#💬｜general-chat

1 messages · Page 68 of 1

fervent thunder
#

ok thanks so finally one more question? let's say I didn't download it but someone else downloaded it and used it then will I be in trouble or him?

#

???

potent spire
chrome jacinth
#

oof

#

I guess I need to remake all my graphics for my game then lol

potent spire
chrome jacinth
#

maan all this time I spent on with AI image generators has been for nothing 😭

potent spire
bleak bough
#

"Hello everyone, I'm new to software development and would greatly appreciate a mentor to guide me in my learning journey. If anyone experienced in javascript, react is available to provide guidance or recommend resources, please reach out to me. Thank you!"

potent spire
#

although for the future and serious stuff generated AI images would mostly not land on the canvas for me

#

itself*

potent spire
chrome jacinth
#

but it does make sense though cuz SD for example generates images by using copyrighted images so the generated images are technically x% copyrighted

#

but then what happens if I draw my own image but I look at other images for inspiration? Isn't that the same thing?

potent spire
#

so yeah it doesnt mean you will generate a spitten image of a copyrighted material

#

but the risk is there

chrome jacinth
#

so you are saying not all generated images are copyrighted but there is a risk

fervent thunder
potent spire
#

yeah because the AI generates images "randomly"

#

its a roulette if you will lol

chrome jacinth
#

can't the devs develop a feature that says how similar the generated image is to a copyrighted image in its database? 😛

fervent thunder
#

By the way, I'm sorry, my English is not very good, so I'm speaking in translation.

chrome jacinth
#

so we can have like a risk factor: x%

potent spire
#

if you provided him the image thats the question

potent spire
chrome jacinth
#

oh

#

so the only way for me to get away with it is to make sure nobody knows it was generated by AI 🤔

#

and that seems scary

potent spire
#

we speak about commercial use here ofc

#

if you generate images for fun its a different thing

#

and as long as its not too similar to a copyrighted image that might be in the AI image itself and you dont claim to be the owner and copyright holder of it

#

i mean you dont have copyright over it anyway

chrome jacinth
#

but since I've only used SD for my game it means it was for nothing cuz now I'm not allowed to upload it on steam and I can't ensure that I will be sued for AI art or not

potent spire
#

your best bet would be Adobe Firefly in the future (not now tho as long as its in beta) but even then youd have to retouch them enough so you can have copyright on them

#

if you wanna upload those to Steam

#

unless Steam changes something

#

you could and i dont and will not want to encourage you to do it sell your stuff anyway which some people definitelly do

#

there is just a risk tho

chrome jacinth
#

wdym by retouch?

potent spire
#

editing them enough for example

chrome jacinth
#

and enough is another scary word

potent spire
#

change form, colors, material, etc.

potent spire
chrome jacinth
#

interesting

potent spire
chrome jacinth
#

steam should have some inspection thing where you can send in your art to determine if they are allowed to use or not before you sell

warm junco
potent spire
chrome jacinth
#

oh that's cool

potent spire
#

generative AI stuff is (pretty) new

#

it will take time

chrome jacinth
#

yeah I just found out about it in 2022

potent spire
#

we dont exactly know what will be within the next 10 years for example

potent spire
chrome jacinth
#

lol

potent spire
#

i joined Midjourney end of july 2022

#

Stable Diffusion a bit later

chrome jacinth
#

but now I have another question which might be philosophical.
AI are trained on images (mostly copyrighted) and they generate images based on them.
If a human wants to draw a house, the human might google "houses" for inspiration and look at those images while drawing its own image.
Wouldn't the human be sued for generating/drawing an image based of a copyrighted image the same way an AI does it?

potent spire
#

i mean those cases definitelly happen where people get sued for basically making too similar images to copyrighted ones

chrome jacinth
#

oh 🤔

potent spire
#

ps: Not just making too similar but selling them

#

or claiming to be the owners

#

if you draw and paint spider-man similar character with suspiciously similar logo and look and say you are the owner of the character you might get in trouble

#

you can do fan art tho

#

eventually even publishing on art sites

#

can be risky but doesnt have to

#

Nintendo didnt react pleased when they found out about those "fan art" video games of Pokemon

#

even tho those werent really sold

chrome jacinth
#

what if I google "tree pixelart" and then accidentally draw a tree that's too similar

warm junco
potent spire
#

xD

potent spire
#

very unlikely

chrome jacinth
#

let's say hypotethically, in 1000 years we might have so many images that it will be impossible to not draw something that's too similar XD

potent spire
#

unless its some Harry Potter living tree or something with same spitten image and name xD

chrome jacinth
#

nah not a specific tree 😛

potent spire
#

nobody will sue you for trees i believe

warm junco
chrome jacinth
potent spire
#

like not just the tree but the image itself

chrome jacinth
#

so my AI art might be safe then 🤔

#

I would still ask Steam before uploading tho

potent spire
#

if i was the head of such a corporation i would act aggressively as well...not on fan art tho unless specific cases

blissful minnow
#

Hey,
Can Anyone here help me out regarding some deforum things? TIA

bleak matrix
#

Good morning, everyone! sunsmile How are we all today?

uneven harbor
#

Hey guys, anyone know what the best way to find someone for a project ( paid well of course), to help with prompts with SD? Not sure where the best place to post this is.

bleak matrix
compact nimbus
#

can i add the bot to my server?

compact nimbus
uneven harbor
#

Thank you @bleak matrix , prompting-help is a a bit too casual for what I'm looking for but I posted in community projects, hopefully I find someone

hybrid wolf
#

Is there any discord for ColmfyUI ?

bleak matrix
#

The bot is in this server at this time. To use it, please feel free to head over to any of the bot channels, and see the pinned channels to learn how the bot works if you don't know already

compact nimbus
#

huh got banned again lol

bleak matrix
#

So, what's everyone up to today?

potent spire
#

a bit of Firefly and generative fill fun in PS, a bit drawing, painting and modeling and later gaming 😄 How bout you? 😄

eager meteor
#

is this totally free？

peak portal
#

in real life

fervent thunder
#

When tf did I join this

warm junco
eager meteor
#

a little regret about a year subscribe for Midjourney hh And I deployed a pre-model of webui in mac-m1 8g, it’s so slow

warm junco
eager meteor
#

Yes

warm junco
#

Then SD should run on it

#

There is a Installation guide for it

eager meteor
#

I will check it out, thanks

warm junco
eager meteor
#

Oh there’s one line that I ignored: xforms are skipped

warm junco
eager meteor
#

No wonder I tried pip install but the message still shows up

river blaze
#

Is this free?

eager meteor
river blaze
#

Stable Foundation is better than midjourney I've noticed

river blaze
serene swan
#

comfyui vs auto1111?

eager meteor
#

It seems mac m1 couldn’t use gpu, I got this error: RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check

warm junco
eager meteor
#

It takes too much time for the webui，about 15 mins. tired of trying

serene swan
#

where can I find people to train characters? Anybody charging for this?

forest trout
#

Check places like Fiverr.

serene swan
#

gotcha

short tapir
#

Is there any way to bring up the old images I just made? It just finished generating like 30 for me, but then my power went out

forest trout
#

What program are you using to generate?

serene swan
#

should be saved in your txt2image folder in ur sd installation

#

oh I assumed auto1111 sry

short tapir
#

Saved? Like the images?

forest trout
#

What program are you using?

short tapir
#

Where is that file?

#

Stable diffusion.

#

With the auto thing

serene swan
#

it depends on which program ur using to generate

forest trout
#

Ok so automatic1111. If you navigate to the program, there's a folder called "outputs". It's in there.

short tapir
#

And no... The tex to image folder has a bunch of old ones.

#

But not anything even close to recent :/

serene swan
#

yea sry bout that. I forgot I set mine to save automatically

forest trout
#

Were you using img2img? Those are stored elsewhere.

short tapir
#

No.

