#๐โai-images
1 messages ยท Page 11 of 1
hello
hanako doing homework
Real Saya
I thought you did it
@rustic raft first try 
fixing some prompting and it should come out well, if that is what is already coming out with no tuning 
oh damn
blue hair? 
this shit is fire
yeah like I said, needs some fixing 

chunky chucky? 
it's getting there, horns still a little volatile
daaaamn look at that
non pointy ears?
blue eyes?
:0

its so cool
besides broken foot pretty nice 
anything else from the one you posted that you want to take over to a new one? freckles?
baggier clothes
ripped clothes
ceptum and lip piercing
older face
shaved eyebrows
chubbier
freckles
that too crazy?
also
WOW
u got my hair perfectly
i still get amazed every time i see what u can make with SD
Like actually that shit is fucking nuts
i can taste the new pfp
didn't even know what septum was before today
always something new to learn, yeah I need to experiment
ear piercings work just fine, others are ignored atm, I will see if I can force it
the rest should be possible
BRMDJSHSKKSKS
sorry that was what the AI interrogator got from the image you send me ๐
you're spot on dw
i have done visual measurements on squishy
and she do be tall
yeah im like 9'11
not that 
she was like 160cm
with a plus minus of 10cm 
pffffffttttttt

i cannot confirm nor deny i signed an NDA
@rustic raft mh I think we're on the way to get the feeling right
OMG
THATS EVEN HOW I DO MY MAKEUP WTF

the AI is loosing it too apparently 
jgmdkdjsnlsjsjsks
well I do try to catch your vibe after all 
hehe
AI knows how you look 
face + hair is a lot more accurate
its crazy how far ai has come
this one turned out pretty well! almost no obvious AI flaws
sad that the piercings don't work out of the box

ayoooo
btnrkduebem what is this second head xDdd
first try with an Anime model 
aaah
WAIT OMG
I DIDN'T REALIZE THAT WASS
I THOUGHT THIS WAS AM EXAMPLE IMAGE FROM THE LORA AND NOT WHAT YOU GENED
Ayo? @rustic raft level 68 !!! 
thats fucking wild
damn
With what? Did you want me to prompt it into a image?
basically try to catch Squishy vibe, no matter img2img or another way 
Iโll see what I can do
@rustic raft very volatile Anime model, sometimes great sometimes just broken 
this is so interesting
noone
were both stuck on mobile and cant change it
Ah
weve been in here for hours
Thanks lol
what LoRa's did u use for this?
- checkpoints
no Loras, besides xl_more_art-full
checkpoint AlbedoBaseXL 1.3
hmmmm
had to force it hard 
y does it always have an image of a head on the shirt?

I would guess the chucky thing might make it do that
huh
but I had the same thing happen with my other experiments the last few days, so also could be a recent update problem
some color bleed, but I love the vibe



why does it look like jinx
I know right! I have no idea but I love the gremlin energy xD
You have ABS??
i have a belly, literally just like that
this right here is my body type
something more realistic again

Hehe smurf shirt :3
fjkkcthnkhv
realisian is good checkpoint for realistic portraits etc
P.S. don't ask me about the random belly reveals, no idea 
btnfkjdjdksjsmslsl
gbjyvbmlibnhgfbnjv
me as a teen
ong
AAAAAAAA i look like my sister my age wth
Does anyone know the prompt to do a character fusion?
ongoing or a character that is the fusion of 2 characters?
rarely does that only with a prompt work, there are ways like ipadapter where fusions are easier
I tried to make one until it worked but it wasn't quite what I wanted
Yamato X Zero two
so unless you have checkpoint that is trained on that or you use 2 Loras of the 2 characters and force them together
but how can I force them to unite
Ayo? @past moth level 4 !!! 
well portrait ratio, 2 loras, only describe 1 character but trigger both loras with their respective trigger words
wait to use lycoris there should be page lycoris page right?
y2b x yamato(went wrong)
I put mine on Lora and it works
ye me too and it works but look at this
could be outdated? comfyui also just throws all those things together into Lora
ye maybe i watched tutorial how to use it and tutorial guy just use it as lora
send me the link
Oh yeah, i thought with newer versions of A1111 theyโd automatically put the lycoris extension in, yeah it works under lora but it canโt use all of the lycoris training data
@rustic raft I have most of the multiple faces prompt text squashed for now 
Ok nvm I was wrong about the Lycoris Lora thing, they did incorporate Lycoris directly into lora
Although if you want to force lycoris mode make sure to put lyco::1*yourmodel*
good to know for me too, learned another thing about A1111 
nope using fluffyrock as a model and using my prompts to get that style
Totally... not imagining that to become reality
chillin watchin the crops
congratz on helper buddy
thx! just trying to help keeping this channel clean from Discord TOS violations ๐ซก
awesome!
im not active enough to do this stuff x3
im more than happy being a contributor XD

