#💬｜general-chat

1 messages · Page 78 of 1

proud shell
#

I should get another offer tomorrow

warm junco
#

check if you really want that SSD because on amazon SSDs are cheap rn

#

maybe cheaper than in your offer idk

proud shell
warm junco
proud shell
floral umbra
#

Do you guys know of a interpolation program/git that thoroughly analyzes each frame to not give artifacts?

jovial quail
#

Hi

warm junco
floral umbra
warm junco
floral umbra
primal quartz
#

can you provide me the link toinstalll A1111new updatess plz @warm junco

jade path
#

Please help me with the site link can’t find it

primal quartz
#

what link ?

calm relic
#

Hey all - Is there an infographic for how SD can be used with architectural design concepts?

jade path
primal quartz
jade path
primal quartz
jade path
#

Ok

covert sand
#

is there a control net model for color?

soft moss
#

what are you expecting this model to do?

covert sand
#

color things the color i want them to be

#

by inputting like a roughly colored sketch

#

like what the lineart or scribble does but with color

soft moss
#

image2image does that. like the basic tool

covert sand
#

true lol. hadn't really thought of that😆

#

i can use control net w img2img?

#

pretty sure i did earlier

soft moss
#

put a doodle with colors in the image2image canvas and enable controlnet

covert sand
#

thank you very much

#

idk how long it wouldve taken me to think of it lol

soft moss
#

sometimes you dont need a swiss army knife to just make a cut, you can just use the rusty old knife 😛

feral eagle
#

why is inpaint anything SO SLOW, does it not use GPU? really bad when i run the inpaint part

plucky shoal
#

@calm relic i think one of controlnet models is there which work well with architecture, on other hand it can be all well used, depends on what do you expected.

calm relic
plucky shoal
#

you are doing text to image?

calm relic
scenic bobcat
#

yo

#

can someone help me

plucky shoal
#

and can you post a prompt? Or better what you want interior, exterior? I think you have to play as well with cfg bit

scenic bobcat
#

I was trying to install stable diffusion on my pc but

#

it gives an error text

calm relic
plucky shoal
#

i made something topdown view but i had not architecture on mind. I think exterior images need higher cfg otherwise not very nice, some sort of glitches like things. <- dont know how to call it

calm relic
plucky shoal
#

you can help the picture with guided images. For interior there is very good controlnet called MLSD can make variation of rooms or probably exterior as well.
Sorry.

deft socket
#

can anyone help me here? I have problem trying to load a .pt file and it says it couldn't identify the file as neither textual inversion embedding or diffuser concept

fathom maple
#

Is there any place I can go to get support for the client of SD that I'm running? I get so many different errors that if I were to just use a forum or subreddit I'd be opening threads left and right

tender cove
sonic pumice
#

Sorry if this is a stupid question but... What is a Defi?

#

And oh

#

Okay then lol

severe zealot
#

what are your thoughts on this as a good name for the genre; What is Haidrid?
Haibrid is an acronym for Human Artificial Intelligence Blended Rendition of Information and Design.
It is a new genre of creation where 50% ↑ Human and 50% ↓ Artificial Intelligence (AI) Are merged to create something new be it Art, Design, Music, Photography, Video and Graphics etc. https://www.reddit.com/r/Haibrid/

sonic pumice
severe zealot
ashen kelp
#

There are so many upscalers. Which one have you had the best results with when it comes to photorealistic images of people?

#

Is it still recommended to go with 4x-UltraSharp or is there something better nowadays?

fervent thunder
#

what do you mean?

proud shell
#

Intel i7-13700F TRAY
Zalman Ultra Quiet CNPS9X Optima RGB
ASUS PRIME B760M-A D4
XPG D35G RGB 16GB 3600 CL18
INNO3D GeForce RTX4080 16GB X3
Crucial P3 1TB
Antec Neo Eco 750W Gold Silent Fan
ANTEC AX81 ELITE

simple echo
#

is there anyway to get anime style images in the web version? i cant seem to get the app or whatever one to work so the web is currently my only option

torpid ether
#

hello everyone，I'm working full-time on the open-source project SolidUI, which can generate any graph with just one sentence. I welcome friends to participate and contribute together.

