#🍥|anime
1 messages · Page 16 of 1

might have to explore that one day
llm?
large language model
ah
basically text generation
lots of way tbh
I just downloaded a bunch more esrgan
they're 1x "upscalers" made to just fix things
might need to try that
waifu
no you!

https://techgaun.github.io/active-forks/index.html#https://github.com/AUTOMATIC1111/stable-diffusion-webui if automatic stops updating his webui, these are the top forks apparentlt
control net faces
learning to draw them properly as well will give more potential for creating characters
@lyric zealot
ruined 
pretty 
much better than chillout lol
now go to reddit and make a post about it

but i agree

they will complain i made the 9999th asian doll clone
exactly 
but your one definitely looks good tho
hello everyone, please tell me how to use the bot?
there is no bot, you have been living in a dream neo
:с
I wanted to make a portrait on the avatar, but free attempts were removed everywhere :c
currently no, no bot around. You can check the #1080946152318443610 but here is the FAQ on it
Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.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!
automatic1111 is free, you just have to invest time on it, assuming u have a gpu
(and money for the hardware though)
but yeah, even then there are notebooks, or websites
it just takes time
Ok, thx
Guys, what prompts do I use for the eye to look like this, in a happy expression?
tag should be happy closed eyes (confirmed)
what is danbooru
an image board, turns out the tags are pretty unreliable imo
Back on the a100 I see
FINALLY
Damn thats op 
now i can do 2560x2560 img2img sd upscale too 😄
will it make me win pow
no.
i suppose i'll be dangling somewhere down below
She has a 3rd leg, literally 
WHUUU
perhaps such a high batchsize wasn't a good idea after all
did you remove my message guizmus?
nope, the bot got it
this is a blocked word
but I'm here to remind you that even as a joke, this isn't ok
you can see when people use it?
just don't use that kind of language at all around, sorry
What word? 
nope
I missed it tho, what kinda language?
changing subject, this isn't something I'll have fun around
sorry, lips sealed, guizmus wants to keep it clean 🙂
it hasn't happeared
Fair fair :O
and i'll respect his authority as a mod
It got caught
ahhh, it got flagged?
yeah
although it's basically street language where i live, that's why i just used it
but... yeah.
as international channel, perhaps not so much
Not making any big deal out of it, I removed the alert for us. But let"s move one.

Fair :O
lol, img2img 2560x2560 = almost 60gb vram needed xD
Why would you do that tho 
Happy Friday everyone!
O wait showlove is still a thing? I thought it was like a valentines only thing :O
Happy Friday 
just because i want to. 😄
❤️

it originally was but why not spread love on the other 364 days of the year 

see, that's why the other machine broke down, it didn't get proper thermal safekeeping and was allowed to be cooled during jobs

Because that would mean i have friends :(
so now i make sure this machine doesn't die, by generating waifus
Oooo, thats pretty nice then 
nom
Fair XD
Evening chat
Now i need something so that i can parallelize the inference so i can just throw all the A100's under the bus for glorious waifu generations
you can set up your pc with --listen
make vpn to your router
unlimited generations on your phone!
4 horns 
isn't that the whole goal of Accelerate though? Doesn't it handle multi GPU ?
it sets up multiple instances...
the inference step is not parallelized 😦
unless it changed when i last checked
training can be parallelized
well, if all gpu's load the model, it shoudl be doable tho
instead of 1gpu doing 8, make all do 4
or just return 4x8 results
for the time of 1
yeah but you're right, you are already multiplicating how many time Torch is loaded for example
it's not the same
Set up a feedback loop with a Llama to generate waifu prompts then have it run them
would need to set up the oobabooga container then, not allowed to flat install on that machine
but i do suppose it can run a 30b LLM model
Sexy
hmmmm 🤔
too bad llm's are now limited to 2048 tokens of memory 😦
Highres pixel art 
I have right now the LLM + SD plugged in, with a custom prompt for the LLM and that model
works great

ah, plugged it into comfyui?
Aurora is so OP
Old gen 

does anyone have any material on "finetuning" with prompts, hair lengths, hair styles, eye shapes, irises, face shapes, etc.?
So hellou everyone ! I need your help ://. So you maybe had saw a videos there AI (They say its stability AI) create a photos of anime girls. So I want to know how to make that kind of stuff on stability ai. Now you thinking that im simp. NO. I using renpy game and I need original photos. So maybe can somebody help me tell how to use it ?
If you simp, we support the simping 
But also
Smile thx <33
is this the pixel extension?
https://civitai.com/models/40199/aurora I just use a chad model