Lora master of AIHub 
this is crazy
im just so fucking amazed how y'all r able to make shit that looks like me without fucking seeing my face
like i'm actually fucking blown away
this looks like a younger version of me
glad you like my efforts 
it's fucking crazy
well a day worth spent I would say 
now you have lots of PFPs to choose from

Proof?
Were you born with blue hair?
ofc
jjfkfjdjd
nah ppl aint born with anime hair but i got really into CRISPR tech a few years ago and it now grows blue
its not that complicated
its kinda like getting a tattoo
but its more complicated
if ur actually interested in knowing more u can read about the basics here
https://www.yourgenome.org/facts/what-is-crispr-cas9/#:~:text=CRISPR-Cas9 is a unique,buzz in the science world.
WHAT
That's crazy ๐ญ
Can I grow my hair to be transparent?
:3
i guess
but it would probably suck to maintain
like, it'd kinda be like seethrough but not "invisible"
Hope the CRISPR doesn't affect hair health or thickness
it CAN if you want
Can I translate your how to kiss boys guide??
I haven't even finished that one yet
trying out SVD
turns out, you don't need 24GB of vram if you use comfyui I guess
this shi cool
And taller version of you
(this is purely sarcasm and exaggerated due to seeing a wall of komodo)
Ayo? @wide mesa level 2 !!! 
alr more posting time
china
why does the chair look like that
this is fun
the face on this one lmfao
pain

i literally use comfyui cuz somehow it runs quickly on cpu and since im not depending on my igpu i can use all 14 gbs of my ram
gonna try sdxl turbo btw
7 gb is crazy tho..

my poor itnernet
mine is like 60 gb
yeah I remember the good old times when that was the case for me too, not 48 checkpoints to test 1 prompt with 
f
Wth
LMFAO
Spyro looking cool here ngl
What's that 3rd leg doin there 
Smacked ur ass like a drum now u got role
Not a chance
I say we eradicate the whole human race
What the hell is that ๐ญ๐ญ๐ญ
I will instead rule the world
Okay babe I go buy the battle pass now
Good for you

Bro what is that server icon for where that sticker resides ๐๐

โYou can be so flexible with a bodโ like mine!โ
/generate peter griffin gyaattt image
||/generate furry toes ultra realistic RTX dolby atmos||
naah sus prompt
bruh
that isnt a bod lmao
Donโt care lol
bro has an 8 pack
Thatโs what the AI gave it though
mhm
oh, ok
CivitAI or tensor.art
lot's of options
thank you ,I'm new In all of this
this scares me
fr
its ok
i meant to reply to u @left carbon

we all know this is true for @shadow bane too
fr
muscluar men

All my animal OCs in one shot
idk
honestly idk how to train lora's
and even if i did i prob couldn't be bothered
breakfast cereal mascots
im sorry but all look like fortnite skins
i cant keep that thought away
how to generate picture?
if you're a beginner, just use Bing Image Generator
to get a feel for prompts used for ai image generation
https://civitai.com/articles/2982 People can start linking my guide now as it is mostly finished
Pinned 
Thanks, it does still need some work to get to the nitty gritty of AI Image Generation
yeah step by step, all this stuff is complex af, so making sure it's all correct and understandable is more important 
At the minimum it tells people where to go and gives them options for what they want to use

