), (low quality, worst quality:1.4), (bad anatomy), extra digit, fewer digits, (extra arms:1.2), bad hands, by (bad-artist:0.6), bad-image-v2-39000,
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 22, Seed: 212511795, Size: 512x768, Model hash: 931aaff934, Model: Newly Downloaded_expmixLine_v3, Denoising strength: 0.53, Clip skip: 2, ADetailer model: face_yolov8n.pt, ADetailer conf: 30, ADetailer dilate/erode: 32, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint full: True, ADetailer inpaint padding: 0, ADetailer model 2nd: hand_yolov8n.pt, ADetailer conf 2nd: 30, ADetailer dilate/erode 2nd: 32, ADetailer mask blur 2nd: 4, ADetailer denoising strength 2nd: 0.4, ADetailer inpaint full 2nd: True, ADetailer inpaint padding 2nd: 0, ADetailer version: 23.5.19, Hires upscale: 2, Hires steps: 10, Hires upscaler: 4x_fatal_Anime_500000_G, Dynamic thresholding enabled: True, Mimic scale: 7, Threshold percentile: 96, Mimic mode: Half Cosine Up, Mimic scale minimum: 4, CFG mode: Half Cosine Up, CFG scale minimum: 4, Version: v1.3.1
Template: (detailed background, best quality, masterpiece:1.2), cinematic lighting, extremely detailed, hyper realistic,
1girl, yellow hair, bob cut, hime cut, detailed glowing red eyes, low contrast, flat color, upper body, high-waist shorts, closed mouth, light smile, galaxy, neon,
Negative Template: (negative_hand-neg:1.2), (low quality, worst quality:1.4), (bad anatomy), extra digit, fewer digits, (extra arms:1.2), bad hands, by (bad-artist:0.6), bad-image-v2-39000,```
Ty, Bruh theres so many things i dont know lol
well, take your time to learn, this technology isn't going anywhere
if you need help, you can ping me, I'll try my best to help 
thank you, im kinda stuck on normal stuff, as theres not a lot of tutorials that i can understand
We're here to help. Let us know what you're struggling with and between all of us we can probably get you on the right track. https://cdn.discordapp.com/emojis/1007968094486003752.gif?size=48&name=thumbsUp&quality=lossless
there's some tutorials out there that works well. You can also read automatic1111 features here https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Features
it's kinda a good way of understanding some stuff built into auto1111
Boy this really likes navels for some reason. I even put it in the negative, heh.
every time I find a style I like I start experimenting with other stuff 👉 👈
yoo ty, some days ago i saw this https://civitai.com/models/30683/automatic-wallpaper-generator
And i wanted to try it, so i spend the next 40 minutes trying to discover how to use it and in the end i didnt get it to work
Important instructions below! Make sure you read the "How to use?" section! What is this? This is a collection of wildcards with extensive, customi...
I've been genning for like 8 months and I still don't have a visual style to claim lol
thanks, imma check it out
well, it's always good to try new style
you can try to mimic the author by using the same model, same prompts and stuff. From there, you can change stuff 1 by 1 to really understand how stuff works
^that's what I do
oh, anyway what you linked is a wildcard, so you bascially need the dynamic prompt extension for that
yeah but in the civitai it said i had to download some extensions and put some stuff in folders, its labeled as "wildcards"
tldr for wildcard, is that you can create a "list" of stuff like example colors. And inside this list can include a lot of different prompts, example blue, red, yellow, .... When you use this wildcard, it will randomly choose one of the prompt inside that wildcard
i actually tried that one day, but my results were not neraly as good as the autor's, idk maybe im lacking some resource
Oops i forgor ^
Today's creation 😁😁
yeah, i do
then you can make sure it's working when you trigger the wildcard, check the metadata of the image
im not home right now but ill try it again when i get home, ty
Then what are you doing here??
Just kidding. Hope you're able to get things working.
lol, i just got off the pc
DAMN
Then how are you T Y P I N G ? Everybody knows you can't type without a keyboard, duh.
my brain is connected with my keyboard
We got new Controlnet Reference preprocessors a month ago and I just now noticed
https://github.com/Mikubill/sd-webui-controlnet/discussions/1280
i can type with my mind