Omg this is actually better than saucity sauce which is better than kenshi
How pixel?
(pixel art:1.5) 
Yis 
Is there any way to filter for checkpoints only on civitai?
On what model? Does this work on most danbooru models?
Not sure, works great with Aurora tho
IC
Can anyone please guide me how to create art through SD
the non-start-here method? get A1111, get model of choice, set up, hope your gpu can handle it, and go to town!
Water colors easy pass
oh A100, i missed you so much
1 batch of what resolution?
just 640x640
where's that test site?'
beyond 640x640, i feel images can become unstable
the better part: 8x, 640x640, 40 steps = 20sec
just over 1 minute, now i'm paralyzed with choice again... which concept do i want to attempt something with?
40 steps? Euler a?
dpm++ 2m karras
nah, karras still has stuff left to resolve even at 60
Bruh
karras is about completely convergent at 90
But it remove da detail
Waifu diffusion tagger
Everyone uses it
Or what did you mean?
What site is that? :o
Bruh
I just found it and was about to send
Oh wait
This is something different

she cheated. but did you notice?
I guess what it wants to think something is AI, is that it looks too good or has artifact flaws
she has distracting. features.
what happens when you send it something hilariously bad?
dunno don't have anything that bad

goodnight 
Good nite! 
when yandiffusion ? 😮
She do, it is a 500 IQ strategic play 
good night !
?
What model is that sdxl?
nope, this is Perpetual, sun version (unreleased for now) and coupling it with the custom LLM prompt they are giving for it
should be out soon though
they did a "Sun" and "Moon" version with different offset noise
It doesn't look like an anime model
it's not
what do you mean ? I'm a outcast for using standard models ? :p
No
Just that you are like eface in that regard
He also mainly uses like general purpose models
I didn't know
yeah, I'm not really good at prompting on anime model
but to be honest, all the prompts for the pics I have shared right now are
(before LLM)
best 1girl waifu anime girl
Good way to be better is to steal
And also danbooru tags
👀
some other seeds for "best 1girl waifu anime girl"
danbooru tags can be hit or miss tho in my exp
ComfyUI
You know most (like 99%) of anime models are at least based on danbooru
You know i was joking right lol
Oh
Ah yes I can read that
She chuckles bashfully, i should start doing that
the thing first adds my real prompt, "best 1girl waifu anime girl", gives that to the LLM, I get a real prompt back, I run it on SD on batch size 4
local llm?
yes but there are also nodes to use chatGPT api instead
What model you using?
Pyg6B best
gpt4-x-alpaca-13b-native-4bit-128g currently
shouldn't you go rest 
if only i can run that 
I mean. I am in bed so technically half there 
how much vram do you have?
and do you use this oobabooga thing?
Meanwhile I am here with my expanded family in a restaurant
yes i am 
use --pre_layer 30
it should work
it will be a bit slower than usual tho
a full run of the thing peaks at 24GB VRAM on my system right now, in around 10 seconds (4 times 768x768 pictures, it's a 2.1 model)
Man I got an error last time I tried installing oogabooga and gave up, maybe I can solve it with the new aitrepeneur video
Though I need to wait until Sunday to get home
you can run the LLM on CPU and RAM though
Does alpaca only work for nvidia gpus or did they get a way for amd to work? Cos having it ran off llama and cpu is horrendously slow
it's already taking one lifetime to generate a response tho 
that's gonna burn my house down

use that --pre_layer or cpu mode
Takes around 30 seconds per response in my experience, which is about what I get with gpu
(on Pyg6B CPP version)
with the model already loaded right now, the prompt is made in around 3 seconds. the picture takes more to build
Ooba?
for CPU, no Llama CPP
Gpu?
yes, 3090TI
Oh nice
Yeah i got llama. Takes literal ages for a reply tho. 15mins for afull hello back
wow
I didn't know that
Right?
So 3090ti is 10 times faster than 3060ti?
yeah, not really an option
not sure it's that much, we aren't talking about the same model either
What is your cpu?
Ryzen 5 2600x
Well yeah
I get around 10 tokens per second
Bruh, I think it's time to upgrade
Im on disability i cant just go upgrade whenver i want lol
Disability?
Yes
"Hi" generated in 10 seconds
What that mean?
So why can't you upgrade?
Yeah this convo going nowhere XD
I don't understand you
PICTURES