food
โYou want flexible? Iโll give you flexible!โ
-colab
@fossil ingot wrong chat
sorry
is there a stable diffusion collab?
u can probably ask an @SDcreator
im not sure
oh okay
that's all right
@tawny peak im going to make a drop page with information on how to get started with download links to SD aswell as how to use the Bing bot.
can i use ur SD guide? this will become a -aiart command for @woven tulip
Update April 2023: Google has blocked usage of Stable Diffusion with a free Colab account. You need a paid plan to use this notebook.
dang
if you want to use SD it's probably best to do so locally, with LCM and Turbo you can get some cool results even with low end PCs
what PC do you have?
the online civit one
is there anything wrong with it?
atleast ur selfaware
๐
nah nothing, using SD online with CivitAI or tensor.art is fine
sometimes a curse...
is it lacking some stuff?
i know it doesnt have img2img
never used it in detail, so I don't know the feature pallette, I just know that it's fine for basic usage
oh okay
btw: https://stable-diffusion-art.com/how-to-install-comfyui/#Installing_ComfyUI_on_Mac_M1M2
if you have a M1 or M2 model
oh okay thanks
ill check it out
let me know how it worked out 
@shadow bane
im gonna ask u to help me set up the drop page for the -aiart command
seems like u know a lot

if ur down for it
i just need links
ehh sure, what do you have in mind?
JVG for local normal
online
local MAC
something like that?
also btw do you know what a seed does in stable diffusion?

yeah
just wanna make ur life easier
so u dont gotta copy paste shit every time ppl ask
you would know what to add here
seeds are the base for putting a number on the randomness, if you use the same seed with the same settings you will normally get the same image, if you change the seed, you will get a new random image for the settings you have
okay and what about CGF?
CFD*
CFG***
OOPS
how much will the AI follow to the letter what you wrote
low CFG AI generalizes more and dreams up wild things
high CFG AI takes words more literally and is more restricted in dreaming
Ayo? @shadow bane level 24 !!! 
oh okay and i understand that steps is how hard the ai works to generate it right?
what is clip steps?
clip
clip steps
that's a very vague description, it just means how much steps the algorithm is being run through the AI to calculate your image from the noise seed
100 steps doesn't make the image magically awesome, but more steps does give the algorithm the amount of times it needs to run to make something coherent from the noise
it depends/varies greatly on the sampler and scheduler combo since they decide how the noise is calculated into an image
I use SDXL checkpoints nowadays and they don't use that anymore, so meh
it's an old way to make the algorithm more vague towards your settings, most people that use SD1.5 just put it to -2(or 2 depending on your interface/UI) and never change it from that.
it's also more known as clip_skip
oh ok
ty
@shadow bane @left carbon @tawny peak since u three are really active here. what are some of the most frequently asked questions people ask you about ai art / making ai art?
im making a list of answers
promptsd
you can take some of the stuff just above here, those are very basic SD UI questions
people ask for prompts
ahhh right good idea!
done 
exavtly
probably training stuff too, like Yume does with Lora training
sadly not much knowledge in that myself, but maybe we can hit Yume up about it 
Whats up?
@quaint tiger
Ohh
yeah
I mean, people dont ask me alot, mostly what programs i use, however i will link the sources i have used to learn about lora training
Once im home that is
Theres some really useful stuff that people cud use
that would be great, please dm those to me
Will do

Since I can't reply to two messages at the same time i'll truncate it here. Yeah 100% go ahead and use my guide for it it'll be a whole lot better than copy pasting the link hundreds and hundreds of times. And the question I get asked a lot is basically "how did you do this" or "what program is that". I think a good question is "how can I update" or "what can I do to optimize the generation speed."
can someone make red circle from shape battle ai voice here is an icon of him
"Hi there. Would you like to sign my petition?"