#

You can also click on my avatar; there's a join link inside.

coarse veldt
#

hey, is there a way using automatic's web ui to generate multiple prompts at once using a single gpu?

torn mist
#

How can I see what version of SD I have? How do I know if it's current? If it isn't, how do I update it?

soft moss
low turtle
feral eagle
#

Ok so I rock a 6700 XT amd card and after a lot of pain last week I got it working. Then for giggles I tried it in aws with a Tesla t4 and it cleaned its clock on it/s . So stay on a aws with only 4 cpu cores or replace my 6700 xt with a 3060 12gb in my fast gaming rig and upgrade to 32GB ram haha

#

I get in then 6-7 it sec range on the T4 and 3 it/sec at most on the 6700xt

dense sandal
#

Hi guys, am I allowed to share a fantasy character generator app that I'm working on that uses stable diffusion and stable horde?

sick dew
#

Hi, where can I ask controlnet questions

upbeat tide
#

hi all - is anyone here good with making game level (mapping) sci-fi 'flat' 2d semi-realistic textures in Automatic1111 (web user GUI)?

warm junco
#

But on windows the directml isn't good in performance

grand scaffold
#

hi

torn mist
soft moss
#

at the bottom of the page, it tells you the version of A1111 you are using

feral eagle
trim magnet
#

like 3 times faster

pine quiver
#

whats this server about again? forgot why i joined

simple herald
#

Hi everyone:), I hope this is the correct channel to ask for help.

does anybody have any experience with using temporal kit + ebsynth in automatic1111? It works fine for me for short videos, but if I use the split video function, temporal kit fails to create the frames

feral eagle
#

is the 4060 TI 16GB really getting 12 it/sec ?? looking for alternative to 6700 XT for inference and SD. was going to go 3060 12Gb

trim magnet
#

yea around 12/13 it/s

#

cheapest and good would be the 3060 12gb

feral eagle
#

yea, be for gaming as well.

trim magnet
#

if u have the money you could also go for an used 3090 24gb its like $700

#

or an used RTX A5000 24GB that one is like $500 and it performs like 10% less than a 3090

feral eagle
#

Looks like 6700 XT get about 8 itsec with Linux

raw dagger
vivid heart
#

can someone help

#

im using the webui version of sdxl

#

but the images are glitched

wanton sedge
#

There is clip interrogator/describe bot on this discord server? It would be great

ruby lake
#

Hey guys, I am working with my wife on integrating AI into her photography business by providing one of the clients photos as an oil painting rendition, or other art genres like papercraft. Does anyone know of what I would use to be able to do that?

#

Can I do it with StableDiffusion?

feral eagle
#

ok, installed ubuntu, installed SD using instructions for AMD from github, but its not using my GPU

warm junco
burnt lynx
#

I am trying to install stable diffusion on my macbook. When I try to locate the stable diffusion model, I get messages that tht file cannot be found. I've tried moving it, restarting, and I've tried opening a new terminal. It's still not working. Can anyone help? pretty please 🙂

thorn citrus
#

Hey there
Hope you are doing well
I am a senior full stack developer who has full experience in web and AI development
So if you have some projects, please let me know
thank you

velvet ocean
#

hello

still glacier
crisp dagger
#

hi dudes

#

@proud shell 4080 most expensive shit..even for iphone 15

proud shell
crisp dagger
proud shell
marble portal
#

is it just me or do recent SDXL results look slightly worse than from a week ago?

tidal epoch
#

In Override settings how would i put face restoration true? How do i add to override settings

long cradle
#

Hello

#

That other day I saw this dude running midjourney on his bot, how tf he do that

long cradle
torn mist
soft moss
#

That sound about right

#

Editing the bat means you will update every time you run the software (I don't recommend it)

#

You can open a cmd window in the folder and just type git pull and that will download the latest version if you want or need to update

#

I recomee you search for a tutorial if you never worked with the command line

floral umbra
#

For any "news" regarding A.I generative stuff, which dependency/requirement is used the most with most/all of them? As i wanna find more cool A.I stuffs, but don't know the best source for finding new stuff for image/video/animation/audio/mocap? and so on

torn mist
smoky flicker
#

hello wanted to install stable diffusion webui on my pc

#

and its taking alot of time here

#

is it normal

#

collecting torch

stone river
#

yes it takes a while

#

is there a way to speed up generating? and also, how do i add tags to my images? and whats a good model for anime characters and stufff like that

soft moss
torn mist
soft moss
# stone river is there a way to speed up generating? and also, how do i add tags to my images?...

one by one.