#

Text to image

forest trout
#

Is your folder structure dumping the files into folders by date?

short tapir
#

Nvm

#

I found them 😄

#

For some reason I have like.. 4 text to image files

forest trout
#

It saves images by default so I'd be surprised if you couldn't recover them.

static tendon
grizzled fractal
#

HI where can i share my art works here _

forest trout
#

Any of the dreamer communities.

warm junco
#

Try diffusionbee maybe

willow coyote
ocean locust
#

I'm finally jumping into stable... the stress is palpable. Reminds me of trying to set up disco diffusion for the first time. I just want to make pictures, but there are so many words and numbers in my way

wintry stream
#

I'd definitely suggest at least giving it a try soontm ....for no particular reason 😉

shy lagoon
#

is it possible to create higher resolution pictures with your own models like 5k times 5k pixels or above , or do you allways have to use some upscaling tools ? Because when upscale to 10k pixels it allways look quiet fake even when the base picture looks realistic

royal falcon
#

Where can I start using and experimenting with Stability AI MODELS?

wintry stream
forest lily
forest lily
royal falcon
#

So is Stability mostly like Midjourney? i.e Image creation?

Does stability not do things like chatGPT, Bard, LangChain?

#

@wintry stream

forest lily
#

it just does image gen

#

If im wrong i blame the government for using 5g to brainwash me

river blaze
forest lily
#

No mor stable... Hahahahha

wintry stream
# royal falcon So is Stability mostly like Midjourney? i.e Image creation? Does stability not ...

There are quite a lot of different areas Stability focuses on & releases products for!
We have released image models, Language models, Animation, and more!

You might find these channels helpful to browse and take a look:
#✨｜sdxl
#1098025024541167646
#1108055793674227782
#1098959457964277831
#1042896447311454361

Plus we also have the blog at https://stability.ai/blog if you wanted a specific timeline to the releases themselves!

Stability AI
Announcements — Stability AI
forest lily
#

Wow thats not a lot of text. I thought it would be longer

karmic brook
forest lily
#

Wow ok

median peak
#

Is there a way to run stable diffusion in a 2gb vram graphics card?

hidden sundial
#

has anyone here made an AI anime like corridor

winged vine
#

xd

harsh fox
#

There is another option but I can't check now. Not at home

drifting quest
#

Hey hey everyone! Quick question:

Is there a way to not have all of the images come up when I'm inside one of the Bot Chat rooms? It uses up sooooo much data & I'm wondering if there's a way to do that & then I know I can do a simple search to find anything I've submitted.

Thanks!

forest trout
thorn knoll
drifting quest
#

Yup. I know how to do the searches but actually, a buddy of mine just showed me how to not let all the images come up and it helped a ton!

Thank you for the reply.

forest trout
#

Considering those images don't stay, I'd just recommend saving them. The bot will dm you the images you keep.

drifting quest
#

Yup. Thank you again!

short tapir
#

Question.

I know it's possible but I just can't remember how. I want to try and create some art that has like the Diablo style. How do I do that.

#

I've tried in my prompt to put in "drawn by michio okanura"

pulsar basin
#

help

#

how i add stable dreamer bot in my server

forest trout
pulsar basin
#

how add bot in my server
help plz

fervent thunder
#

you cant

#

gotta use it here

short tapir
vestal dew
#

I don't remember where

short tapir
silent fractal
#

How does Clip Drop's Style Transfer work? Because the results are amazing! I recently tried out Clip Drop (https://clipdrop.co/stable-diffusion) image generation and it was fantastic.

But what really blew my mind was their amazing style transfer capability. Can someone explain how do they do it in detail?
Or what kind of style transfer they are working with, are they adding the style after the image generation or do they alter the prompt before?

vestal dew
short tapir
#

Yeah. Followed exactly how they said

vestal dew
#

what lora is it?

thin parrot
#

Hi all

short tapir
vestal dew
#

It's not applying the LORA for me

#

It saved with an odd name D14BL02(1)CVTAI

#

same problem yesterday with LORAs...they wouldnt apply, but then today they applied

#

rebooted, but it's still not applying

harsh fox
#

Any tips for making the subject look down and the camera angle upwards? I can't seem to get it to work at all.

unkempt hatch
#

I want to do an x/y using Loras, but without the Loras at fulls strength. Can the strength be lowered in the x/y grid?

deft kraken
#

I'm new here and im wondering if anyone knows if stable diffusion can mix images on its own or needs move2move with it?

inland turret
#

Anyone can tell me how to do a art

vestal dew
vestal dew
warm junco
flat steeple
#

Is there a tutorial for doing a face-swap type thing? I have a picture where the face is not very good quality, and I have another picture where the face is very good quality -- I would like to swap the high quality face into the other picture

signal halo
#

How do i fine tune this to use for my own project

young ivy
#

muhuhurara

#

I forgot, if I have a dream (is it still called that?) account that's active, can I use it as an API for photoshop plugin?

vestal dew
shy lagoon
#

what do you guys think is the current best practice to upscale creations ? I have been the website "imgupscaler" upscaler but for creations with fine line and uneven details it often creates an unrealistic mesh on these arreas ?

young ivy
#

hmmmm dreamstudio still quite expensive on the credits

#

well not overly

#

but if I'm generate as much as I do locally 😄

#

and this is weird, prices on 3090 spiked in 2nd hand all of a sudden

young ivy
#

anyone know if stablestudio running with automatic backend is able to run extensions, lyco, lora, embed, hypernet, models, esrgans, samplers, models, control nets etc?

merry sluice
#

Pretty basic question here, but would anyone know a way to show proper branded outfits on an AI generated image, really specific items or clothing sets. I guess this would involve creating a LORA for that specific outfit for instance?

native tulip
young ivy
#

heh sometimes prompt guess makes no sense

sullen lodge
#

大家好啊

soft moss
#

is there a hosting ranking?

#

im thinking on forking money to rent a server or hosting for stable difussion and a google search is kinda overwhelming

wheat falcon
#

Well, guys, when will stable diffusion already beat midjourney?

frank badger
#

What is the best settings for AUTOMATIC111 with RTX 4090?

wheat falcon
modern ferry
#

can I turn around the same image that I get but it has to be same seed/image, is there a way to do that?

For example: I got a building image and I want to see the same building on the back, side etc.

hollow beacon
#

is there a training suite for SDXL yet?

covert radish
#

Are we currently able to use SDXL 0.9 through the Stability API? Anyone know the engine ID?

vestal dew
#

SDXL is available for download, for local use?

viscid jacinth
#

can we add any of the bot in our own server?

#

can you help me with this please?

#

can anyone?

wild steppe
grizzled fractal
#

jiojfaoiwe

#

hi what has pixiv become so bitchy_ nsfw ai content gets banned a lot

#

why____

vestal dew
#

Is SDXL 1 available tto DL now, like for A1111?

vestal dew
grizzled fractal
#

but ppl turn to patreon

vestal dew
#

it must be legal in Japan, but it's quite shocking

grizzled fractal
#

so it makes me think i need to censor all my content and post em on Pa

#

censor the fuck outta creators

vestal dew
#

Not sure. I am flabbergasted that people post such stuff there

grizzled fractal
#

or post em on opensea

#

some are not kids but got killed too

#

i think thats not the issue

vestal dew
#

Yes, I've read about that topo

#

Pixiv sent an email some time ago about not allowing AI anymore

#

maybe it has to do with people posting lots of AI images

#

if they post too much, it gets removed

#

I need to create an image of me that can express into the real world

karmic brook
#

hi, this server is not a place to facilitate intricate NSFW talk and or debate. Please tread careful when discussing heated topic here, as our server rules disallow NSFW as well. Just a heads up!

vestal dew
#

despite changing my schedule, telling people, they have triple booked me for the same hours at different places

vestal dew
grizzled fractal
#

we are artists

vestal dew
#

Im trying to find out if SDXL is available for dl yet, and using this web site to generate images, but it isn't cooperating

#

For a dragon b reathing fire, it gives an image of a dragon on fire

#

I posted a sample image in general with images

#

that's ON FIRE, not BREATHING FIRE

#

flying Dragon breathes fire onto the city

#

who would smaug have intimidated if he set himself on fire?

grizzled fractal
#

so where r u guys postin the art rg now_

#

if pixiv is being a btch

vestal dew
#

nowhere

#

everywhere seems to frown on AI

#

maybe try deviant art

#

you can make a gallery there...post your stuffs

grizzled fractal
#

na liebr on patreon

dawn sparrow
#

Hi 🙂 is there somewhere where I can advertise an freelance job opportunity?

fervent thunder
#

@delicate oxide je moet discord een ticket sturen

delicate oxide
#

Bit late brother

vestal dew
#

send a ticket for what?

vestal dew
hallow plaza
#

when trying to create a lora is it useful to have images with similar poses from the same subject if there are differences such has facial expressions and lighting?

forest trout
delicate oxide
grizzled fractal
#

can anyone tell me how 2 use the roop thing in sb?

slender field
#

seem like some people find it funny to vote up the worst of the generations...... i have many many images and nearly only the bad ones i made are voted up....... we need an channel where only all pictures are shown from each bot channel without text to only look at them and vote for them, some showdown winner for example have only 2 or 3 votes

grizzled fractal
#

some ppl just dont get it why ppl use ai art bc obvoiuosly this beauty does not exsist in the real world and those hypocrates try to moral singling the hsit otta it

#

so pathetic

#

moral police

fluid charm
#

do you have to download negative embeddings in order for them to work, or can you just copy paste them in? for example bad hands

crystal crown
wild steppe
solid geyser
#

Hey folks

#

how long does the stable diffusion's bot take in general to generate a single image here on discord ?

forest trout
#

Maybe 15 seconds?

