#🍥|anime
1 messages · Page 195 of 1
there are no bots here
Then how
Ok
here human ear hide hair
just putting 'ears' in the negative seems to at least hide the human ones
🤷♂️
girl with cat ears
Negative: ears, cat
model name?
blazingDrive V1.3 + LoRa`s
thank
look PNG-Info
there's no bot here. If you have a decent gpu, then try downloading local installation of Stable Diffusion. One of the most popular choice would be automatic1111 webui. You can probably google that. If not, you can try using dreamstudio by stability, you should be able to get some credits for use. The status of the bot in this server can be found #1047610792226340935 here.
some of my loras not showing...
anyone know what is this or how to fix it
not sure maybe like this ishttps://github.com/DominikDoom/a1111-sd-webui-tagcomplete/issues/207
have a same message... but extention work good
where?
Maybe not compitable with your current model
a1111 hides xl loras for sd1.5 and sd1.5 loras for xl
oh I see
You can disable that from settings if you want
@gloomy scroll you know how change size title preview img in models, LoRas galereys
Not rly but this could maybe do if you know css

hmm not sure
hmm how do we get dark background or dark images
its nearly impossible for me
how do we even get images like this, it dark
I need DARK
cdtuner extension,doesnt need a lora
interesting
🗿
Komi Can’t Communicate?
cool anime! 😁 👍
tell me pls how it use? not understand it in overview
So cooked
Just tried to move my git cloned webui folder to an external drive and broke it entirely
It's so gojover for my shit 💀
anime character
(Because I deleted the original folder)
your best way of moving, is usually just getting a new installation, then moving your embeddings/lora/models/upscalers, and others
you can try fixing by maybe deleting venv folder
I did
rip
So cooked 💀
stoopid
hmmm, i think it's fucked
Yeah nah it probably is
stupid hair
I decided to read it
https://www.reddit.com/r/StableDiffusion/comments/zi9dll/change_automatic_1111_location/
i found this old reddit post, the top comment might be worth trying
It's now cooking something
you might wanna try the post i linked
can't confirm if it works or not tho
Yeah that's what I was looking at before
All they did was drag and drop
I tried it initially
Did not work
this
I was considering trying that
But I currently have something cooking and
Doing something
read through the whole comment first before trying, cos someone added more stuff at the bottom
👍
automatic 1111 location change? hmm I think my case is different
I think this is not what you're looking for
my bat file look like
set A1111_HOME='e:/3D_Projects/stable-diffusion-webui_v_1/webui'
@REM
set VENV_DIR=%A1111_HOME%/venv
set COMMANDLINE_ARGS=%COMMANDLINE_ARGS%^
--ckpt-dir %A1111_HOME%/models/Stable-diffusion^
--hypernetwork-dir %A1111_HOME%/models/hypernetworks^
--embeddings-dir %A1111_HOME%/embeddings^
--lora-dir %A1111_HOME%/models/Lora^
--vae-dir %A1111_HOME%/models/VAE^
--realesrgan-models-path %A1111_HOME%/models/RealESRGAN^
--esrgan-models-path %A1111_HOME%/models/ESRGAN
call webui.bat```
Man
oh woow
try change your
eh? what do I need to change? it works fine tho
then ok
btw do you guys know a way to pin this .bat files to start menu?
as far as I know .bat file cant pin

try add in windows startup folder
ehh I dun need it to start with windows, I mean startup folder is doing run those apps while starting right?
yes
no I dun need it like that
I mean like this, so we can quickly click the file and launch it
right click pin to start ?
nope it doesnt show pin option to bat files
Desktop shortcut it is then
I dun use desktop
on the right click
Then taskbar
but how do we pinnnnn
clean, Im navigating all the things from start menu
Drag and drop
ugly
nope
hmm not for ME
you are the ugly one 
Try right click the bat file and create shortcut, then pin that
I did try that, for some reason it didnt work aat that time, lemme try again now
Google then. I had to do some trickery to get a bat file on the taskbar but do t remember how
win10 lets me just drag 'n drop it
I’m not at home so I can’t check it
deym
I just googled, this is the one that showed up lol
If you don’t figure it out in the next 2 hours I’ll let you know
@junior quest try this
Renaming your .cmd / .bat to .exe
Right-click on the * .exe and choose fix in the taskbar
Rename it back to .cmd / .bat
Shift + right click on the task bar icon (for the exe) and choose Properties
Changing the reference of * .exe to .cmd / .bat and change the icon as necessary
okay, now thats a masterpiece
thanks a lot