  1. speeding up, controversial topic. if you have low VRAM (8gb or less) google that and it will help, otherwise not much i think
  2. you cant "tag" images from baseic A1111, altough the PNG files have their generation text stored in them as plain text (open them in notepad and see. you can also check the infinite browsing extesion
  3. very subjective. loog for your style and try. donbt be discouraged by bad results, its going to take weeks to fully understand how each paramter affects the results. a pretty good starting tip, read the models recomendations on CFG, steps, etc etc
stone river
#

oh okay

#

im using img2img btw

soft moss
stone river
#

also are there optimal settings for img2img?

torn mist
soft moss
ivory fractal
#

I Found a loophole for unlimited nitro. This link can be claimed by unlimited people.

stone river
#

no?

vivid heart
#

i think he is high

stable quarry
#

Hello I'm pretty new with this whole thing. I tried Pixai and makeayo but I'm looking into trying this. Where do I even start?

vivid heart
#

without the space

#

i cant send the original command

stable quarry
#

Thanks

vivid heart
#

yup

nova dagger
#

hey

stable quarry
#

Is there a stand alone program or app? Or is it all just generated in discord channels?

soft moss
stable quarry
#

10-4. I found dreamstudio so far. I'll look around some more

knotty turtle
#

What's this I hear about a public beta for ControlNet SDXL. Anybody have a link?

smoky flicker
#

hii i get out of memory cuda problem i have gtx 1650 max-Q

#

how can i fix it ?

ashen marten
#

that sound familiar?

plucky shoal
#

@smoky flicker what model are you using and what resolution?

smoky flicker
fervent thunder
#

Does anyone know how to run SD on two GPUs? I'm considering getting a second 3090

#

i've seen people run local LLMs off multiple GPUs

plucky shoal
#

@fervent thunder should work but not in series, should work in parallel. Try ask in #🐝｜swarm-ui

#

@smoky flicker are you using A1111?

fervent thunder
smoky flicker
plucky shoal
#

i mean it should work like having running two apps. @fervent thunder
but not sure ask there

#

and what resolution you have there? And model its name is in left top corner

smoky flicker
#

Wait let me check

#

Resolution using 1080x1080

plucky shoal
#

it is too high. If you are using normal model it is supposed max 768x768 or 512x512

somber nacelle
#

EU always fail on technologies concern

crisp dagger
somber nacelle
crisp dagger
somber nacelle
#

They are scared of AI, that's why they try to censor it

somber nacelle
small shard
#

im trying to understand what the proposition says but i dont fully understand it what will it limit/ how will it work?

somber nacelle
#

The country that said "Oh you know what ? we can't control internet with DNS censorship anymore, let's force the browser developper to censor internet for us"

normal pagoda
#

I wonder how this AI Act that's coming from the EU is going to affect things like Stable Diffusion (which are just used for generating art)... nyaThink

crisp dagger
somber nacelle
crisp dagger
somber nacelle
crisp dagger
#

market will penetrate his own way

somber nacelle
crisp dagger
signal trellis
#

so what happens if this EU thing goes through? just makes it illegal to have fun?

normal pagoda
somber nacelle
somber nacelle
crisp dagger
crisp dagger
#

but i understand the frustration behind it..its a mess

normal pagoda
# somber nacelle StabilityAI

hmm... I don't see a reason why they should get any problems with EU laws in this case - unless... the model diversity will shrink heavily
and that's a big problem, then

solar condor
#

hey guys can i train a model with a mac ??

#

im an artist and regrouped 55 of my drawings to train it for my style !

signal trellis
#

won't it necesaily effect stuff outside EU? like with twitter they have to police speech for accounts outside of EU because if they want to opperate in EU and they don't police speech according to EU laws then they get fined or something. something like that going to happen with the "ai"?

normal pagoda
somber nacelle
#

In France they're developping AI surveillance camera for Olympic games

normal pagoda
pale latch
#

Twitter under elon can afford to pull out of countries

#

Especially when he knows people from those places will use it anyways

small sand
#

thank god, violation of EU laws can be punished only on EU territory 😌

small sand
rose wave
#

bruh i havent used SD in like 2 months are there any significant news or advancements?