#

Less than a minute but much longer than instant.

#

You could always just measure it yourself if you want to test it out.

solid geyser
#

well, I tried but my results didn't show up itself

#

hence asked

wild steppe
solid geyser
solid geyser
#

cuz I'm actually stuck in the middle of somewhere and just needed some help.

wild steppe
solid geyser
#

oh cool will head on and drop it there. thanks mate!

wild steppe
#

Cheers!

young ivy
#

hah seriously @wild steppe you guys blocked the word uh well I cant say it

#

cause it'll block me

#

but you blocked that but not 100 words that fall into it?

#

a 4 letter word...not even profanity

karmic brook
#

wasn't intentional blocked word

young ivy
#

nope

#

still doesnt work

#

Poor Al

#

can't even finish his name

karmic brook
#

ok one sec (should be fine)

inner crest
#

I would like to create an AI generated human and then create pictures of the same human but in different scenes, scenarios, however, I am new to this and have absolutely no idea about coding, I spent the last 30 hours trying to do this but nothing worked for me, could anyone PLEASE help me with this? I would appreciate it a lot🙏

young ivy
#

@karmic brook did you get like a generic list to start with?

wild steppe
young ivy
#

ah for a minute I thought that your list of banned words for the SDXL bot had somehow gotten applied to the whole server instead of just its channel

fluid charm
#

Hi, I've downloaded the 4x-UltraSharp upscaler and put it in the models > ESRGAN folder, but when I go to upscale my image I can't find it in the drop-down list

#

Does anybody know why?

broken smelt
north sentinel
flat steeple
#

Does anyone know why the img2img mode here is painting outside of my mask?

#

huh, can't seem to upload a photo

dim nova
#

I'm looking for a model that attempts to replicate phone camera quality, something that's not too well done. I'm currently trying to lower the quality while using https://civitai.com/models/55705/crystal-clear-photorealistic as a starting point. I'm open to any new models or prompt keywords to achieve this affect.

wooden briar
frank spruce
#

how do you decide whether to keep generating, img2img, or inpaint

stable pawn
#

I use the bot-1 here to make some images, and the best feeling is when you write a prompt and you then see some picking up that prompt and make it better, when you know you inspired someone. That feel better than being on some voting list or score board.

grim frigate
#

Ah anybody have an idea why sometimes hi-res fix resolution isn't exact?
Example 512x768 and I upscale to 750x1125 but the resulting image has a resolution of 744x1120. What's happening?

tough bough
#

Image dimensions must be multiples of 8. It actually used to give an error if they were not, so now it must just change them to the closest multiple.

grim frigate
vestal dew
#

choose dimensions that are multiples of 8

grim frigate
#

No, need to be toxic dude. I was just asking if I can bypass the restriction. I know what numbers are multiples by 8

vestal dew
#

Not toxic...that's a valid suggestion

#

stay with multiples of 8, then it won't change it

#

I posted the video because it's a fun song

tough bough
#

Your best bet might be to crop the image after you are done, assuming that there is nothing important in the 2 or 3 rows of pixels you would lose.

burnt cliff
#

I tell ya, sometimes it is so out of this world what the AI can generate, especially in the dark, horror department.....I think my eyes seen enough red for today lol~

vestal dew
#

it's frustrating because some parts are so good, and some bad...like it's a great image except for the third leg, or 7 fingered hand

burnt cliff
#

I agree, sometimes it comes out great but then that one other detailed messed it up

brave matrix
#

there really should be a channel for SD vs MJ battles

#

would be fun

lusty dirge
#

Can i use a CHECKPOINT TRAINED in model_path and a LORA in lora_dir together?

coral orchid
#

How can I enforce a text on an image?

#

I would like Latin numeral vi to appear on an image, how can I do that?

shadow igloo
#

anyway to use stable diffusion more as a filter than generation? As in, generating art in a certain style and wanting to convert another image to that style as well?

crimson belfry
#

on the stability AI site it says that we have the right to the images we generate, Im getting mixed mgs about the copyright with SDXL, Why is this not more clear?

midnight verge
#

Can you use muliple models at the same time?

pine fiber
#

we need a super pantheon

#

valhalla

inner crest
#

Pls help me someone🙏 Ihave created an image of a young girl standing in a flower field, how do I now create multiple variations of the same girl (with the same face) but in different scenarios?

sudden ruin
soft moss
#

i need a website that converts my broken inglish into the real words into the correct prompts. i just had to ask chat gpt for "how is this called" and he rplied "that is called panties"

#

my next question to chatgpt was "im trying to figure out the name of a pose, to use in stable diffusion as a promt. its when a person is sitting, leaning backwards and resting their upper weight on their hands "

crimson belfry
#

i would google poses, im not sure what that pose is called

soft moss
#

well, chat gpt says its called "leaning back" and i feel kinda dumb cause i named it on the question

crimson belfry
#

its difficult to name them because there are many styles in that same situation, It could be relaxed or exaggerated.

#

could be leaning backwards on one palm while having a drink in the other

#

fun times

soft moss
#

i need an image editor, a fre one, thats easy to use

ocean locust
#

photoshop is pretty cheap if you get the photography bundle, $12.99cdn per month for photoshop, lightroom and a few others. Not sure of the USD

junior venture
#

Hey guys, I’m looking for video editors who are proficient in the usage of deforum stable diffusion to make videos similar to the ones trending on TikTok. Dm if interested.

karmic granite
#

I wanna thank the stable foundation and the entire stable diffusion community. Learning this tool has been so much fun and ive had so many awesome resources to learn from through out. Im just happy to be apart of this awesome open source movement

mighty delta
#

hello guys

wild steppe
crisp dawn
#

can anyone help me

wild steppe
crisp dawn
#

if I use any textual inversion it is like that

#

everything in place

wild steppe
cosmic light
#

Hellow

west jasper
#

hey guys after some really interesting conversations with some folks on here i have a questionnaire! i would greatly appreciate anyone to fill it out. should take 3-5 mins. thank you!!! ❤️ : https://forms.gle/kAWVExrCVd71mN9q9

faint garnet
#

how do y'all avoid deformed hands lol

fervent thunder
#

Hey anyone need a custum stable diffution model dm me

pallid harness
#

Hey there, is there a way to have stable diffusion in private servers?

cosmic light
south veldt
#

hi ppl, am new here, have some ClipDrop related questions. Is there any place around here where I can find CD_SDXL support?

wild steppe
south veldt
wild steppe
pallid harness
young ivy
#

You meant SDXL right? not just SD?

wild steppe
#

For @radiant meadow? Yes

young ivy
#

Hmm invokeai coming out with nodes soon, I think gyre ai is too except theyll communicate with photoshop

hasty hornet
#

🫠 why tho

split plover
#

Hey guys-

Nathan here from Landslide Network, we're in the web3 space. We are bringing IBC (from Cosmos) to Avalanche.

Where is a good place to start in the stability AI space? I'm very familiar with GPT-4, but I'd like to understand what type of models and work is going on in the web3/stability space.