triple miko won't hurt you. Triple miko :
use adetailer with eyes model
ahh okayy
Is there a way to provide reference images in order to generate images of specific anime characters?
I'm currently using Automatic1111 but open to switching
check for Loras on CivitAI
or train Loras yourself with Kohya_ss
use img2img, which restricts your options quite strongly
kk thanks
Hey baby yume I hope ur having a great day
holy grail those are awsome
bruh
Ayo? @granite flame level 1 !!! 
Bee attack
"HUG LIFE"
awwww 
I need some help
with?
um
i cant paste image for some reason
i wanna see if i can get someone to make the napoleon meme in vlandian gear for my clan
I trust this man
+1 Les
the trusting'
so you mean with a non-sdxl based model?
yeah sure, I'll do that tomorrow 
dogshit "Artists"
Literally an Overwatch pfp lol
This reaction had me rolling lmfao
Anyways back to the VoiceAI stuff where I belong I go wheeee
naruto on green platic mountain
SAKURA MIKOOOOO
what the hell man
more miko
beautiful girl holding cream bottle
f โฌโโฌใ( ยบ _ ยบใ)
"Hello! Everyone welcome back to my sponsor. Today I have this cream which does stuff and it gets rid of stuff. Thanks for watching"
butt nair cream
wanna do me next? i love the hyper-realism
Since I can't use the same Loras and Embeddings I just opted for a "add-detail" Lora on 0.5 strength for this SD1.5 image
it's the "same" model just not SDXL:
xxmix9realisticsdxl_v10 vs xxmix9realistic_v40
someone help this man
these are crazy
STABLE DIFFUSION BLOWS MY MIND EVERY TIME
had to face fix 2 times, far away faces in SD 1.5 are much harder to fix than SDXL
but man, there are just so many Loras missing in SDXL, like the Dyed hair Lora...

this was the initial image btw xD
Huge difference when u zoom in
since my SD 1.5 and SDXL WFs were different and incompatible, I build a completely new WF that is nice and compact, but basically does the same thing as the SDXL one. 
This is why I donโt want to get into stable diffusion
pov "anime girl" in prompt
Me when applio installation error
took me like 5 min

10 for all the little details like adding the checkpoint name to the filename, but that's nothing important
can I ask here for help?
yo ai is crazy
for stable diffusion
ye
oh hey, comeback of a classic! 
theyre not that bad
sdxl is crazy
this one a SDXL model?

yeah I'm confused too ๐
u said "
"
i dont want to do it again


What do u guys use for ai art?
the most common ones for local generation are A1111 and ComfyUI
Online it's google AI, Bing AI, CivitAI.com and Tensor.art
then there is a way to use google collab to generate images, which is a little grey area, Sztef is doing it at the moment for testing
Ok thank you
Ayo? @lusty nimbus level 1 !!! 
what's your PC/Laptop like hardware wise?
3060 ryzen 5600x
hey that's not too shabby, you can run locally just fine then
depends on your brain really, A1111 is UI based, ComfyUI is node based, like Unreal Engine or Blender
you can go to Stability Matrix and look at the screenshots of the different UIs that they offer, maybe that will tell you more:
https://github.com/LykosAI/StabilityMatrix#๏ธ-one-click-install-and-update-for-stable-diffusion-web-ui-packages
we also have a work in progress guide in the pinned messages
Sztef likes A1111 more, I enjoy ComfyUI more, since it works well with my noggin'
i mean i cant use comfy sadly there is no amd support๐ญ
Ok trying to learn more. New to all this tbh
when i was trying to render something it looked like this
no problem, you won't be an expert any time soon, it took me half a year almost to talk this confidently about all this 
Step by step
Yeah thanks for help
whooops, I remember now sryyy
Comfy only works on Nvidia?
which you thankfully have.
but there is a hacky way to get ComfyUI working on AMD, with ROCM, works best with Linux atm. Also some MAC M1/M2/M3 and Intel GPU users have got it working as well
always on linux ๐ญ
it will all probably get better the next few years, but yeah NVIDIA went ahead with tensor cores and CUDA
I worked 10 years in IT and neither back then, nor now do I have any interest in running Linux
enjoyment/time is horrible for me personally with Linux, so a big nono
Yeah tbh not a fan of Linux
Gn
#๐โai-images photo Range Rover