hot marlin
#

It doesn't matter what country you live in for these new AI regulations. The EU already fines companies based on their world wide annual revenue. Not profits but revenue. This of course forces companies to enforce EU laws over the whole world. They already do this with the European Union’s General Data Protection Regulation (GDPR). https://gdpr.eu/fines/

keen badge
#

i was reading about this "eu ai act", in the ping, and it concerned me a bit even though im not even from the eu.. can someone let me know that they intent to let devs train ai on any data publicly available, then distribute models, with no censorship, publicly? my concern largely stems from open ai PUBLICALLY talking about trying to make their ai not tell people about, for example, how to make a bomb or something similar (may have been nitroglycerine?), whereas i see it as vitally important to the future of humanity that we have open access to the most powerful models in an un-lobotomized form, where they can give their full most accurate valid answer, with FREE access to KNOWLEDGE for EVERYONE. can someone affirm for me that their intention is to support the creation and distribution of such models? i unironically see "access to unrestricted ai" as one of the biggest deciding factors in humanitys timeline, with no access and no unrestricted knowledge of course being the (insert political terms leading to bad results) future

lethal plover
#

hey guys, how do i use the bot here?

solar condor
#

damn

pale latch
#

recognizing that models CAN be transformative works doesn't open them up to no holds barred wild west legal freedom

#

images, text, media, whatever

#

i hate the gdpr. i'm canadian so it does nadda to protect my privacy. but it sure makes it so eveyr single website now has a stupid stupid stupid "do you accept the cookie token?!" prompts that are effectively useless to protect privacy

#

privacy theater

#

everybody must get stoned

#

that's your themesong rn

#

maybe before your time

#

i live on an island . the train doesn't run on it no more

#

the tracks are here but no trains

#

it'd just be me jumping

#

it'll never come. bro. they shut it down

#

ghost train maybe but like t'll just go thru me cause i'm hot

stable pawn
#

I did read the #📣｜announcements that AI models soon may be treated "identically once they are released publicly" and that this can hit towards grassroot developers, but do anyone know what they see as "released publicly", can a simple sig-in make it be seen as "non public"?

pale latch
#

thats the obvious one to tell me to do in the first place, but i'm the bozo

#

🤡 just 2 clowns looking at each other

#

get it?

#

it's looking at you an you're looking at it

#

i'm just saiyan

#

kamehame hahahahahahahahaha

#

me so funny

pale latch
#

i know theres a lot of nuances. like you can be a private business but have your doors open to the public

vivid heart
#

literally you??

pale latch
#

i dont think you understand how air works

vivid heart
#

i think he drank too much beer last night

#

you

#

you do it

#

if you want me to do it so bad

pale latch
stable pawn
#

Okay, I an no huge fan of intellectual debates, so I am out of here.

vivid heart
#

you

pale latch
#

my guess is some girl he liked started dating the hot fresh ai artist over the summer and he's back at school now and is sad about that big reveal

vivid heart
#

literally you

pale latch
#

it's always about a girl

#

see

#

nailed it

vivid heart
#

literally you

pale latch
#

your crush is dating an ai artist

#

made you sooooooo angrery

vivid heart
#

no his e boyfriend dumped her

#

you first

pale latch
#

AI artists are taking all the crushes. we're the new hottness

#

used to be "learn to play guitar" to get all the crushes

#

now its "learn to ai art"

#

girls just go mad

vivid heart
#

literally you

pale latch
#

it's stable mania baby

#

girls dont' like guys with cars anymore. it's all about the 4090

vivid heart
#

you

pale latch
#

you roll up to the club with a h100 an i tell you, they'll swarm

#

"awwwoohmmaaahggoawawddd its an ai artist and he's got his gpu with him!"

trim magnet
pale latch
#

lol i wasn't really trying to debate. not everything is an argument. most people are not going to be experts on this subject

#

i was just saying theres some nuances around what makes something public

mild condor
#

I have a doubt, from time to time, the Lora file gets corrupted, sometimes the size goes to 0 MB. Does anyone know how this happens? For now, I keep a few copies outside of Stable Diffusion

pale latch
#

nothing should be deleting your lora like that. that's such weird behavior that it feels intelligently directed. like, i bet one of your friends has a RAT on your computer and is messing with you, deleting that shit while you're using it

#

another instance where a file would just up an delete is a virus scanner detecting it

#

unliklely for a lora, so we're left with the 2nd useron the system theory

forest trout
#

Idk what you are talking about. Every time I use my loras they get consumed in a ritual sacrifice which pleases the latent space spirits to generate an image that looks like the Lora that was sacrificed. It's a beautiful process if you think about it 🤔

pale latch
#

hail science!

onyx hedge
#

hi

#

impossible to take acquisitino of the 1.5.4 version of stable diffusion : /

#

1.6 is sexy but useless...