I just noticed
who can stand start menu in the middle

You're not normal I should've known sooner
you got that right
I'm used to it after switching to windows 11
isn't that a norm now in windows 11
we can change it I guess like windows 10
I moved it to the left on my desktop but I left it on my laptop since I usually use it in tablet mode
oh really, im definitely doing that then. But still sticking with my good old windows 10 for now
If I can't slap my cursor down left and click to open the menu
its not windows for me

yeah yeah I remember those good old days
I never use it anyway
@junior quest I like the raiden image
most of what I use it for, its in my desktop
just leaving things on the desktop is easier
I haven’t seen my desktop in months except the few times I restart
what are you
stuck on cmd screens all over the entire time
I hardly keep stuff open
Only browser and a game, its basically what I do in my pc
I usually have a lot open
Most of my extra windows are file explorer and command prompt windows, task manager, notepad
Chrome in the middle, chrome also on second monitor playing YouTube music or a stream
Plus discord
My main display is a 65” oled 4K tv though. Easy to see everything 😁
no comment
Yes but those that don't change it to be in the corner are psychos
za hando
hmm this is no actually doing dark, I mean its not what I was expecting, guess I have to stick to the lora
it can do white up to a completely dark pic,its just a matter of skill
but its doing almost like lower the brightness and increase the saturation
hmm thats not what I want
skill issue
saturation and brightness are different sliders u prob playin with those thats why
u only need to increase contrast-1 to increase darkness
I need like... okay lets say I need a dark room and there is a person in it, but for me no matter what I enter it just dont make a dark room, it always create somthing with lights and other stuff
nah its not like this how I want
I think I can even do this on PS
PS is post production this one modifies the latent noise
okay lets think bout this, this photo get dark, sure. but its not how I expect to be it. I need it to create a character with no lights just darkness and dark color environment
hmm im not confidence bout my english
hmm not kinda
anyway I'll figure out
thanks for the extention, its still can useful in my case
How?
contrast is for darkness,detail is like add detail lora
the color tab is for sending color to the latent noise
but how you take a dark?
increase contrast
con1 or con2?
con1
what about con2? when it work?
i think con2 applies darkness after picture gets generated but con1 applies it directly at latent noise
same and D1 and D2?
D1 is stronger D2 is lighter
what about it?
at how many samplin steps should it stop sending data to the latent noise
For some reason cdtuner doesn't work on forge for me even thought it should be fixed for forge
it works for me maybe u have another extension that cause conflicts with it
Do you have different versions?
its the same one
Torch also?
lemme open it again
I'm just wondering since it's throwing error from torch script
Maybe I need to update torch and xformers
But I still don’t quite understand the details.
and forge
you have 0.7
i am 0.10
and @native halo same have a Forge 0.10
Well I'll just update my stuff at some point
animatediff is the one that doesnt work on forge for me
for me "all-in-one", many errors everytime
also i enabled xformers just to test speed difference now i cant disable it if i disable it it gives me error
cant do it too
if u up the slider to the max it causes grain cuz of the latent noise
🤨
not sure what it max
also you have hr-enabled try to disable it or play with stop-step and hr-stop-step
what are we trying to do?
i love a forge! if you change something in the settings of some add-ons, then with the same seed you get a different result!
deym thats dope
um... whats this
?
this model?
yeah
Steps: 40, Sampler: Euler a, CFG scale: 11, Seed: 3816223248, Size: 1024x1024, Model hash: ab4ea1ebbe, Model: sugoiXL-v1.fp16, Clip skip: 2, Version: v1.7.0
hmm fp16 and this one different or smthing?
are you using forge?
tbf it takes a high ammount of skill to create a portrait
Hour five of no Stable Diffusion
mean while me: spamming some words