again?
u are lucky I dont want to spend all night sending white screens
Launch webui-user.bat instead
this image is surprisingly coherent other than that weird short lamp post by the wall
looks good
Ayo? @echo coral level 13 !!! 
is it possoble to generate realistic AI art with our logo/store on it?
Ayo? @ebon lichen level 1 !!! 
like this one, i want to make this girl dancing in front of store while holding a cup of coffee with my brand logo
Ayo? @left carbon level 36 !!! 
Thats a nice artstyle
Seems very familiar but I can't quite put my finger on jt
This is lovely lmao
Sleeping girl
I know why you find it familiar
How to use prompt for ai images
Search clipdrop on google or use bing
Okay sir
if it's just a random store/coffee/girl then that will be easy, if any of these are supposed to be specific, it becomes more complicated
if it's just one specific thing with random things, I might advise you to check/google out inpainting, that could help with that
then it depends on your PC if you can do that locally or you need online solutions
love how she is presenting one of the antlers xD
She was on hunting 
modern day picasso

unironically I really like these ngl
well taste is subjective after all, I like some odd things that most won't like 
its Billie Eilish done in the style of David Choe
Gigi Hadid and Rihanna
Hi <@&1159579982683701379>'s
we're working on an RVC Suite and Rentry site for Ilaria RVC + other projects and we need someone to help us create a banner 
if you're interested please dm me
What kind of banner exactly?
What are the dimensions?
What theme?
What style?
What does it need to contain?
The more information, the better
We need an "16:9 outpaint" for a 1:1 image"
- 1st image is source
- 2nd image is an example (it didn't come out right)
that is all
high quality, good lookin
the image needs to be in that style
If anyone got finger fixing tools, let me know lmao
Try using the bad_pictures and negative hands prompt combo
bad_pictures and what else?
negative_hand
Does that work for all Stable variants?
only SD1.5
Is there a variant of that prompt that could work in Stable XL?
i can't run XL on my laptop so haven't made one yet
you are using TIs/embeddings?
thankyouuuu!
just use adetailer
here is the wide one fixed:
some photoshop here and there and it's pretty perfect ๐
you forgot to add 8k HD hires, octane render, shizuki ryoumi, masterpiece, (best quality:1.4) ๐
nope, since its web hosted based not local.
still need something? looks like what Khazar did is what u needed?
well it was just an example image, so it's not like it was a finished product, also it's good to have more options, also switching up banners every now and then is not that unusual
i did not fully understand what he asked for with those 2 images, but i dont mind making a banner if they need
Ayo? @teal sphinx level 13 !!! 
yeah I too can't make out if all they want is just a extended image of the example image, or if they want images like that in general plus that...
At first I thought that they want both, but now you made me doubt it a little xD
||/generate skibidi gyatt ultra realistic||
no
Can you make it for me pleas
end it
RVC Chat can properly understand and fix images now
ok
trying to generate reference art for a pose and instead I get a janitor taking a piss and a mop noclipping.
lovely
Ayo? @vivid mango level 5 !!! 
/generate que
Hai pizzacool
I know you from the other server from a while ago
:b
pizzacool92, right?
i love trees
xd
I love trees ๐ค