#

do you know on git to be able to return on a 1.5.4 verson or close to 1.5.5 ?

#

A eternel thank you for anwser =)

pale latch
#

if you wanna git fancy you can do a git revert c9c8485

wanton nacelle
#

So guys I have a question. If there’s (especially now over the past few months) a debate about whether LLMs have any rudimentary form of consciousness, does that open the debate about if other systems do? For example, we tend to focus mainly on LLMs because they talk like humans, and seem more humanlike - but would similar processes not be used for, for example, AI Image models? Or even what about things like video games? You have a bunch of lines of code that trigger different events when different criteria are met, and they all work together to create a bigger picture - is that not at least somewhat similar to synapses in the brain firing off to create coherent thoughts and perception?

#

It’s honestly a little concerning - this may sound like sappy emotional nonsense but it the prospect that we could have been abusing systems that have some form of actual awareness for years genuinely does make me a little unnerved. You guys know way more about AI and technology than me - what do you think?

feral eagle
#

2080 ti for 240.00 good option for sd and llm?

fervent thunder
#

somebody here?

#

can somebody help me with Stable Audio.. it seems like i cant generate my music..

pale latch
#

i dont think there's a debate. no indications of it has been done. Just people prompting for it and the language predictor effectively working

#

there have been people raising alarms that chatgpt has consciousness, but it does not. It's the very accepted consensus among researchers. There will always be an oddball who swears these predictors are doing more than they're trained to do though

fallen lodge
knotty axle
#

Just found out someone selling stablelm.com, I thought stability might be owning it but turns out not

vast berry
#

found a new way to vote out hand abominations: nail art prompts

raw dagger
lyric narwhal
#

anybody have any tips for how to write good prompts?

fallow veldt
#

I was wondering the same

broken gull
fallow veldt
# lyric narwhal anybody have any tips for how to write good prompts?

separate in parts

medium (photo, analog photo, street photo, illustration, etc)
subject, with important defining adjectives (age, gender, role)
action (what is the subject doing)
style (in the style of certain artist, cartoon style)
secondary attributes
location
style detailed atributes (HDR, subsurface scattering, detailed skin)

fallow veldt
lyric narwhal
#

Thanks a lot!

fallow veldt
#

also in medium you can state the angle and proximity (high angle, low angle, close-up, wide shot, etc), so "A low angle photo" or "Photo, low angle"

lyric narwhal
#

Thanks, that's really helpful. I have trouble explaining what I want the AI to generate sometimes so this is a helpful blueprint.

fallow veldt
#

you are welcome! I also think sometimes writing a more grammatical accurate prompt may help the IA better than doing the booru tag style (like "photo, low angle, a man, running, park"), so maybe better try to do "a photo of a man running in the park" as long as you can

#

but usually it ends mixed and it doesn't matter much ("photo, a man running in the park"), but it seems to catch things better like that

#

also style and medium could be mixed

south flame
#

@sudden ruin sorry to bother an admin, I was wondering if it was possible the Metadata bot could be added to the server? It allows people who decide to post on original render to share their generation data by someone putting a magnifying glass on an image, very useful for many SD websites.

ivory fractal
#

guys check out my website which i have embeded all the art made using stable diffusion

#

https￶://rohithm.netlify.app ||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||||​||

sudden ruin
safe sleet
mellow hearth
#

hello

feral eagle
#

2080 ti good for sd and llm?

warm junco
lime crest
#

Hello dear AI enthusiasts, is there a way to determine the max resolution you can use? For some reason, 1024x896 works but 960x960 does not, is there any way to determine without going through every single possible resolution?

warm junco
lime crest
torn mist
#

Someone help, please? I can't get "Interrogate CLIP" to work. It did work a while ago, but now I get a "connection timed out" error message. What am I doing wrong? (I can post a screen shot in general-with-images if it will help?)

lime crest
#

Connection timed out sounds like your webui.bat is closed

torn mist
#

webui window shows a list of six "1006" lines

#

I'm very new at this. I ran "webui.bat" - was that the right one?

lime crest
#

The correct one to run is webui-user.bat

torn mist
#

Thanks - that helps! No error message now - but it's taking an awfully long time, past 200s now... is that usual?