Your notice is my noticing. Then again, I haven't used controlnet since it stopped working for me a couple months ago. I need to see if it'll install this time and work properly.
BotW era Ghirahim redo
yooo i got it to work! thank you
nice nice nice, great job! 
nice colors
ugh where are people going with all the subreddits closed? this is driving me nuts.
time for a new reddit alternative I guess
Nah. As useful as the platform can be, let it burn a little more. The fun only just started

Chaos is fun when spectated
with the IPO around the corner and them pulling stuff to be beholden to shareholders instead of users, I'd like to see an alternative pop up
their platform is valuable because of the sheer size of the userbase. The software side is frankly pretty bad.
there you go ```(detailed background, best quality, masterpiece:1.2), cinematic lighting, extremely detailed, hyper realistic,
1girl, yellow hair, bob cut, hime cut, detailed glowing red eyes, high contrast, (closed mouth:1.1), mask on head, hair ornament,
Negative prompt: (negative_hand-neg:1.2), (low quality, worst quality:1.4), (bad anatomy), extra digit, fewer digits, (extra arms:1.2), bad hands, by (bad-artist:0.6), bad-image-v2-39000,
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 22, Seed: 3674822200, Size: 512x768, Model hash: 931aaff934, Model: Newly Downloaded_expmixLine_v3, Denoising strength: 0.53, Clip skip: 2, ADetailer model: face_yolov8n.pt, ADetailer conf: 30, ADetailer dilate/erode: 32, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint full: True, ADetailer inpaint padding: 0, ADetailer model 2nd: hand_yolov8n.pt, ADetailer conf 2nd: 30, ADetailer dilate/erode 2nd: 32, ADetailer mask blur 2nd: 4, ADetailer denoising strength 2nd: 0.4, ADetailer inpaint full 2nd: True, ADetailer inpaint padding 2nd: 0, ADetailer version: 23.5.19, Hires upscale: 2, Hires steps: 10, Hires upscaler: 4x_fatal_Anime_500000_G, Dynamic thresholding enabled: True, Mimic scale: 7, Threshold percentile: 96, Mimic mode: Half Cosine Up, Mimic scale minimum: 4, CFG mode: Half Cosine Up, CFG scale minimum: 4, Version: v1.3.1
Template: (detailed background, best quality, masterpiece:1.2), cinematic lighting, extremely detailed, hyper realistic,
1girl, yellow hair, bob cut, hime cut, detailed glowing red eyes, high contrast, (closed mouth:1.1), mask on head, hair ornament,
Negative Template: (negative_hand-neg:1.2), (low quality, worst quality:1.4), (bad anatomy), extra digit, fewer digits, (extra arms:1.2), bad hands, by (bad-artist:0.6), bad-image-v2-39000,```
ever since updating torch my initial gens look better than my old upscales lol. (I'm slow to update everything!)
these are the most pissed off girls lately lol
death stare
👀
changed models, got happier waifus
With a mecha-boyfriend
or is that a waifu too?
just a bit....masculine and metallic?
rip hand
un-rip hand
may i know what is Adetailer?
To the market!!!
These silly double umbrellas haunt me
I swear, umbrellas are the bane of my existence lmao
umbrella-ella-eh
Whenever I want someone standing in the rain, no matter how much I make "umbrella" negative in the prompt, it still shows up
in yall's opinion, whats some must-have extensions and embeddings for image quality?
same, ugh
I miss Napster. 😛
that's a huge umbrella
ty
After Detailer.
it provides additional end inpainting.
i.e you want to add eyeshadow, fix the eyes, etc you use AD
Interesting
Speaking of giant umbrellas 😂
That's huge
lmao
I like coloring them myself
Better
This is great.
That prompt is almost a classic lol
still looks nice
try https://civitai.com/models/49037?modelVersionId=53632 it is my favorite 2.5d checkpoint. I've made some disgustingly good images with it.
This is a personal merge that i've spent several hours tweaking and testing. A good number of people requested an upload so here it is. Its good at...
Moth girls are based.
Why am I getting these weird spot stuffs in img2img generations?
do you have any thing in your prompt set to a high modifier like :1.7 or above)
No but I am using the segment face fix (in ComfyUI)
I don't have experience with ComfyUI, but as a A1111 user weird artifacting is almost exclusively caused by either using a low quality lora/checkpoint, or by putting high weights on a prompt. But in a img2img context your denoising strength might be too low to get rid of the splotches.
put cum in the negatives
Cum in the negative prompt? That's a crime.
Hmm. I am using Mistoon anime, .5 denoise, default weights on prompt
adetailer is like an auto inpainting extension that can detect face, hands, and body. It's especially good for face tho. Not that good for others.
ty, im going to check that out
some interesting colors being posted
evenin'
hey, this extension is really good but im not sure what settings to use, could you give me some advice on that?
big ears
damn what model is that?
use the normal settings, with face adetailer model
that's what i do anyway
do you use it on every gen @lyric zealot ?
custom mix, might upload later, mixing another one now
Sir Mix a Lot
almost always
I need some new wallpapers
At 3840x1600.
need some new stuff
riding a bicycle made of pickles
using a large slingshot to launch a cooked ham
com'on, let's see something new