I'd like to use Stability AI to help us integrate existing Cosmos chains with our opensource SDK to bridge it to AVAX.

hasty hornet
surreal perchBOT
split plover
hasty hornet
split plover
#

We want to use LLMs to help increase adoption from existing Cosmos Dapps

hasty hornet
#

try it, you'll see

Last time I tested it it wasn't even close to gpt3.5.
Idk if it got better since then

#

adoption from existing Cosmos Dapps
-no idea what those words mean tbh

split plover
split plover
hasty hornet
#

ask in it's channel, there should be people more familiar with it, idk

high ruin
#

I am new to this what is the best one click install SD for creating people models that works with AMD video cards I have a RX 6600 XT ??

fervent thunder
#

Hey everyone

trim minnow
#

Is there a way to use a photo as base image to the ai generation?

frank spruce
trim minnow
frank spruce
inland path
#

Hi Friends, just wondering if someone knows why my Stable diffusion is always generating green images. I'm aware that this is common on the GTX 1660 series. I've seen some fixes but nothing seems to work. Can anyone help?

lucid void
lucid void
austere plank
#

Hi guys i need to ask something.
i Have Automatic1111 but some time ago i saw an engine with better ui and better ability to outpaint (not Inifinity) and now i can not remmeber what that was
So if anybody know what i have in mind i would love to know the name 🙂

#

And still i think i have outdated A1111

broken smelt
#

invokeai mabey?

hasty hornet
snow minnow
#

Hi everyone, what are you all doing?

novel fog
#

voting A or B all day

inner dagger
scenic gate
#

Where can I see landscapes

hasty hornet
#

thinking outside

analog cove
#

😂😂

#

Can I get some one to help me with generating images a1111

inland path
snow minnow
#

Hi everyone, what are you all doing?

zenith harness
#

Does the script have a way to give weights to prompt terms?

bleak matrix
#

Good morning, everyone! sunsmile

#

How are we all today?

bleak matrix
cyan comet
#

How much of it is img to img?

pearl minnow
#

Its been 14 hours and the 3d inpainting mesh is still running under depth map tab. Time to give up or this is normal?

fervent thunder
#

About this announcement: #📣｜announcements message
I think it follows the prompt more than the doodle. And if that's the case, I would rather just do a prompt, without a doodle.

compact nimbus
#

is anyone here on the mj server?

arctic lantern
#

I like how stable is taking me like 8 minutes for a normal picture at default settings

novel fog
#

weird question but will there be a end result about which person voted highest?

vestal dew
#

yes

wise belfry
#

hey all good morning, got a question. does anyone know how to get the styles for sdxl to work with comfyui? or is that only a feature supported by the bot?

vestal dew
#

SDXL is not yet released

wise belfry
#

oh, yeah. my bad

#

would it be possible in theory?

wise belfry
#

i would love to know if the styles are built into the model itself, or if a prompt is being injected by the bot before generation

crimson belfry
#

so everyone is releasing sdxl 0.9 on thier platforms, Are we able to use our images for commercial uses?

#

playground ai just released it

crimson belfry
#

What is happening in the courts will not stop AI

#

This is why it is great to see open source SD

#

They told us that this will be disruptive and we are now seeing the ones disrupted start to fight back.

waxen pasture
#

Is there a way to report images that are nsfw? I am pretty sure this server doesn't really condone that...

crimson belfry
#

Mega corps will try and stop us from being free to express ourselves

crimson belfry
waxen pasture
#

Because it's breaking the server rules.

crimson belfry
#

they dont control that if that is what you are asking

#

ive generated some leonardo da vinci style stuff and gotten some nsfw pop ups before

waxen pasture
#

Ok, thanks

crimson belfry
#

Its an ai and im sure that it cant be perfect but if you see people breaking the nsfw rules on purpose inside the prompts i would say something

crimson belfry
waxen pasture
crimson belfry
#

yeah i would report them

waxen pasture
#

But how exactly?

#

I feel like you used to be able to react with a warning emoji to report but you can't react in those channels

crimson belfry
#

i think you can report to staff in the 3 dots on the right side of posts

#

im not sure but i do know they have a report system

waxen pasture
#

Alright, I will see if I can find those gens again

#

There isnt a way to report it for nsfw stuff, and I think it's idscords built in reporting system, not something for this server specifically. Oh well.

crimson belfry
#

there is a report to staff feature in the apps section

#

in the 3 dots on the right side of posts

crimson belfry
waxen pasture
#

Well, it was in the showdown channel and that just got reset :/

eternal violet
#

wen stable doodle for local installs? 🙂

#

what a great meme tool

vestal dew
#

are they abusing sdxl?

prisma elk
#

Hey, I'm getting this issue pop-up when using stable diffusion and I can't find how to fix it. Any help: OutOfMemoryError: CUDA out of memory. Tried to allocate 9.90 GiB (GPU 0; 11.00 GiB total capacity; 2.22 GiB already allocated; 7.17 GiB free; 2.24 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Time taken: 22.39sTorch active/reserved: 2317/2344 MiB, Sys VRAM: 11264/11264 MiB (100.0%)

vestal dew
#

are you running with the lowvram parameter?

prisma elk
#

Yeah, the issue persists

crisp dawn
ebon sinew
#

The command line

primal storm
#

Why do some images not show up?

safe sable
#

is there a way to evaluate models like realistic vision or F222 and compare them?

unkempt hatch
#

so you can extract a lora from any model, right? but what about models that were trained on top of merges? Something like the new Photon.

Could you extract is as a model a minus the 1.5 based, and have the entire cumulative training in lora form?

safe sable
#

just simple performance evaluation

#

like F1, AUC metrics

serene pasture
#

hello 👋 is there a staff here I can talk to about the stability.ai website? There's a bug I'd like to report.

karmic brook
serene pasture
brazen orbit
#

hi

#

im og

#

yes

frank musk
#

oi oi oi

brazen orbit
#

how do i use the doodle

#

and how do i get old style og

jovial sable
#

Hi guys, is there a model for Vampire Hunter D?

brazen orbit
#

why

#

same prompt

#

why is bot doing stock images so much

#

NWM

#

NVM

daring turtle
#

Hey 👋
Which stable diffusion models are for commercial use?
No license issues. ??

bleak matrix
#

Good evening, everyone. How are we all doing?

frank spruce
daring turtle
#

@frank spruce 😊

keen birch
#

how much free space do i need to install a dual boot linux alongside win10

#

i have about 200gb free that should be enough right

soft moss
#

linux is very "disk space friendly" if you know what you are doing and take some time to administer some stuff (when it inevitably gets bloated)

#

200GB should be more than enough, unless you start installing every software under the sun and dont un-install it afterwards

#

i mean, i usually allocate 32 gb for the operative system (this does not count user files, like movies, and SD models)

fervent thunder
#

ing for good placners or creativee proep up with ai art stuff on internet and can direct to rod inspirations a

broken smelt
#

theyre the highest voted for pictures from the sdxl bot

#

has the prompt too

forest trout
vestal dew
tidal lion
#

Hello friends . Does anyone know which dreambooth colab uses UI??

#

I'm following the instructions of training with a lot of data and in the manual, rank = 128 seems to be only available on the dreambooth version with ui eg kohya_ss on local

#

Anyone know any dreambooth version similar to local kohya on colab please let me know

final dragon
#

Anyone around to voice chat to help me learn more about stable diffusion?

#

I just joined

pale latch
#

wonder what Stability thinks of Adobe's content protection system that they want to integrate with, now that Adobe has revealed their intentions and what they're lobbying for

#

asking the us government to create laws that copyright entire styles

vestal flower
#

What are the best settings to have EVERYTHING large on a different drive while my core SD App sits on a smaller fast SSD

#

?

#

I have Characters/models on a different drive

#

set COMMANDLINE_ARGS= --share@#$$@#$ --opt-sdp-no-mem-attention --no-half-vae --opt-channelslast --update-check --autolaunch --ckpt-dir "S:\Stable Diffusion\models\Stable-diffusion"

#

Maybe this is already correct? are there other --_ _ _ _ -dir "S:" I can set to my S drive?

warm junco
potent spire
cobalt tulip
#

Hi I need help

#

I am trying to create images but its not happening can someone tell me where i am doing mistake?

lavish lake
cobalt tulip
#

bot 1

#

what should i type?

lavish lake
#

the correct syntax starting with dream

cobalt tulip
#

I am typing imagine prompt and the content i want

lavish lake
#

imagine is for midjourney hehehe, try dream

cobalt tulip
#

but its not happening

#

Ok only dream?

lavish lake
#

slash dream

cobalt tulip
#

Ok i am trying with this

#

wait a sex

#

sec*

lavish lake
cobalt tulip
#

thank you its working now

#

thanks

gaunt belfry
#

im using colab
So i have open pose editor, according to the video i have leave the pre processor tab saying "none" and select a model.

I dont have any models and I dont think i can download them locally like that guy is suggesting.