PICTURES
the last smile, kinda scary
it decided to do a prompt about fishgirl

Do stag girl
I'm running the same prompt since I started lol
I am also technically disabled
that's what is funny with this
a style each time
but I'll add "stag" because it's you ❤️
before the stag prompt got run
it doesn't want to make a stag girl for now 😢
what model are you using?
darksushimix
amazing
a chicken. in a bird suit. logics.
how is this generated again?
#🍥|anime message
this workflow, LLM + a custom model
he just prompts "girl with stag horns" AI does the rest xD
probably... dont'shoot me
prompt is "girl with antlers, anime style, stag-girl"
Oh a comfy ui thing
and you need a model which actually understands it

A luminous and enchanting girl with antlers, painted in the striking and expressive anime style of Yoshitomo Nara.
example of a real prompt the AI makes
here's some absolute unit chibis with an attempt at stag horns
thats cute as hecc
I redid my gpt primer and its been giving better results for prompts
glowy
Anthro wyvern, stag and shark girl furry character into >
Anthropomorphic wyvern, stag, and shark girl, hybrid creature with wyvern, stag, and shark features, combination of dragon-like wings, stag antlers, and shark-like attributes, unique and fantastical appearance, feral yet humanoid, fierce and majestic stance, powerful wings with intricate details, majestic antlers on a deer-like head, sharp shark teeth and fins, sleek and streamlined body, combination of scales, fur, and shark skin, dynamic pose with a hint of danger, fantasy art style, vibrant color palette, detailed rendering of the wyvern, stag, and shark elements in the character's design, evoking a sense of awe and wonder.
thanks to your model
lets feed it into my model 😄
abstract :D

yis 😄
uhh... the word anthro and horny models... yeah...
yeah be careful i use allot ng inversions to keep it sfw lol