I'm so cooked
nope
yeah, spent all my efforts on girl, dark, soft
one day I'll prompt as good as you
maybe just reinstall it in new place?
thats the determination
I'm so cooked
@native halo "1girl cat outfit", bit suspicious this one isn't it 
if it's missing things, it didn't install everything properly
it shouldn't be that hard
So delete everything and clone again?
now do this one ( 1girl at astroworld statue,portrait,travis scott statue)
how slow is your internet? a git pull should take like 2 minutes tops

yeah, that can take a bit
I waited two hours to get an error
have you tried turning it off and on again
@junior quest I just regenerated that image in forge and it's exactly the same, so it's not my a1111 doing something, yours must have a setting somewhere that changes seeds
like cpu seed generation instead of gpu or something
maybe it CDTuner
could be
i dunno, everyone around here hides their metadata like it's a national secret or something 😏
u like it dont lie
I don't even know who that guy is 
save me 
get a faster potato 😄
Also send that contrast extension
i dont have one
The one for this
i'm not using any extensions, that's straight from the model
Nothing super special, just merged a few models together to get a specific look I was after for chibi characters. It can do more, but I haven't ful...
God fucking damnit
are you sure the cmd still has the focus ?
different error at least, that's called progress in IT
pretty sure --deepdanbooru is not a valid command line argument
just try starting it again, it should finish the install
I've had it on there since even before I had to nuke my entire thing
sure but it s not listed anywhere
So it's probably maybe not the issue
now thats fine
someone stole your vae
🤨
right it was vae, I didnt use vae, now it fixed after vae
needs more colour
👍
come here she will save u
go away ugly fake elf
perfect gen
@lyric zealot
bard died
wait what, the bot or bard itself?
new bard
now do old 
you can try their gemini ultra for a period of 2 months for free
kunoichi 7b local is the answer

can it roleplay as a spaghetti god ?
it can give u the spaghetti toppy
😄 Hail to the King - Avenged Sevenfold lyrics

ugly 
hes been using reddit for a whole week
doubt it, he still looks sane
lol

whose dad is this
@nova remnant
definately not mine
rin overrated 😒
is it must to use more than 1024 for sdxl?
true
agree
no
but in all seriousness, i didn't watch much of fate series so 
i dont actually have a problem with rin,it just usually triggers a bunch of people when i say that 😄
or Jack
Prompt
(wife:2.5)
fingers really fucked but...
guy`s tell me RTX 3060 12Gb for SDXL good and fast??
which char is this
yes,good fast and cheap
are you asking what the character you generated is from?
any 30/40 series is 'fast', it's the vram running out that slows down SD
or are you memeing about rem 
yes,its waifu education day
sakura
imma be honest, i have no idea
will you get one ? 

the funniest part about this is that, using google lens, your image only comes up with other ai sources
what mean 3x, TWIN, GAMMING?

AI will take over anime
praise ai
its Birdy Cephon
how dare u to forget Birdy /ban
who?
is it in the name of the gpu ?
asuna?
😄
GeForce RTX3060 12Gb GAMING OC 2.0
3X meaning 3fans cooler, twin I guess 2fan coolers, gaming is just branding or smthing
brand?
the guy who made the fights on naruto,JJK did that anime 😔
gigabye?
yes
gigabye lmao
the brand shouldn't matter too much
yeah, almost 40 years ago
i think change my 1660 6Gb on any 3060 12Gb
it matters smtimes not always tho, like there are few fake or underperformance gpus in market, but yeah if yu get gpu from a good trusty shop its fine I guess
well, the anime isnt that old
3060 12gb is like the best cheap choice 
thats... madara
:clueless:
eh? thats pain
hmmm
I don't watch I remember one game for playstation
or zombie yahiko?
so.. on amazon right now a 3060ti is $475, but a 4060 ti (16gb) is $430
not sure why you'd get the old version
4060 8Gb and 4060 12Gb what faster?
number go up = better
4060 16gb not 8gb
4070 12gb
hmm its not bout faster, only vram is better
ehh I dun recommend it, I mean I have a one
maybe 4070 super will be fine
are those card only can do work things? I mean can we do editing and other stuff too?
i dunno bout those cards
i mean, why not a K80? 24gb
because the k80 is old and is not RTX
me I got a TI 3050 4gig and I do fine 😄
a used quadro rtx 5000 is also old
and rtx only really matters if you're playing game son it
just keep a normal gaming card and add something for AI
the 4060 16gb is the best deal tbh
yeah
my 3080ti only had 12gb
and upgrading to a 4090 was not worth the price, i would have been better buying a 4060ti and running both
the quadro rtx 5000 is the same as a 2080 it uses the same chip only difference is that it has more ROPs and more cuda cores it performs the same
hmm not what I wanted but ok
but it's still old, and at least on amazon is more expensive than a 4060ti 16gb
so it's not a good choice unless you can find it cheaper
yeah, i'd pay the extra $70 for a new card in warranty
someone sold one the other day for 150 usd cuz it had a broken fan
i'd maybe buy the broken fan one for that
Buy the 4070 the 4060 sucks because of it 128bit memory bus,when it hits the limit it slows down,u can have 24gb or 48gb of vram but it doesnt matter when u crippled by a shitty 128bit mem bus
hm can't seem to get it to make her out of crystall not holding crystals or wearing them hmmms
@magic wave I heard you like Kobeni
eh? why me
I do 
Zelda too 