Any tips

twin aurora
#

hi i am using stable diffusion using Automatic1111

#

at 1st when i start the application i can generate upto 4 images with no problem then i get out of memory error
saying that no space left in my GPU (GTX 1650 4gb) how can i resolve this
Plz help

gaunt belfry
#

I NEED THIS ON collab

#

ControlNet-v1-1
/
control_v11p_sd15_openpose.pth

cosmic light
#

Imo there should be a way to upload a sketched image into the new stable doodle

#

Rather than needing to draw in the ui

cosmic light
#

4GB is very low, if you have access to a different graphics card with more memory that would be preferable

#

Also, avoid large batch sizes to avoid out of memory errors

smoky palm
#

where do i go to talk about setup discussions etc?

placid kettle
#

I want to merge two cars into one unique looking car. whats the smartest way achiving this?

thorn pilot
thorn pilot
placid kettle
#

but the results suck

thorn pilot
#

which model are you using?

placid kettle
#

I trained my own with dreambooth but right now I use the stanard one (SD15NewVAEpruned.ckpt [27a4ac756c])

thorn pilot
placid kettle
fervent thunder
#

Hey, whats the min req gpu for sd?

vestal dew
#

car [(dune buggy:1):(vw bug:1):7]

The last number is the number of generations out of the total that go to the first thing. I have 20, so in this case it's 7 for dune buggy and 13 for vw bug

vestal flower
warm junco
#

normaly i would recommend only using --xformers OR --opt-sdp-attention for the best speed

#

and for your question there are also:
--lora-dir and --embeddings-dir if you want to move them too

vestal flower
#

This is good info... Somewhere online it said this might be better for my GPU... I have been thinking of trying Xformers again

vestal dew
#

be careful with xformers. It's more than meets the eye.

sage gorge
#

can I copy paste prompts?

nova hawk
#

Hi guys, does anyone has advices to have the same quality than warp fusion on stable diffusion ? Or at least get closer to ?

cosmic light
#

So just to clarify, what IS xformers? Does it just like work a generative process into a two for one deal or..

broken gull
hasty hornet
signal lynx
#

hi guys ✌️
I m kinda stuck with a picture. Its 90 % finished, but I cant figure out how to get some last details right and need help.

Where would be a good place to share my progress to get some feedback?
Thanks

turbid plinth
#

Hello every one. When I start Automate 1111 for Stable Diffusion 1.5, there is many time update when starting command screen webui-user, and some times that days it s write" enumerate... etc..." "=> new branch : SD XL" : what it means ? New functions are arrived already or it's for an extension, or for prepare for the arrival of SD XL? If yes for last : it will be in my same Automate 1111 installation by auto update ?

serene swan
#

so.... hows the patreon pipeline going for ai artists these days? Has anyone been harrassed by the luds, or is it dying down?

olive zealot
#

Anyone going to DEF CON?

#

If so, there's a BTV/AI Village Pool Party

fervent thunder
#

Hey can I run sd on my laptop with 4gb NVIDIA RTX 3050?

warm junco
warm junco
#

These are needed if you installed automatic1111 webui

delicate remnant
#

I have noticed that stable diffusion has trouble accurately depicting many kinds of vegetables. Anyone else notice this? Is there a stable vegetable model?

turbid plinth
warm junco
#

no problem!

midnight verge
#

Why do people use stable diffusion and generate images?

#

Earn money?

broken smelt
#

same reason people like to paint

#

its fun

slim summit
#

what's the longest day of the year

ebon sinew
#

June 21 in the northern hemisphere

cosmic light
cosmic light
wicked granite
#

Hey

fervent thunder
vernal storm
#

Hey hey.. very new to this and a bit overwhelmed with just how much information I need to catch up on to get up to speed.

#

In the short term, I'm eager to produce images based on a specific art style to generate characters that have a very specific aesthetic from the 80's/90's.

#

Do people offer commission services here?

tulip yoke
slim bane
#

Is API down?

broken smelt
vernal storm
vestal dew
#

hmmm...using vladomatic. Downloaded SDXL .9. When selected, error: model not loaded

barren sentinel
#

question is it possible to use stable diffusion offline?

pale latch
coral lark
#

StablityAI supporting surveillance tech?

tiny swift
#

for clip drop function, once I clip the image from my phone, how do I drop it to my PC? is there a tutorial on this?

minor crow
#

hi

hollow flame
#

Does anyone know how I can use gimp for basic inpainting? I just want to be able to use the lasso for precision instead of the automatic1111 inpainting UI

vernal storm
hollow flame
vernal storm
noble lily
#

anyone else getting poor results from bot past few days? either image. trying to generate realistic portrait photos of women

fervent thunder
#

I'm trying to install Stable Diffusion with Google Colab and I keep running into this error. It's saying the file doesn't exist when it does.

python3: can't open file '/content/gdrive//sd/stable-diffusion-/webui.py': [Errno 2] No such file or directory

unkempt hatch
#

any practical use to putting words with negative weight into the negatives?

e.g. I want a slender person, and add into the negatives (skinny, slender:-1.3)

chilly sparrow
#

Hey!

broken smelt
bleak matrix
#

Good morning, everyone! How are we all today?

buoyant remnant
#

guys can anyone lend me some help

#

ImportError: DLL load failed while importing cv2: The specified module could not be found.

#

this is the error i am getting

compact nimbus
bleak matrix
compact nimbus
compact nimbus
astral goblet
#

it might at times. you'll notice this when you are doing a high memory operations like large batches or high resolutions, and it goes extremely slow. It gets very crash happy when it starts playing in the realm of shared mainbus memory too. Since it has to send commands over the mainbus and use the system memory, that's where the huge slowdown comes in.

compact nimbus
#

i have found some people that are there here

unique hamlet
#

Any tips/models/extensions useful to generate closeups of hands? 👋

astral goblet
unique hamlet
astral goblet
#

openpose has specific nodes just for hands that the model understands

bleak matrix
astral goblet
#

there's also a preprocessor that is trained specifically for hands

unique hamlet
bleak matrix
astral goblet
# unique hamlet Thanks. Which one is the preprocessor?

the extension has it all built in. when you select openpose it'll show in the dropdowns. the preprocessor will look at a photo of a hand, and create an openpose image from it. If you've got your own custom posed hand as an openpose image already, select none for preprocessing

compact nimbus
bleak matrix
frank spruce
#

is there any practical difference from putting a negative weight into the positive field and putting the same term with a positive weight in the negative field

neat silo
#

Does anyone know a potential reason for my stable diffusion not saving anything?

#

it loads the image

#

and as soon as it's done it vanishes

#

and there's no way to save it

#

the directory that it should be saving to doesn't exist which is even weirder

ancient condor
#

Welp, it still has issues generating "wingless dragons". It keeps giving them wings.

neat silo
#

lmao, to be fair, the source material for wingless dragons must be small

#

does anyone know of a model that doesn't require your graphics card to support "half type"?

lost marsh
#

Are all the images we create using Dream shared real time or am I the only one that sees what I create? I mean do you have to share it for it to become public?

warm junco
warm junco
neat silo
warm junco
neat silo
#

lemme check, it's quite old

#

it's an NVIDIA GeForce GTX 1660

#

I've been meaning to get a new one

warm junco
neat silo
#

ah right

warm junco
# neat silo ah right

You need to edit the webui-user.bat and at the line Commandline_ARGS= you add: --xformers --medvram --no-half --precision full --autolaunch

neat silo
#

cheers dude

warm junco
#

No problem

neat silo
foggy spear
#

salut

#

pas mal

#

pas mal

#

et vous monsieur ?

#

ou dame-moiselle

slender field
thick bone
warm junco
vestal dew
#

I wasnt using it because I thought it sent launch codes to a silo somewhere

hidden pivot
#

monster-labs/control_v1p_sd15_qrcode_monster is insane for qrcode renders

sage lion
#

so how would someone get sketch to image on Local system? or is it even available yet?

warm junco
sage lion
#

would sketching be better or drawing out an image, saving it, and using the Image to Image?

#

beacuse installing ControlNet looks complicated

warm junco
#

Or you can do sketch, controlnet to image and use that for img2img

sage lion
#

oh thats a good idea, Thank you. Trying to download the models now, but I dont know if i can. they are huge.

warm junco
sage lion
#

i was getting them off huggingFace. are this not the right ones?