not entirely bad tho
it's to protect us, not them
XD
fridaynight = drinking + lyrics night 😄
The last March of the Undead - Machinae supremacy
Adding some highres fix to go to 1920x1080
innate artstyle, prompt, model, or too little denoising?
seems a tad blurry
god thats nice
still tuning the params, but not enough steps I think
this was 30 for lowres initial pic and only 20 in highres fix
I need feedback guys
- Make meme image
- Download kohya ss gui
- Natively fine tune tab thingy
- Make folder correctly for kohya ss, put empty txt file along meme
- Put learning rate 10000, epochs 348
- Publish model
- People use model only for it to reveal meme
- Success
make never gonna give you up
... i have images so red, i can't use the excuse ketchup incident anymore 😦
I am not on my PC until sunday
Raise your banner - Within Temptation
this is animuh channel, i dunno what kind of girl that is.
waifu-sense doesn't compute
xD
you are right
stagfu instead
as long as it's not going to be the main focus, but showing the alternate outputs shoudl be a fresh nudge, at times?
@wind girder what you think?
She wrote a message with the ketchup of her enemies
ngl, i like the output these lyrics give
imma move on tho, i heard mods are in this channel, dunno how long i can keep the ketchup going without a warning xD
1/ have never done lora myself, no idea about the good LR or epochs
2/ meme = text or very repeated frames, can be hard to train without burning it
but outside of that, all good
You misunderstood my purpose
Also not lora but model
Purpose is to burn in image into model so hard the model becomes deterministic
well
So the model can only produce the meme image
They seem determined to take back their throne!
I tried that at some point.
I wanted to push out a model named SD 2.0 when it was taking ages coming out
I wanted to burn rick roll into it so that it was all that would come out of it, watever the prompt.
This didn't work.
Either the LR was too high, and the model started to diverge (not learn anything, go into NaN state, bug out)
or the LR was too low and even on around 30k epochs it wasn't showing if I prompted "table"
I would have needed to do a "full fine tune", using thousands of tokens, all linked to the same picture
Sakura - Clean tears (Ft. Hatsune miku) will not ever not produce lovely results
My tutorial is for full fine tune
Or natively fine tune
Or whatever it's called
STOP IT. MY FROZEN HEART CANT TAKE THIS MANY FEELS
this was my model card ^^
I know I scared our community manager with that :p
can someone give me a good fantasy background only prompt?
"fantasy forest background"?
xD
Do you mind trying my tutorial?
i said only background, no characters
i mean a full prompt
it's not really complete but sure, I could give another try
finally... masterpiece actually enhances an image?! ( i love the heart in #1)
Wdym not complete? I just skim over most parts and don't go into detail, but it's practically the whole process
can ya give me the full prompt?
We are by chance Our steps are different, but we're holding hands It's a short but long moment If the direction is the same, we can move on
| A vast, vast world, a missing star puzzle As if to fill my weaknesses, I'll meet you Many stories that continue to shine The next page is when the cherry blossoms bloom
| we keep repeating Spending time together talking about memories I felt it even though I couldn't understand the language We understood each other Heart to heart
| Someday she will meet again I believe that time will come The days of the cherry blossom road trip illuminated by the moonlit night Give me flowers in memory Thank you thank you thank you
| We walked through different times Overcome sorrow and suffering My end point is a little before that you were by my side Thank you thank you thank you
| Give me flowers in memory thank you
} ```
yeah but I mean, some of those details you skim over are quite important in the process and in the success of it :p like I said, not sure if this would work, I failed the few times I tried.
I see new ways to tackle it now, I learned more since, but I'm not entirely sure it would work
ok nvm that wont work
#1 was surrealism, desktop wallpaper, 8k, masterpiece, Someday she will meet again I believe that time will come The days of the cherry blossom road trip illuminated by the moonlit night Give me flowers in memory Thank you thank you thank you BREAK We are by chance Our steps are different, but we're holding hands It's a short but long moment If the direction is the same, we can move on
you need it a bit more precise?
nah, i need a booru based prompt
ya... i got none of those
even worse -> better, i actually made my model worse on those lol
Ok
tried mixing blacklight + sugar water style
ima need that prompt
seems pretty good, but most of the time I get pink skin 😔
hold on
thx
I'm basically testing old prompts that I used on NAI and adding some new stuff. I don't even know if prompt mixing works on SD. The prompt is basically a fucking mess:
Masterpiece, High Quality/High Definition, extremely detailed CG unity 8k wallpaper, intricate details, Illustration, beautiful detailed, high resolution illustration, Extremely detailed, beautiful, Anime Style| Brilliant Colors, dark intense shadows, Depth of field, solo, 1girl, yukata, Character focus, Character Enhance, long hair, hime cut, dynamic effects, cyberpunk city, (upper body:1.1), <lora:Black Light:0.4>, blacklight, neon lights, neon graffiti, <lora:Sugar Water Style:0.4>
no prompt is a mess, all prompts are equally beautiful -> Said the madman
because dumb prmopts get the best kind of shit 😄
Actually true.
Break through it all - Sonic Frontiers
Tranquility - Sawano Hiroyuki
Horizon - Clean Tears
i like how she's standing 😄
And the Hero will drown - Story of the year (BEST need for speed song, FITE ME)
hmmm... might explore another extra layer of rng 😮
damn I couldnt sleep
Messed around a bit with models.
katekyou hitman reborn apparently getting a new adaptation
Usual model for comparison.
what model is that?
Cetusmix.
k thx
Been told it gives a blue archive style look.
what's that?
Last bit of testing was with trying to get glasses to look good.
which pic is better(both look kinda shit and desaturated because i am using auto generation function of kohya ss)
Eyes in 2nd is much better.
ok nice, then this lora serves its purpose
Lighting and colors aren't too different although that's probably the vae.
nope, both have no vae applied(because i am using auto generation function of kohya ss)
A3CAEWC?