Battlebit 
Mikoo 
anyone know whats the model? there is no info bout this img
I'm enjoying my time in helldivers 2 
imma be honest, just looks like very generic anime SD1.5 based model
hmm
breakdomain
hmm where can I find sdxl control net models
sus, but need to try
the one i linked is old... dunno why the github still links to it
I just checked Illyasviel's hf
I rember someone making control loras for sdxl, but I can't rember who it was now.
Found them. They were official releases from stability. https://huggingface.co/stabilityai/control-lora/tree/main
only a few available though.
CDTuner gives error when I change brightness value
create an image of imran khan standing in wrestling ring with nawaz sharif.
no bot in this channel
Imran khan my beloved
how to create image ?
that depends, do you have a good GPU?
I'm on phone
Your best options will be to pay for a NovelAI or Niji Journey subscription then. There are also some free stable diffusion based sites that will give you a limited number of images per day such as PixAI and TensorArt.

Almost everyone here has stable diffusion installed locally and uses their own GPU to generate images.
ok thanks alot

Nubby being helpful so here's your reward

Do these work for you on discord mobile
For me they are like big pixelated images :b
They don't appear as small

Pixelated but it still works
idk, she's not a 2hu

hmm... it's like a triangle made of 4 triangles. some sort of tri-four-ce
i'm stumped
who could it be
That just makes that one even better
hmm I've been wondering, is that a lora? white hair vampire girl?
Nope
That's Yue. Vil's OC. She's just easy to prompt on most models.
is dat it?
1girl, white hair, blunt bangs, hime cut, sidelocks, long hair, blue eyes, fangs, pointy ears, skinny legs, oversized white hoodie, sleeves past wrists, choker
I see thanks, btw whos yue?
Original Character
This is Yue
eh I cant find her
Yeah I haven't uploaded her anywhere
original character means that vil made her up.
ohhhhhhh
Just my imagination 
I tthought she was a anime character

I'll imagine her with you
It's fine
meanwhile me:
some day anime girls will be real and then yue can be real too
That's where her name comes
yeah some day
I see
I made a friend for her
Time to start writing 1k pages of her lore which I can then some day feed to computer and get 3d printed waifu
Hopefully she doesn't find the gens in your PC

ABORT
don't prompt chibi on autism mix

XD
now which breakdomain? Im currently downloading 1st one
just makes children
The last 3 are probably all the same model (sd 1.5 based) the first one is the sdxl version.
I constantly keep thinking that I'll check my ai image folder and sent one of my new XL images but then I remember that this is family friendly server
And because of that Yue needs a hug now
Thank you for coming closer