The Greek God ares, wearing a helmet, red glowing eyes, shaded head, Roman armor, fingerless gloves, a glowing red spear, and a golden circular shield
Didn't really fit the prompt, but it looks cool
Least gigachad crab:
heyy, i actually only meant for the small image to be set to the correct size.
the large image that i uploaded was just an example of what we were trying to do but as you can see the pixels kinda flopped
16:9
ur funny
Looks like the artwork from some heavy metal song cover.
lol yeah im bad
i just wanted the first image extended
bc the quality of the one ila did was trash
Hey you made a game of find the matching pair at least
i made my profile with a bunch of stable diffusion and photoshop. stable video was a fkn godsend, i used it on the shit i already had and omg i love this
Ayo? @flat shuttle level 1 !!! 
its literally easy lol
A man with a strong body, a fierce and realistic face, fire coming from his eyes.
/generate cat flying a plane
so what about my fixed version? #๐โai-images message
@odd lance omg
that's a yes
Ilaria!!
OH MY FUCKING GOD
well..
im gonna replace the old logo ig..
even just that is a new pfp fr
ohio
Ayo? @mighty mulch level 1 !!! 
Would anybody know if Kohya_ss supports using .safetensor models to train LoRas? It seems to only accept ckpt files
Ayo? @hardy zinc level 1 !!! 
Since youโre on iPhone, try out draw things in the app store. Youโll have access to SD 1.5, 2.0, SDXL, SDXL Turbo. And no account needed as it uses your phone as the generator.
Yes it supports safetensors, your instance may be out of date.
You may have also gotten ip blocked for using multiple alt accounts
Nah Iโm good
Also no itโs not an IP ban
Other people are having this issue as well
huh ok, I'm using the version downloaded from StabilityMatrix
/generate cat couch
oh wait I think I was using the wrong button to select the file
SVD with rife
I got jump scared by AI
Yo Wsp
they really have to tell people that you canโt generate images here - https://civitai.com/articles/2982 use my guide if you want to generate images.
Can you create a guide for rvc chat's image generation?
RVC Chat isn't primarily used for image generation it's up to the character model and the chatting you do that generates the images. At least from what I've seen of course, I don't use RVC Chat. And even so it uses ComfyUI as it's backend for generating images. ComfyUI or Automatic1111 alone is much better for actual image generation.
Unless I'm completely wrong, and there is a way to change sampling settings and other aspects of the model, I guess I could
it's supposed to be simple for noobs to create good looking images so the sampler settings are limited (like the civitai online generator):
Ah, okay. Trying to install it right now. The easy install is broken at least for me
if you have conda installed you can just run conda-start.bat
I've got a few issues with RVC ComfyUI, Currently the process to getting to the point of getting the SD server started takes way longer and is more complex (if conda doesn't work). Second, the requirements.txt file installs the wrong version of torch, it installs 2.0.1+cu117 but requires torch 2.0.1+cu118 so things like xformers don't work. The torch problem isn't that hard for an experienced python user to fix, but for new users, it might be a put-off. Yes the UI is definitely simple in it's design, but it is missing crucial features to make good AI images like VAE, Clip Skip, Resolution Selection. It definitely could use some work, but the concept is there.
Thanks for testing. Do those VAE people create actually work? I don't notice that much difference between those and the default mse/ema ones.
The checkpoint you used i'm guessing had a baked VAE
yeah, I always bake the default VAE in my models
@dawn valley @shadow bane @lean plover
this is some deepdream type shit
oh shit im not mod anymore
gold star though because im epic
Oh you are FDG?
I heard your name before but haven't seen you yet ๐ญ
Hai FDG
hello
i was relatively important back on the old AI Hub discord but now i just hop in every week or so to say hi
Is this accidental?
Ok
i was seeing what would happen if i generated the image one time and then iterated over the image again with really small tiles of it
it made this trippy sort of effect
like a mosaic kind of
it reminds me of deepdream
I would love to hear the FDG lore one dah
Day
:3
all i did was write some guides and have a bot that ruined the server for a few days
i was almost nominated for admin but just barely lost
that.. is cursed
Anime creature
Ayo? @floral rover level 3 !!! 
What 
wendy?
nuts in yo mouth
smokey
A 20 years old teen wearing samurai clothes with 2 katans on his waist looking at the wide sea. Cartoon style