its a model name
ahhh, then it's not 


smol oblivious 
My mix isn't popular enough
adding abit of 3d lora
grrr how was it you got a control net pose into a regional prompt or latent couple or multidiffusion region?
wasn't that an additional extension...?
or smth
is that a ... uhh... skullfck horn?
the other one goes into...
How do you get the buildings looking so nice
If you are asking from me then it comes with the model
Whats the model
A3CAEWC right?
sakimori style is a success
Yes
nice, can i merge your model into mine?
Go ahead
I've renamed my model but I haven't decided on a new name
those who know what it can do, any suggestions?
what model?
nah thats good
it has a crazy range
when will you release it?
but its no longer just a dark one
I probably won't be releasing it, it'll be a very limited share with people I trust won't post it
cause i don't like civitai's new TOS
I'm master of model naming just look at A3CAEW and A3CAEWC so I can give you perfect model name 
lol
what are the new TOS?
what changed?
to make my beef with civitai's new tos simple...they can do whatever they want as they gain non exclusive royalty free full rights to it
they can sell it if they want
without giving me shit
HF?
no, civitai
CivitAI could just take it from there and post it back there
sounds like folks need a new site
I meant like uploading to HF
Thankyoy
this is bad news
they chose the corpo lifepath
If you uploaded it to hf and someone then uploaded it to civitai couldn't you sent message to civitai to take it down since you didn't give permissions for sharing it?
no cause civit may be the ones doing it
they're kinda sneaky lately
but I don't think they'd bother with mine
unless it became real popular
or at all
not really worth a lawsuit
but they've posted models and then given credits etc and said it's a repost
they are allowed to share it using the same licence, if the licences allows it (yes by default)
But they can't really change your license without your consent
source : they imported my models before contacting me, and respected the licenses
but if I don't allow them to even import it then?
not sure which license that would be
you can do that yes, creativeml-openrail-m allows you to specify conditions in your model card
and I just might as well post it myself then, but if they import it, I won't give them nonexclusive I guess
they don't import anymore from what I saw
well, they can try
how? They can't sign my name on a contract
that's fraud
they changed all the licensing rights to defaults for them when they added such a system, and that meant this for all your model if you didn't check at that time :
I mean, legaly no, they can't. but ... shortcuts are taken, like I just showed in this screenshot, so it's good to keep an eye on them still imo
no, they aren't eating all the models everywhere
they imported things from HF when they started, to build up a library
but they stoped quite fast
they have the people uploading it there now
then they can use the models I have, but as per their TOS if I delete my models
the contract is cancelled, for all future stuff
what's posted is posted though
The contract itself is clean yes, and they seem to follow it to a T, and follow takedown notices too for now
I can post my next model and call it Everdark since it's no longer related to my other one
it's no longer specific to darker stuff
its a whole new world
thats cool, on civitai?
I don't know
I will have to decide on this after determining its value to me, cause right now I'm in dire need of cold cash 😛 If there's a way for me to sell it to a subscription website or something, I dunno. Others have done that
but then I guess I need for it to gain popularity
so I would need it on civitai anyway
i dont really care about what someone does with my model, the only important thing is that they have their fun with it!
dont be like this bro
I can't be that carefree atm
when you're healthy and well then you can be the cool hippie
you will recive a LOT of backlash from that if your model is a little popular
thats what i want to say about that
backlash isn't really something I'm in a position to care about
I have cancer and got bills to pay of all sorts, if the opportunity to sell a model of mine comes up
I will take it
- this will make not really much money, if you really need the money sell pictures or game assets(i know its "hard" to do, but yea)
nevermind i said nothing
it makes a good bit of money by selling to certain sub based sites
it's alright, no worries
I have heard it's in hundreds
everyone ive seen selling models said something like "doesent make much money"
👀
FR WTF?
Beautiful
to make professional models ? or sell model rights of popular models ?
neverfuckingmind is said absolutely nothing
first one, I saw run from 300 to double that rencently, per model
to sell the rights to use my model at subscription based services for AI imaging
it was what I got as offer for an icon model