#

yea no, i dont know whats going on now i fked up something and i dont wanna troubleshoot it. -.- i give up.

warm junco
warm junco
sage lion
#

all it is doing is inverting the picture from white to black

frank spruce
#

yeah im literally a CS major and linux user and they poohpooh me for not being nerdy enough and willing to jump through hoops to get my install to work, and when i complain they tell me i'm simply ungrateful, it's like, no, if you want your product to be accessible, you have to change this process, deal with it

sage lion
#

i cant post image

frank spruce
#

like to use canny you need both preprocessor and controlnet model that are named canny

sage lion
#

im using scribble_pidinet as prepros, and sd15_scribble as model.

frank spruce
#

okay so does the preview look good in the controlnet panel

#

does it look like an appropriate corresponding scribble

#

and if that's good then it might be an issue of CFG or controlnet weight or something messing up the final image

sage lion
frank spruce
#

looks good, so how does it affect the final image then

#

you can also experiment with other pre-processors, ive had luck with simply inverting the colors of my sketch or using the "realistic sketch" function, try them out sometime

#

if you end up with excessive lines, try removing the controlnet like 1/3 of the way through, or reducing its weight by half

#

and dont forget your prompt and model still matter

sage lion
#

doesnt feel like the controlnet is adding anything it just does what it wants

frank spruce
#

check your python console window, the black and white text thingy

#

look for errors in controlnet, if any error occurs controlnet will look like it did nothing

#

if no error check appropriate weight, start, and end values

#

and obviously enable it

#

also the way controlnet adds lines and such might be different from how you might imagine

sage lion
#

im goiing to watch someone else do this, because the images created dont look anywhere close to anything im using as reference.

#

thank you for all your help and suggestions, but im just going to work on something else now.

fluid summit
#

im trying to install sdxl in automatic 1111 i keep getting Image Browser: ImageReward is not installed, cannot be used.
Using SDXL 0.9
Loading model stabilityai/stable-diffusion-xl-base-0.9
vae\diffusion_pytorch_model.safetensors not found
Loading model stabilityai/stable-diffusion-xl-refiner-0.9
vae\diffusion_pytorch_model.safetensors not found

eternal violet
#

Don't think it works in a1111 yet. Only comfy right?

vestal dew
#

it works in a1111 and Vladomatic

#

it has to have the model, secondary model, and then the SD run in a different mode

#

diffusers

#

probably easier to copy your whole A111 dir to run it for sdxl from the copy, because it also uninstalls controlnet extension

waxen flint
#

Hello! New here - just saying hi 🙂

pale latch
#

oh thanks for updating a whole day after i reply discord. thanks for showing me that when i brought my pc out of sleep. sigh.

sage lion
wooden tangle
#

does it take days to get a deforum video?

pale latch
#

np heh. i thought it was happening now. i never read time stamps

wooden tangle
#

wait if i use google collab would that make the video faster?

#

to generate

whole mist
unkempt hatch
#

once a lora has been trained, it retains no info of the original images, right?

asking because, if you were to include the same pic twice in a dataset, it'd overtrain that pic. but what if you trained 2 separate loras, which each had a pic in common, would combining/merging them cause some kind of problem there?

astral goblet
unkempt hatch
#

Like, a funny hats model, and a top-hats model. I want to combine. They both contained a specific funny top hat pic

cursive sparrow
#

/dream猫

#

/imagine猫

late holly
#

sup yall

valid sparrow
#

hey guys, i want to download the latest, newest, best version of SD.
since there are so many versions on the internet, Idk which one to download, can someone help me?

pallid knot
#

copy it and remove all space

valid sparrow
#

guys
whats the problem here:
"venv "C:\SD Files\stable-diffusion-webui-master\venv\Scripts\Python.exe"
The system cannot find the path specified.
Press any key to continue . . ."

vivid stream
#

I'm new to Stable Diffusion and I've only used MidJourney for now. I'm not quite sure where to start to be honest. I specifically want to do image to image generation with some creations in Minecraft. So take screenshots and transform those images into realistic photos or paintings/artwork, basically.

lavish lake
vivid stream
crimson pawn
#

Hi, good morning everyone, I'm new here and to Stable Diffusion, I'm having a bit of problem and your help would be greatly appreciated, when I click generate on deforum it comes back with "Error: None. Check your schedules/ init values please. Also make sure you don't have a backwards slash in any of your PATHS - use / instead of .

I've checked everything and csnt seem to find the problem, your response and help is highly appreciated.

lavish lake
crimson pawn
lavish lake
#

check that the init img path has forward slashes

crimson pawn
lavish lake
#

it can be multiple things, check that you have the correct syntax also in the schedules, values such as frame numbers, etc. i.e.: (0) : 5, (10) : -3

lavish lake
crimson pawn
karmic island
timber geode
#

any idea when we will get img2img in this server if at all?

lavish lake
wicked sonnet
#

I've been trying to perfect the generation but sometimes I feel like the model doesn't always follow what I tell it to, is there a way to remove the janky randomness?

nocturne storm
#

How do I get access to sdxl 1.0 Inpainting?

valid sparrow
#

guys pls i need help

#

whats the problem here:
"venv "C:\SD Files\stable-diffusion-webui-master\venv\Scripts\Python.exe"
The system cannot find the path specified.
Press any key to continue . . ."

stable pawn
#

This was a challenge given to me by a friend: "Can you create a portrait that look like or have the same pose as Mona Lisa without mention her name, the artist or any art style".

lavish lake
slender field
#

invokeAI is an free sd tool for the desktop whoi has great inpainting AND get controlNet in 3.0

stable pawn
slender field
#

its like paint

#

its bug free and easy

#

they have an very helpfull discord

bleak matrix
#

Good afternoon, everyone. 😶‍🌫️ How are we all doing?

bleak matrix
#

Type of paint, type of brush stroke, canvas, etc

lean field
#

Hello everyone, just joined the server
I don't know much about AI, I am learning it

Many of stabilityAI models are open source right?
How can use that locally?
Can anyone guide me a simple process to run that, i just want to try. Maybe running in python ig

bleak matrix
crimson pawn
lean field
bleak matrix
placid sage
#

I'm seeing people using {} in their prompts (on auto1111?)

Anyone know what their purpose is?

#

{shy face|curious face|happy face}, outdoors, (cartoon:1.2), looking at viewer, cowboy shot, (masterpiece), (best quality), ({black|white|ivory|dark|light black} theme),

#

example

broken smelt
#

That's with the dynamic prompts extension

#

It'll pick one of those options between the |'s randomly

#

So sometimes you'll get shy face outdoors, others you'll get curious face outdoors, and others you'll get happy face outdoors.

#

The | and {}s go together

astral goblet
#

RuntimeError: Couldn't install torch. << thats the meat of the error. if you look before the traceback code where it started, it was at Installing torch and torchvision Looking in indexes: https://pypi.org/simple, https://download.pytorch.org/whl/cu118 ERROR: Could not ind a version that satisfies the requirement torch==2.0.1 (from versions: none) ERROR: No matching distribution found for torch==2.0.1

so we can infer from that, you don't have hardware compatible with a cuda 11.8 enabled version of torch 2. AMD hardware?

lavish lake
#

you should try with Python 3.10.6, anything after that has documented issues and performs worse

frank spruce
#

"impressionist" means "doing things sloppily on purpose and pretending it's a good thing" change my mind

astral goblet
#

with most art history its a lot of hoity toityisms. but the impressionist movement was actually key to getting us to such an expressive world that we have today

vestal dew
astral goblet
#

art is subjective. thats a very cool thing about it. the context of a style matters a lot. The spiderman spiderverse movies purposely detune and use lower frame rates, objectively less quality, for artistic aesthetic. it's widely considered to be one of the best works though.

if you don't think impressionism is nice, thats' cool. There's a legacy to it though, so ....

vestal dew
#

Art can be ugly, but still have meaning

sudden pulsar
#

i installed the Lobe extension or "kitchen theme" and now I can't start stable diffusion anymore. Any breadcrumbs I could follow?