jokes on you
long tail
do we have only one vae for sdxl?
I know someone is working on a new one, but I don't really see any need for it. The official one has good colors and can handle various artstyles well. I'd recommend just using the fp16-fixed version of the official VAE. https://huggingface.co/madebyollin/sdxl-vae-fp16-fix
yep
nice
The version that stability released produces a NaN in VAE when used at fp16 which forces it to be upcast to fp32 which increases vram usage and drastically slows down the vae encode and decode operations on most hardware.
That one just fixes the NaN error so that it can be used at fp16 with no discernable difference in quality.
oh I guess then this is it
Yep. NAI vae also had a similar problem for sd1.5 and was redistrubuted under about 2000 different names because everyone was uploading a copy of it with their model for a while back when vaes had to have the same filename as the model in order to manually load a vae in A1111.
I see I see
Back in my day~
whats ip adapter doing? following art style or smthing?
This is fantastic
anyone?
@junior quest #✍🏼|rules-and-tos
Is it a lora?
It's EasyFluffV11.2 with hll6.3-fluff-a9 lora for style. Saria is just prompted and then I drew the ocarina because the model couldn't get that right.
Ah ok
I am author of this lora btw https://civitai.com/models/183538/saria-ocarina-of-time-xl
0.7 weight is recommended
very nice!
https://huggingface.co/KSD2023/Nyan_Mix_230303_absurd2/tree/main
anyone used this vae before?
Here's the original of that image with metadata before fixes. Looks like I was wrong it was the a8 version of the hll lora, not a9.
based on the filesize, extension, and age I'd put money on that vae just being a renamed copy of NAI vae, maybe the blessed up version of it which gives higher contrast.
blessed2?
possibly. in either case those are just NAI vae with the contrast tweaked with this tool https://github.com/sALTaccount/VAE-BlessUp
not worth using due to an unpleasing texture it adds to the image, especially noticeable on foliage and incompatibility with fp16
This is the only vae that I use for 1.5 based anime models https://huggingface.co/nubby/kl-f8-anime2-blessed/blob/main/WD1-4-kl-f8-anime2-bless09.safetensors
It's the wd1.4 vae (kl-f8-anime2) with contrast reduced by 10% which solves the oversaturation and color bleeding that the original version had while keeping fp16 support.
I've heard this one is pretty good for illustrated styles, but haven't compared it yet. https://huggingface.co/RedRocket/furception_vae
Remilia grew up
not your average fridge 🙂
Spooki
ok back to wholesome
O, that is pretty cute :o
What model you using?
did I get banned from this server, I can never find it!
You can only find it when it wants to find you.
I have also image of cat sitting in armchair
That's not a cat. That's an elf wearing a cat hood hoodie.
yue can be anything she wants even a frog
We're so back
So back
The model
Is perfect
Now time to play around with the prompt
Oh boy, it's cooking
Which model?
sugoixl
Man, I cannot for the life of me get dark and soft to affect a full prompt how it would if they were the only two on the prompt
even with weighting?
Yeah
It's the too bright guy again
put 'light' in the negatives
I set them both to 1.5 at one point
Yeah I was considering just setting light to the negative
But something tells me it might not work
worked for me, but i wasn't making a large prompt
my negatives are usually empty, so dropping one or 2 terms in there is simple enough
ultimately, the same prompt is not going to work every time, you have to play with it and tweak things constantly unless you're happy with letting it do what it wants
if you want something specific, it has to be changed
a few days ago i was making a girl with white and black hair, i had to change the weights to balance the white/black, plus changing number of steps for the image for every single image for it to work right
don't know what you're trying for, but there's no bots here
@vital raptor make it animation 
can u afford it 🗿
I have also image of cat sitting in armchair

cat in an armchair?
here
Make it dragon girl


what's your prompt?
best quality, masterpiece, HDR, detailed eyes,
girl, anime, petite, (long ashen-silver hair), hair between eyes, red eyes, black oversized t-shirt, black and red earrings, pale skin, dark livingroom, standing, night, dimly lit,
soft, dark, upper body, looking at viewer, dynamic angle,
Negative prompt: chromatic aberration, cell-shaded, rimlight, light,
Steps: 37, Sampler: Euler a, CFG scale: 6, Seed: 882761691, Size: 640x760, Model hash: ab4ea1ebbe, Model: sugoixl_v10, Denoising strength: 0.19, Clip skip: 2, ADetailer model: face_yolov8n.pt, ADetailer confidence: 0.3, ADetailer dilate/erode: 4, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer version: 23.6.2, Hires upscale: 1.4, Hires upscaler: R-ESRGAN 4x+ Anime6B, Version: v1.7.0
best quality, masterpiece, HDR are all not needed anyway
you don't have to tell SD to make it better quality
it generates what it can generate
helps with some checkpoints because they are booru tags.
Definitly won't help with realistic stuff
it used to be needed a lot more, but pretty much any recent model it's not because they've all been merged into the best they can get
and on top of that, repeating concepts isn't helping much
also, this is a XL model, which doesn't need all that stuff anyway
Hm, alright time to test it without the first two prompts
Paler than a kid with leukemia.
well you told it to do a livingroom, those tend to have lamps
your prompt of dimly lit living room likes lamps
Very true
another monochrome coloring attempt
😄
I appreciate the white fucking background to my dark image
so, going through your prompt phrase by phrase, removing the red eyes and earings immediately got rid of the harsh llighting
💀
not 100%, but quite a lot
it might just not be something you can get rid of reliably though, it's in nearly every image
it's also a very common portrait lighting angle, so it's going to be in most source images
but then sometimes it just doesnt do it
SD when the power of the sun:
'soft shadows' got me this 😄
The power of the sun and random lights 💀
Oh shit and she even has the Sharingan too
Crazy
What's the prompt you used for the hair?
upper body,```
it's also part of just how anime is normally shaded and lit, it's going to be hard to get rid of it completely