kek
well.. umm
me at the grocery store at 2am
da fuq?
keep 'em comin'
No pickles or hams.
hmm...
interesting
same
maybe I can generate one.
my fav from last night
sci-fi-barnacles
Gimme something in 3840x1600 🙂 None of my wallpapers are coming out. 😦
This is generic and boring. 😛
here I'll help make up for it
that is weird
was suppose to be nuka cola
atom punk chibi doll woman at a nuka cola amusement park
sadly I can't gen that high of a res
fun vehicle
I can't quite get there natively either. Have to use SD upscales or regular upscales.
and back to random. hmmm
ok no melting anime girls
That was a pretty damn good seed; I drew in some lighting if you're interested.
you might need to upscale it yourself

oh boy
nope, it's my second waifu 
I got 3 waifus and a husbando myself
i will need you to see these waifu(s) and husbando
hmm... my image viewer gui is messed up when scaling is enabled on ultrawides... just found the bug
need to see if I can fix 😐
My family man husbando has a helmet on 😦
oh cool 🙂 glad you liked it!
little more anime-ish
no
even better
it's a simple prompt though
beautiful girl holding a book made out of human flesh and organs, flowing flowery dress
it usually makes the girl out of flesh though
The photoshop techniques I learned from coloring several OPM chapters have come in handy.
I call this next piece, "Nii-chan, please read to me?"
my model doesnt like creepy flesh books ig
huh?
what about this 
amazing lmao
there we go
is it a boy or a lady. .it's a ladyboy
nice pose
y'all having an ai battle.

Goodnight Smol!
well I got anime 😄
Another edit
getting some odd results but interesting
not sure I wanna know what that is on the right
My bonk bat is ready
Currently gearing up to extract a LoRA for the first time, vibing to
Surprised no one has extracted this yet: https://civitai.com/models/65/complex-lineart
Very nice man. Love the aesthetic.
Extracting a lora is fairly easy with the supermerger extension, but it's only taking the difference between two models. not sure how you could extract a specific style from a checkpoint. is that even possible?
lantern festival,Oddments, array of strange objects, surreal composition, vibrant palette,in slow motion, artwork by makoto shinkai, hdr toning,night, dark scene
Thanks for the prompt! I'll have to try that out. And regarding LoRA extraction, I'm currently following this guide. It uses Kohya_ss: https://www.youtube.com/watch?v=hpdp0ZtnJlQ
very rare to see a generated image with light that is not flat, I usually add that in manually
If you know the base model that the one you want to extract was trained on, then yes. You just difference out the base model.
sneezing in slow motion 😄
Seriously? I guess I wasted my time then, and can just install the mentioned extension and difference it out on 1.5 
Most LoRas I'd image would use the original 1.5 or the anylora model since they are the default options in the dreambooth colab notebooks I've seen
Yep, should work. Won't be as good as training a LoRA on a similar/the same dataset, but generally it's pretty close. Although they do tend to be weaker.