@lethal prairie
why magic? xD
shit actin like Rapunzel's hair
meh stylistic choice really 
but yeah, sad about the finger, the rest of the image turned out very stylish and nice
ye
Ayo? @sharp haven level 1 !!! 
Heres Shrek doing an onion mukbang because why not
it's funny how they are trying to bake the bing ai watermark into images now to prevent people from thinking the images are real
which ones do u use
Ayo? @lean plover level 4 !!! 
Bruh just realized
my racial diversity lora:
hi
Hai
w
erm hello im looking for some help with generating. im using a1111 webui and im trying out this sdxl anime illustration model, and the preview looks great but then the final result is awful. im not using any refiner, no clip skip, no loras, using the recommended vae and textual embeddings
Whatโs your negative prompt?
try using a refiner, turn your clip skip to 2, and what are your sampling parameters
"malformed, zombie, bad quality, low resolution"
What about normal prompt
Try these
the creator of the model recommended no refiner and no clip skip. my sampler is dpm++ 2m karras (as recommended) with 20 steps and 10 hires steps, i tried using both swinir and ersgan, and euler a and both got similarly bad results
"fragmented in shards of mirroring ice + by ziyun"
Your resolution may be too low, SDXL models are trained to work at a base resolution of 1024x1024. If your GPU can't handle higher resolutions, SD 1.5 or SDXL Turbo may be a solution.
i will try increasing it
Oh and I just checked the PNG Info, your VAE is also set wrong, it should be set to the SDXL VAE, which can be found here: https://huggingface.co/stabilityai/sdxl-vae/tree/main
oh thank you!
The file is sdxl_vae.safetensors
sus
i always knew that.....
@autumn moat look anime girls are real
i cant make now why anime girl arent real v2
ive converted the anime deniers with my wise words
their might pales to the power of anime women
all they can do is cower in fear
DarkPizzaXL coming in hot with another great interpretation of the prompt 

Image gen am working on I have 32 gb of ram lmao
it doesnโt seem to be using the gpu or the cpu that much, but itโs using all the ram. Weird
Iโm pretty sure that RTX 40 card will help with speeding up the generation time
It is using torch and torch vision
Building it from scratch?
Ayo? @tawny peak level 11 !!! 
It is set to use cuda if available and yes
Damn, props to you.
Thank you
Plan to make your own UI or run it through cli only
create a bioluminescent fish
Yo I have a question
Yes
sussy towers
How do you generate that?
Bing ai
thx
Ayo? @noble galleon level 1 !!! 
๐ ๐ ๐ ๐ ๐
Goofy a(-( lego dall-e 2
Lego Super Mario full hd,with green backround
@normal cradle @autumn moat
I have put these two white files in my models folder, is this how it should be done or do they belong in any of the other folders? It feels kinda off to have a bunch of folders and then two new added files seperately, or did I put them correct this way?
in
Alright thank you, I am very new to this and trying to set things up for the first time.
๐ซก
So I have just installed and prepped A111, and I have downloaded to models to test things and for now to work with. Do you by any chance know what next steps I should do now before making images? Or should it be good to go now?
nah u can just try to render something
Ayo? @brittle ocean level 30 !!! 
LET'S GOOOOOOOOOOOOO
mine look like this
but i forgot how to
do this things
@tawny peak heeeeeeeelp
but u can try to render something
I select an model on the top right where it says "stable diffusion checkpoint", and then use prompts as accurately and concrete as possible to craft an image, correct?
ah, head on over to the settings tab, User Interface, and under this section search for CLIP_Stop_at_last_layers and add it, restart the UI by closing the CMD window instead of the button at the top.
then clip skip will show up at the top of your UI
I still don't know why automaze has to put a reaction everytime the word "Clip" is mentioned.
oh now it doesn't do it
bruh
anyways it should look like this
yup
New thing unlocked
Achievement Unlocked: Clip Skip 
WOAH
And now I do prompts? ๐
Oh, so did you install SDXL models?

