What sampler are you running?
Happy Lunar (Chinese) New Year to those that celebrates that 🎆
DPM++ 3M SDE Karras
I've been on euler a the whole time
ew
Euler a tends to make smoother images with that model, so it just depends on what you want
this is with euler a
I actually think this model looks better with euler a, unless i'm doing chibi stuff, then the 3M sampler works better
Quite the dark background
hmm
Oh well time to play Minecraft
I'll continue my Stable Diffusion gambling later
CFG 4 (euler a) seems to tone it down a bit
still has lighting, but it's not quite as bad
No, no, that's perfect
so maybe just play with lower CFG
I just want the sun to implode and disappear on it
the model seems to do ok from 4-11 from my testing so far
Yooo
It's properly dark
Wait which prompt did you use
I've sent a few
Unless you mean the only one from today
the images all have the meta in them if you need it
it looks like 'upper body' might specifically be causing some of the issue
not always, as usual, but it seems to mess with it
but yeah, i think lowering the CFG is the majority of the solution
Hmm
12gb vram for a 768x960 on 1.5 on a1111,i heard on comfy is better optimized
yeah, animatediff is pretty vram heavy, and really hard to get anything decent from it

sometimes u get unlucky and no blinkin animation
spent a lot of time trying to make this image blink with animediff and the result was.. less than great
it changes the seed generation, and also destroys the quality
and that was the only working animation out of dozens i rendered, most just had flickering and shifting in the image but didnt actually animate anything
holy
its yu-gi-oh back from the make-up session and gym
wait
hes not called yu-gi-oh
I failed
thanks - that was with SDXL (aamXLanimemix)
jack and the bean stalk
Ohhh nice :o
model?
Its your model
autism
👍
@native halo

eh I dun have dat
go download autism and make dragon girls
souka , hmm i'll try
unlike sdxl normally, this model is pony based so it might like some negatives, mine are (worst quality, low quality), source filmmaker, sketch, lineart, chromatic aberration, unfinished, blurry, monochrome, censored
I believe you can do much less or nothing on autism, find what you like anyway thats the best for all models, always start your prompt with: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_anime, your prompt
I'm using pony, autism is more the generic anime style like what you just posted
, pony is more artsy
score??
cfg?
kinda creepy (ponyDiffusionV6XL_v6StartWithThisOne)
you can use normal settings other than the score prompt
its their quality prompts
there should be pnginfo embedded
oh, you need to have that extension
hold on, let me send the one that doesnt require that extension
um...what extention?
it was to counter that one time when discord stripe off all embedded data of images
and others
but i still download this, cos i can just copy and paste from webui and those with this extension can straight away see the pnginfo

lewds use comfy it doesn't work for a1111
The prompt is in the image...
Just type what you see, nothing complicated on that one >.>
eh but seems like they added adetailer
hmm so you said it doesnt work for a1111...will it work on comfy? I mean is there way to get prompt or its just doenst have the prompts embedding?
put a camisole, not sure it looks good or not
not an anime char /warn
I see
I think it doesnt cause I believe it shows a bunch of random text in a1111
then time to install comfy
real AI gen
👍
vtoob
Ugly 
dark backgrounds you say....
Try these loras together. Depending on the model you may be able to get away with having both as high as 1.0 strength.
https://huggingface.co/Aotsuyu/LoRA/blob/main/diffLora/8dim/[diffLora] lowexposurenoiseoffset.safetensors
https://huggingface.co/Aotsuyu/LoRA/blob/main/diffLora/8dim/[diffLora] dark.safetensors
Is it just me or did those embeds just implode and disappear
I deleted the embeds because they annoy me
Ah, I see
I wish I could make embeds off by default
images can get real dark
which one @nova remnant

UwU
I believe in you
You can do it alone!
I don't know the differences
needs more dark loras together
sigh hold on. let me find them.
3 in this one. I know I have more
I guess I only went up to 3 at nearly full strength
those works with sdxl? 
nyop. sd1.5 only