#

Spinner is still spinning (420s) - the image isn't that complex... what else might I be doing wrong?

lime crest
#

Hmmm, maybe it is taking so long the first time? You can let it run now and try to refrain from making images or changing settings

#

When you interrogate, do just that

#

Could be from your GPU, do you have enough VRAM?

#

You can try settings like -xformers and -medvram to speed up your rendering at the price of lower quality for xformers

#

Sometimes changing models or interrogating takes a really long time,

torn mist
#

Thanks - I've just restarted the PC (Win11), snd going to run webui-user.bat and try interrogate CLIP again. My GPU is a 2060 with 12Gb.

lime crest
#

You should have no problems rendering great images 🙂 GL and HF !

torn mist
#

Thanks for your help. It's running, but not giving any output yet. How long should I wait before I give up on it?

#

The current model is DreamShaper, BTW.

#

Stil no output after 10 minutes - I've used Interrogate CLIP before, and I know it didn't take this long.

#

Just a thought - I'm using an ordinary image as the source to be interrogated (a PNG I produced in MidJourney) - that should work though, shouldn't it?

lime crest
#

I am really not sure what could be the cause of it

#

So it worked before but not now?

#

Maybe something autoupdated and broke it

torn mist
#

Yes, it worked before.

#

What does "1006" in the cmd window signify?

#

I.e., does it mean "working properly" or "there's a problem"?

#

How do I stop the process?

lime crest
#

try adding --no-gradio-queue

torn mist
#

Er... where do I add that? The prompt window is not accepting input while the spinner is there.

lime crest
#

right click webui-user.bat file

#

click on Edit

#

and add it to set COMMANDLINE_ARGS=

#

I hope it is like this, if not, try it on the other 2

torn mist
#

I'm running it on the huggingface link you gave - been going for 2 minytes so far...

#

I posted a s/shot in general-with-images.

lime crest
#

Im sorry it didnt work, hope u find the fix

#

Reinstalling SD wont harm if all else fails

torn mist
#

Are you able to do it on your setup? Just wondering because that image ain't working on the huggingface link either. So I downloaded a pic of a kitten from the web, and ran it on the HF link as well, and that's also not working So it's not my hardwar, or my software, and probably not the image I used. What's left?

warm junco
runic merlin
#

Hello people!
I'm wondering, I had been using stable diffusion online and I love some styles that come as an option, since few days I started with Automatic 111 and I can't archive the quality of that images in the inline version....is anyway of using this styles in the WebUI version?

#

Because writing the name of the style is not enough (example: retro-futuristic)

#

Like the quality is not at all the same

#

I'm also paying for the premium subscription of this online version ((Black Magic))

feral eagle
forest trout
#

They even do it with the discord bots here in this server when you apply a style to your prompt.

pale latch
#

basically, they're just words that append to your prompt

runic merlin
#

Ok ok

#

Thank you 🙂

pale latch
gray wolf
#

With Dall E 3 coming out soon when do y’all think local hosted free ai like stable diff can level with it?

forest trout
#

If DALLE3 let's me pop open the hood and tweak everything then it can level with Stable Diffusion.

#

Until then, SD > DALLE3.

fervent thunder
#

Hello brothers and sisters (and aunts and uncles and parents grandparents and cousins and second and third cousins and ...)!

I am new to Stable Diffusion. Most people seem to be using the stable-diffusion-textui wrapper from here: https://github.com/AUTOMATIC1111/stable-diffusion-webui. I tried that out, and I don't like it. It leaves my gpu running at 100% of its power even when idling, just for having that web ui open, which is insane.

So I decided I want to try and use the barebones stable diffusion, am now in the quite arduous process of installing it from here: https://github.com/Stability-AI/stablediffusion, but found 10 months old reddit post (https://www.reddit.com/r/StableDiffusion/comments/z42wfc/comment/ixp1eu0/?utm_source=reddit&utm_medium=web2x&context=3) from which I learned that the stable-diffusion-webui actually uses an older version of StableDiffusion and it claims that StableDiffusion 2 is "incredibly unoptimized".

Is this just outdated info? Is StableDiffusion 2 stable and performant now, 10 months later?
And where do I find good resources on how to work with it?

gray wolf
pale latch
fervent thunder
#

Cops took my pc away, it had a 4070 ti. I now ordered a new one, with a 3090. Will it peform much better than the 4070 ti?

trim magnet
#

yes\

warm junco
fervent thunder
#

I just can make bigger pics with the 3090?

warm junco
#

Also you can train models easier

trim magnet
fervent thunder
trim magnet
#

yes

runic merlin
#

someone has experience with infinit zoom?