vestal dew
#

delete the extension

sudden pulsar
#

How do i do that? I installed it via the automatic1111 interface

vestal dew
#

it is likely in the extensions dir

#

your base automatic dir/extensions

#

it will have its own folder

sudden pulsar
# 
Startup time: 7.9s (import torch: 1.1s, import gradio: 0.5s, import ldm: 0.3s, other imports: 0.7s, load scripts: 0.7s, create ui: 0.3s, gradio launch: 4.2s).
ERROR:    Exception in ASGI application
Traceback (most recent call last):
  File "G:\SD\stable-diffusion-webui\venv\lib\site-packages\uvicorn\protocols\http\h11_impl.py", line 428, in run_asgi
    result = await app(  # type: ignore[func-returns-value]
  File "G:\SD\stable-diffusion-webui\venv\lib\site-packages\uvicorn\middleware\proxy_headers.py", line 78, in __call__
    return await self.app(scope, receive, send)
...
    language_data = self.load_language_file(lng)
  File "G:\SD\stable-diffusion-webui\extensions\sd-webui-lobe-theme\scripts\lib\locale.py", line 18, in load_language_file
    language_data = json.load(f)
  File "C:\Users\hs\AppData\Local\Programs\Python\Python310\lib\json\__init__.py", line 293, in load
    return loads(fp.read(),
  File "C:\Users\hs\AppData\Local\Programs\Python\Python310\lib\encodings\cp1252.py", line 23, in decode
    return codecs.charmap_decode(input,self.errors,decoding_table)[0]
UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 145: character maps to <undefined>
#

this seems to be some enocding error

#

butIm not sure how to fix it

tardy kestrel
#

Is there a channel on this server where people share json files and also how they setup their sdxl in comfyui?

I was hoping to get some help with a more advanced setup aside from just the base and refiner.

waxen geyser
#

does someone here use the auto 1111 extension for sdxl demo?

nocturne pewter
#

Hello, I haven't been here for a while. Can we now run stable diffusion locally on computer?

broken smelt
rigid cipher
#

sup all. I'm having a hard time with this. Basically im taking a human portrait and selected the background (does it perfectly) via inpainting, then write my prompts to change it to a sci-fi type background. But no matter the prompt or prompt strength (1), it's still influenced by the human in the foreground and keeps adding hair/arms/legs in the background. any way around this? Have tried so many models and so many different prompts/settings

young ivy
#

are you using an inpaint model?

#

they're separate

#

you can't properly do it with a regular one

young ivy
vestal dew
#

Make the background you want

#

not inpainted from that image...just make the background

#

Take your image, remove the background with AI or a program

#

In Photoshop or GImp or whatever, place the person onto the background

#

alternatively, get a program that turns SD images into real people

#

Send the person to whatever planet has the sci fi looking background you want

#

have the person take a selfie in front of that background

young ivy
#

if you have photoshop that has the best method of extracting characters. Their "AI" detection will detect a character and everything that's a part of them, clothes, items etc. Then you can work on the edges etc in the next box. After that, place it above some other background, do some autocolor etc on it.

Do a hue with 0% saturation and I forget which filter, make a curves layer and make the immediate layer under it a slave by using the attach button (just mouseover and you'll find it) then move the diagonal line so it's perfectly horizontal (128 | 128). You can then use I think the top right menu button (of the curves tool box) to use I think shift or ctrl on an image to change only that specific color within a tolerance. Without affecting the rest of the image

supple glen
vestal dew
#

I take it you've tried "flash digital camera"

proper tartan
#

Nevermind (I removed my question), I see in the settings the bottom is a button:
"Unload SD checkpoint to free VRAM", which is exactly what I was looking for.

astral goblet
#

3 - 3.5k for dual 3090's on ebay. i might be able to budget that and eat the resale lose if they don't work. Python can use them as one gpu if they have nvlink right?

#

maybe not

#

pytorch will be able to use both memory pools for a total of 48, but they won't be a single gpu. hmm. okay i'll study technicals a bit more before executing this idea

frank spruce
frank spruce
#

unless you just want to be able to make bigger batches of images at the same time, then it should be okay i guess

astral goblet
#

can get 3090s for cheap rn and they have nvlink capabilities

#

used

#

upgrading to a 4090 is the other option. i don't have a a100 capable system to drop it into

#

oh mb they do come in pcie form

#

still. $10k

vestal dew
#

4090 for 10k?

rigid cipher
# young ivy if you have photoshop that has the best method of extracting characters. Their "...

Thanks so much! sorry, missed your message since nobody replied for awhile. Was hoping to not have to go with a photoshop step in between my workflow...as I'm going to be dealing with 1000's of images this month 😦 is there ANYTHING that can be done via inpaint either via special prompt or negative prompts to make stable diffusion completely ignore anything in the foreground? It's a bit frustrating....

young ivy
orchid sedge
#

hello everyone

midnight verge
#

I want to make a person with AI so I need generations to be as consistent as possible based on a randomly generated person with a model from civit do I have to make a LoRA to do this and if so how

warm junco
midnight verge
quick island
#

Sorry, how do I use image2image? Is it possible to use this here in one of the bots? So to give the bot an image to work with? Thanks

lavish lake
deft marten
#

does naked dress count as nsfw? They are those dresses that many celebrities wear for a gala, red carpet

#

La ia no entiende el concepto de naked dress

lavish lake
bleak matrix
#

Good mornin, everyone

#

How are we all today?

solid cloak
#

anyone got experience with the reference preprocessor in controlnet? It's amazing for imitating a style, but gets the color scheme all wrong. Any way around this?

compact nimbus
#

@bleak matrix dont you hate it when people just turn to hate you bc of past mistakes?

white knoll
#

Отдам бесплатный интериум (кряк) создавал с большой командой там можно взломать кейсы и тд спросите какая нам выгода то этого мы развиваем популярность нашего кряка

compact nimbus
long garden
#

can someone help me with preprocessing?

crude pier
#

hello, recently i joined the stable diffusion.

#

what is importance of animation section of stable foundation? can i generate the animation through it?

lavish lake
calm elbow
#

hello can someone say me why with the 3060 ti i get very low it/s

lavish lake
calm elbow
#

i have 32 ram and the image its 512x768

lavish lake
#

GPU Ram... a 3060 does not have 32GB RAM, depending on the model can be 6 or 8 GB

#

what's your it/s? where are you infering?

#

a lot of things also affect it/s, like python version, xformers, torch version and so on...

dapper ferry
#

My GPU has 32 of RAM

turbid plinth
#

I have read somewhere on internet that SD XL 1.0 stable will go out the 18 july (tomorrow), it is true for the version automate 1111 ? How it will work ? It will need to install from 0 on a new folder ? Or it will be inside the first installation of automate 1111 and i will can switch between sd 1.5 and XL as i want with an auto update ?

soft moss
#

where can i do a bug report? or search if someone already reported and add my own details? i have found that if you interrupt any generation, or if you close the terminal abruptly, VRAM is not cleaned and the performance degrades until something that you can generate in 30secs on a fresh boot takes 5 minutes

dapper ferry
#

its so good having 32 of RAm

#

on this Nvidia card

hasty hornet
#

🫠

hasty hornet
obsidian adder
#

Does a platform for Img2Img exist where you can have a starting image and a target image and save the steps the generator took to get there and save it as a style?

crimson heart
#

who think sdxl is baised towerd latinas 😎

lyric bloom
#

hi, i'm new on sd
i'm getting double element of my desired subject does anyone know the fix?

broken smelt
#

What resolution is the image?

fallen narwhal
#

anyone know how to change body proportions of a person? inpaint or img2img? nothing I do seems to work

vestal dew
#

Yes, that is tough

#

often in outpainting, controlnet inpaint, it makes the legs too short. If I try to fix it, it adds something to block the legs or has the legs inside something

fallen narwhal
#

Which function of controlnet allows for that?

vestal dew
#

toi add to an image?

fallen narwhal
#

Yes, like to adjust someone's arm or leg size

vestal dew
#

Inpaint Only + Lama

#

no, it doesn't adjust their size

#

that zooms out of the image

#

I have been trying to get someone with a long neck. Can't get it. Like you say, inpainting doesn't really work

#

you could try making an openpose , then use thjat when you first generate

fallen narwhal
#

I experimented with openpose but it doesn't really work for existing images, no? Isn't it primarily for text2img? I'm really new, so thanks for the info

vestal dew
#

edit an image to make the arms or legs longer, then run it through again with canny or depth

vestal dew
#

use an original as reference only, and use an edited version with the longer limbs you want for canny or depth

fallen narwhal
#

Alright, I'll try

vestal dew
#

you can even draw a canny or depth map, use it as a template

fallen narwhal
#

I don't know anything about controlnet yet

vestal dew
#

look at some videos

fallen narwhal
#

Will do

#

Very many features

final dragon
#

Anybody up to voice chat tonight about diffusion?

fervent thunder
#

@forest trout
which of these is better
winner-gallery
or pantheon

and what does highest voted from 'sdxl bot' mean
what kinda bot is that, do votes from bot mean anything @broken smelt

forest trout
#

Pantheon winners are just submissions that a lot of people voted on. The Pantheon winners are just people who got a lot of votes on their images during a specific period of time (winners are announced every hour).

ocean mango
dawn lance
#

2nd week using Google Colab Notebook: Automatic1111 SD

Hello, I'm trying to make a custom LORA using Kohya Dreambooth, but it appears like I'm missing something; the image's details are excellent, but the face doesn't match the Data_sets I used. Is there a setting I need to change to make this right?