it really depends on the trafic your model already gets, but it's hard to measure how much is worth a "use" tbh
no idea
well I guess also by how versatile and useful it is to them
why pay for model rights tho? generation services or what?
hmmm how to get my controlnet openpose where I want it
@cyan burrow make a model you have get a ton of traffic then ez sell
yeah generation services
i hate them
heh help me get a ton of traffic ;D
You could pull some "schemes" like how andite did
Move it to right position in pose image
He named his models anythingv4 and anythingv4.5
Even though they had nothing really to do with anythingv3
where do I do that ?
because custom made model in the case of a specific pipeline that makes money, this has value.
Like making a model that lets you do photoshops of a given cloth, or make specific types of game assets. this gets sold at high value and fast
One way is to make it with this extension but I haven't used controlnet much
Oh wait nvm, he was asked that before #1072013871730131004 message
long neck
is that the sakura (song) prompt?
idk some random ass prompt i found on some website combined with my LoRa base prompt
btw, can you even say fuck+ass and so on on this server?
you might, but don't ever do that in horny models
Word.
i know for sure that all the anti-nsfw in the world will not save you from that, in my model
what
i mean say, literally say/write these words
she spends like... 10 hours at the stylist for a 1 hour role in an anime.
Actually now I am also wondering. @wind girder care to clarify?
Like it didn't state so in the rules
They only wrote about nsfw imagery
Guizmus is unavailable to answer right now. it's quite late and has been moderating for about 18 hours today
and not sure in current state what to answer to do it correctly
in simple words "meh... not too much but no taboo. problem with joke insults is that people don't get satire, and it becomes another kind of insults fast"
reprogram your LLM stack so it can answer for you!
LLM what?
"guizmus is done with your shit" XD
I'll make anime monsters in clay
continue like this and I'll post only realistic photos in here HAHAHAHAHAHAHA
Nice idea
Testing new merge
LLM -> large language model, stack -> the whole software stack he has dedicated to LLM
oh, learned something new thx
Why mod when you can feed your "opinions" on most stuff to the AI and then even if there is a unique question the AI can just guess what you would answer based on your other "opinions"
just found out about the latest automatic extension, was quite surprised by it, top on reddit right now
?
because it bugs a lot, and can just say insane things
that's an advert, if i ever saw one XD
Depends what model
not even, I'm still just in shock :p https://github.com/Stability-AI/webui-stability-api
maybe I turn my models insane. they keep friendzoning me 👀
SDXL in auto though
idk whats so spacial about that
and no more GPU restrictions
One plus about AI, it can't say no to you....well unless it's ChatGPT then you're fucked.
no custom models tho
@wind girder so if i get this right. it's about getting dreamstudio/stability.ai "native in A1111"?
nope
From what I get on reddit where they explain more, XL seems around 25% trained currently and this helps get the tokens people need more trained too
taking a mecha to mecca to mech
it's about it yes, run the API in automatic, get access to the online models, and not run on your GPU
i have company A100, i fear no man XD
this is only what I get from the reddit and github, I haven't been specificaly briefed on this, it's just public info at this point
You just made me remember that one time I tested my game I am making rn using Pyg6B where you select between 2 options for a response each time. So without writing anything myself, just choosing options, the AI randomally asked me how my philosophy thesis was going.
well, now you did snowball it i suppose. but if it's not intended, i'm pretty sure stability.ai will shut it down hard
I am actually doing a philosophy thesis (or at least waiting for confirmation to start)
idk i think thats lame, why are people so excited about this
"free waifus"
I use 3 models, all seem to behave differently. I prefer the first one for long term coherency, but the second one is really wild in roleplaying characters, it just tends to go to magic too fast even in real world settings
sounds like overtuned then
is api free?
fun for situations ^^
You heard of tavern AI?
yes, I tried making Aether in it
the guy LIVES tavern ai
pygmalion is an "old" model
Oh right forgot
An*
bruh
Loves?
no. lives.
Ok
it's the must funny one, but it derails fast
Depends on settings
And prompt
yea, but it can do nsfw roleplay pretty good
in one year, we are talking about our favorite virtual waifus for real....
I imagine yes, not a pro in it yet. it makes really long prompts so it tends to hit the 2k limit super fast
"/remindme"
sharing
In one year? You mean right now
😮
I should try gpt-chan
mind control/replace -> after 2 answers, the AI is obedient
no, but text2speach & speach2text mandatory some people told me, not sure why 😮
there is your roleplay
one question i ask everyone: When do you think FullDive(Nerve gear/SAO) will be possible?
5 years ago: in our lifetime, not gonna happen. now? not so sure...
30 years without the losing consciousness
I think this has the prompt and stuff for tavern AI in the metadata
If you wanna try
the moment it's available, it's most probably a place people will choose to die
but controlling it with sensor input etc and very small virtual reality goggls controlled by a neural net
hard to say. I think there is still some progress, medicine wise, but medicine is exponentially accelerating right now
I saw a YouTuber who literally started college just to answer this question and to make it come true. He said, we might get close in like 204x~
do you have an estimated time frame?
"cleavage" in negative prompt. AI solves it by: don't press boobs together.
bruh, if i get my hands on this thing in my lifetime ima do something to the waifus man
Ayo
And then get cancelled for violating virtual girls. Haha
How to remove cleavage : 1boy
shhhhh... didn't we all already do that? ? 😮
LMAO
The correct answer.
I don't have such knowledge. I have hopes and dreams that this happens before I'm too old. I'm nearing my 40s
Heck birth rates already dropping then they drop vr waifus
Imagine aging
Guess I'll clone myself
Happy birthday
yeah, the mirror helps imagining it
can ya send with metadata?
turtle no humans
girl. she's the most flawless being in the universe, eternum angel, manga gradient art, epic, cloth
(eternum is a skinline by League of Legends, associated with mech/demonic/red)
with your prompt
it does't know eternum -> in that case, eternity!
Sry didn't get turtle only tortoise
Sick turtle
turtle, no human on my horny model
get rid of the cow, we need the turtle
lucky hit. 7/8 are just... weird
I just click copy image -> send to discord
ok let's see how this multidiffusion will turn out
fingers broke 
ok, can you send me the prompt and negative prompt?
I catch myself watching hands on any digital images nowadays
SD has tought us to look for the flaws there xD
AI Art 99% of the time
uhh...
Menace to society
Bored....AF.
Goodness gracious
ABSOLUTE. UNIT.
Omg wtf
Voice messages on discord?
Mind if I kill them to grind for exp?
my prompt is so fucking ridiculous, what have i done?
what?
Yes
you know they look dull, but that's the lvl 150 field boss between the lvl 5 starting area mobs xD
i'm looking at you, xenoblade chronicles
Except with Aurora 
Dw the more ridiculous, the better
Oh God not the Adamantoise.....
did you look at it?
Not yet
not that messed up
best prmopt is ridiculous prompt. letsgo!
Unaffected by the cuteness, I'll oneshot her.
wat
anyv3 is so ancient at this point
It really is. lol
It do be
ik this was back last year
There is Chadster Aurora now 
old anyv3
alfahentai. oh. right, i'm the only user.
Do you have it try on phone
Didn't you say it got a bunch of downloads? 
what happend to your perfect english?
jk, dont take it seriously
about 1k downloads in total
Then its good 
Imma send you some real high level field bosses to hunt you down though
Too excited
Huh, looks to be what? Level 4?
XD
Turtle mouse?
finally the prompt is running
are you sure you can even inflict 1hp damage...?
my mecha in progress (legendary powerful construct design of a futuristic war machine:1.26), synthetic mechanical fighter, mecha style, futurecore, (slim design, visible pilot:1.24), quantum technology, (plasma and pulsar weapons:1.19), fighting in the ruins of a (desolate barren cyberpunk mega city:1.23), (sand storm:1.16), ( cyberpunk art Style:1.24), (splash art style:1.21), (hypermaximalism:1.1), (digital art:1.20), (digital illustration:1.13), (cold and dark colors:1.3), hard lighting, volumetric lighting, (masterpiece:1.25), (best quality:1.25), (highres:1.25), (2.5D:1.232), HDR, Ornate, 8K
casts defense debuff
but all mechas are powerless versus chibi
if someone looks at this prompt: dont judge me i have no ideas left
Me putting 8k on my 720p screen
masterpiece, high quality, high definition, intricate details, extremely detailed CG unity 8k wallpaper, beautiful detailed, high resolution illustration, Extremely detailed, beautiful, Anime Style, Brilliant Colors, dark intense shadows, Depth of field, Breathtaking, solo, 1girl, hoodie, Character focus, bob cut, purple eyes, expressionless, streaked hair, cyberpunk city, cherry blossom, petals, (upper body:1.1), <lora:Black Light:0.4>, <lora:Sugar Water Style:0.4>
(EasyNegative, ng_deepnegative_v1_75t, badhandv4, bad-hands-5:1.1), sketch, ugly, huge eyes, text, logo, monochrome, worst face, (worst quality:1.4), (low quality:1.4), (blurry:1.4), (bad hands), (missing fingers), multiple limbs, bad anatomy, (deformed fingers:1.2), bad hand, (holding:1.6)
DPM++ 2M Karras
40 Steps
Highres
30 step
0.5 denoising
4x-Ultrasharp