PS: I'm using Toolpie to determine whether the generated images are similar to the reference image I use through Control Net, but even though the results are 91% or higher, sometimes it still appears to be a different person, (Not Reliable). I must say that Human Checking is still preferable to using Toolpie, (I wish I had good eyes).

Potential causes
Data_sets:
Configuration?

Questions:

How many images should i use as a Data_sets? ( Should i used 512x512 to each images? )
How many sampling steps are advised?
In how many epochs?
Is there a Resolution settings? For instance, should I reduce my data sets from 720x920 to the normal 512x512?
After generating the xxx.safetensors, do i need to use the default image enhancing prompt settings again? ex. 8k,Raw photo, HD , HDR , etc etc

tribal fjord
#

Is it just me or is the quality of generations significantly worse than yesterday?

remote urchin
#

Hey do I need to keep the command '--lyco-patch-lora --lyco-dir "C:\A1111\stable-diffusion-webui\models\Lora" --api' in my user.bat file after I've reinstalled the lyco or whatever this was supposed to do? I dont remember

#

updated my loras or lycorishes to be in the same folder I think

#

Do I need to keep the command line?

#

or rather, lines

elder sleet
#

can anyone advise on Loras, how many images I should have and repeats? Im in VC as well if anyone wants to chat about it (diffuser 5)

cosmic light
elder sleet
#

i have 200 images, all high quality but I dont understand it

elder sleet
cosmic light
#

I'm not really sure, tbh (⁠^⁠～⁠^⁠;⁠)⁠ゞ

#

I just used the default settings on the trainer in auto1111

#

But that was for a hyper network, not a lora

#

I'm not sure how different they are

sacred jay
#

Does anyone know of a discord channel for meta AI models? I’m looking for help using their open source musicgen model.

magic void
#

hi is anyone here? 😄

#

i dont understand the difference between the "normal" stable diffusion and deep Floyd

#

or deep floyd "IF"

cosmic light
vestal dew
#

DF takes a lot of VRAM

cosmic light
#

Yeah that too

#

It's a beast

#

Though the 4bit version takes significantly less

#

I hope to try it out soon

warm junco
magic void
#

yes thats what i saw, that DF is better with Texts in images

cosmic light
#

Yeah that's a big part

magic void
#

but in general both are just different T2IMG Ai models?

cosmic light
#

Yes

magic void
#

ok thanks :))

#

im a total noob

cosmic light
#

IF can also take existing small images and use it's upscalers on them

#

I haven't personally used it tho

magic void
#

nice so i can give IF an bad quality image and it restores/upscales quality etc?

#

when will it be released? 😄

cosmic light
#

I think it's already out? People are using it

#

I was looking to download it earlier but I couldn't find it tho

#

It's not just a checkpoint that can be put into auto1111 tho

crimson abyss
#

Does anyone know what's the difference is between the "original" and "diffusers" backends (in Vlad's for of AUTOMATIC1111)?

astral goblet
#

it just seems like a different technical implementation that really provides no differences to the end user

crimson abyss
astral goblet
#

people got really angry at me for asking around about this this week. tread carefully with this question. there's a lot of heated emotions around the topic and i really have no clear understanding why

finite cloak
serene swan
#

K need some feedback. I have a long term project for creating a very long kinetic novel using ai. The plan is to use a realistic model to generate images of my custom characters, train loras for those characters, then once I have them, use the loras in "anime-like" models to get the style I want, and the usual photoshop processes. Sound about right or too convoluted?

delicate basin
# crimson abyss Does anyone know what's the difference is between the "original" and "diffusers"...

I think there are some benchmark done available in their discord but basically the original is what automatic has been using and was coded by some code from SAI way back. Nowadays diffusers take the job at implementing the means to use the models, loras and so on so using diffusers means leveraging the oss community more and have some things developed quicker without diverting resources. There shouldn’t be any differences apart that atm in the current fork you cannot use SD XL or kandisky without diffusers

crimson abyss
brave jacinth
#

do u need seo , web designing services ?

light solar
#

has anyone been using the stable diffusion add on for blender?

#

specifically the depth and control net features, ive just tried to update mine and it still seems to be the old version though im not sure it still points me to the updates page and i cant get my normal control net model to run on it

turbid plinth
#

I have read somewhere on internet that SD XL 1.0 stable will go out the 18 july (tomorrow), it is true for the version automate 1111 ? How it will work ? It will need to install from 0 on a new folder ? Or it will be inside the first installation of automate 1111 and i will can switch between sd 1.5 and XL as i want with an auto update ?

astral goblet
#

search engines actively de rank you if you do optimization shenanigans to that regard. don't believe these scam artists selling seo services in 2023

vestal dew
#

Stable Foundation people say they want people to vote on the model version that will continue to release. How can we vote if we can't try the models?

lavish lake
fervent thunder
#

Hello AI people

hasty hornet
#

search enginges de rank only if you spam shit too much.
Keywords are still a thing as far as I know.
(texts and meta tags)

astral goblet
#

that's UX design. that's a legitimate service but there's a ton of snake oil in that market too. SEO is a horoscope level enterprise

hasty hornet
#

nah, UX design has nothing to do with SEO.
People looking for seo when they need to get higher positions and usually nothing else really.

UX is opposite - just pure user expirience, don't care about search engines.
While search engines trying to boost "good UX" practices - that's still not the same thing.

astral goblet
#

google's algorithm is highly proprietary and intentionally kept secret. anyone claiming they know how to optimize to it are selling you a bridge

#

there was an era of Google bombing, when the corners of the internet like 4chan learned how it worked for a bit

hasty hornet
#

They have whole youtube channel for it

#

I'm not seo guy, I'm mostly doing visual things, but my company do that, was learning few things just out of curiousity

astral goblet
#

it's more how to optimize for ad exposure and how to keep content coming out everyday. always updateing. training people to do a feed of information that ad's can go into .
sure fire way to get higher rank is more traffic. the algorithm is entirely secret except that we can count on the high traffic high ad revenue and exposure links getting ranked higher.

hasty hornet
#

sure fire way to get higher rank is more traffic - that's just one part...there are alot more variables

astral goblet
#

ad exposure is a big part of how google ranks. Higher traffic pages that are utilizing google analytic systems rank higher. it all has a long tail effect of making more ads expose to more traffic. Google is the world's biggest advertising company. this is their meat and potatoes

#

they keep the actual machinations of how it works under wraps so that people can't fuck with their advertising revenue and they maintain control

#

thats why we don't see google bombing anymore

hasty hornet
astral goblet
#

marketing works. you can decide not to go to a place because you saw an ad, but that ad will work still

hasty hornet
#

You just have no idea what you're talking about....sorry, I'm not gonna continue...

astral goblet
#

Oooo i'm gonna get out of here and not touch this room with a 10ft pole now

#

gotta be using a weird hacked out discord script to react to my message while blocked 👀

willow coyote
#

yep speed is crucial

astral goblet
vestal dew
vestal dew
#

NMV, I think I found em

bold solar
#

Are we able to watch the court trial between Stability, Midjourney and DeviantArt vs Artists tomorrow?

#

I’m very surprised more people aren’t talking about it

#

It could have very big implications for the future

vestal dew
#

the elites will take the tech away from the hoi polloi

bold solar
#

They can’t take away what’s already on many people’s computers

#

The cat’s been out of the proverbial bag for quite some time now and it’s not going anywhere

#

I just want to see the outcome of tomorrow’s court trial

wild bluff
#

👋 Do you have insights about the backend tech stack of stability ai?