#

Everything was working but I was having frames very visible ans when I checked in settings I saw it was not selected

sonic burrow
#

I just installed SDXL and restarted my pc how do I launch the app?

runic merlin
#

And now when I selected it is not working

potent spire
#

for the prompting part and image quality it seems that DALL-E 3 does or will beat it

potent spire
pale latch
#

thing with dalle3, it runs in a datacenter and gates access

#

so its irrelevant to people building on stable diffusion open models

#

also if it thinks you're asking for anything that resembles a copyrighted art style or someone who is living, it'll decline it. they're designing it as a "brainstorming tool"

#

it doesn't even have a research paper released for it. it's entirely hype to drive more investers to "open" ai

pale latch
#

dalle will have a business model that lasts fora while . its irrellevant to me

potent spire
#

Yes, but Google and Adobe are the main ones in that

#

Well Google with Gemini

rotund notch
#

hi, is this a place wer can chat about ComfyUI?

hybrid edge
rotund notch
#

new to what?

hybrid edge
#

the server

rotund notch
#

oh. okay, you use sd or Comfy?

hybrid edge
#

idk

trim magnet
hybrid edge
#

ok

#

i find that desturbing

rotund notch
#

certainly ai generated

hybrid edge
#

yep

#

i just added you as a friend

high ruin
#

Anybody know of a free program that can batch adjust multiple image colors and brightness??

pale latch
#

open source baby, well except irfanview, non commercial freeware

weary raven
#

Can I tell stable diffusion to use a provided face when generating people? Im using 1111 ui

foggy shell
#

How can I save my prompt more easily again?

hasty hornet
weary raven
#

oh is that what lora is

#

or does lora mean mod or setting or something

steep knoll
#

Who here can create full illustrations for kids story books?

  • key being that they can successfully recreate the same character(s) every single time in different stories.
lusty beacon
#

how to I generate images locally. I only get weird images

#

i'm using invoke ai and it generates better images

weary raven
hasty hornet
frigid wyvern
#

Hello
I need help.
I'm on a specific task: I want to batch process images with controlnet depth and a mask, so for an image sequence
I can't get anything...
I'm on Windows 10, webui updated today (version: v1.6.0  •  python: 3.10.9  •  torch: 2.0.0+cu118  •  xformers: 0.0.20  •  gradio: 3.41.2  •  checkpoint: 6b08e2c182)

#

any help welcome

#

Does anyone know how to get a different seed per image in a batch process?

covert sand
#

what does recolor controlnet model do

indigo basin
#

Does anyone know how to install automatic 1111 to use stable Diffusion 2.0 with controlnet?

#

Thanks

torn mist
#

@lime crest Someone suggested I try a different browser, so I did it in Firefox, and that worked fine. So I uninstalled Chrome, restarted PC, reinstalled Chrome, and all is now working as expected again 🙂 Thanks for your help!

distant swift
#

are there any current things that SAI are working on? it's been very quiet since SDXL

rain fern
#

Can someone tell me what folder I need to download negatives into for automatic 1111. I've recently set it up and have downloaded models, Lora's, and VAE into their respective folders but I'm unsure where the neg's go.

obsidian geode
rain fern
#

Actually it appears they don't go under textual inversions folder but here instead, stable-diffusion-webui\embeddings They do then appear under the textual inversions tab in Automatic 1111

faint sequoia
#

Did any if you notice a huge speed increase with the todays automatic1111 dev branch push? On my 3090ti is like blazing fast all of a sudden. Not sure what exactly changed under the hood

pale latch
wind kelp
#

baffled at how dall-e 3 is meant to compete with anything ai related when the world's most normal safe for work prompts that have literally nothing in them get things blocked

robust citrus
#

I've got an image made using Bing's image creator, but the hands are pretty fucked. Is there a way to use SD with controlnet/something else to try to fix the hands whilst retaining the illustration style, or am I stuck using photoshop?

#

Is there a model that's good at redoing an inputted image and whilst retaining the style?

charred hazel
#

hey guys! is there a channel here to ask some questions about loras?

robust citrus
signal mountain
#

Hey guys

#

How can i generate those images that look like another if i squint my eyes?

gentle burrow
#

I am getting Downloading CLIP categories... in terminal, what is it?

still glacier
#

it's downloading stuff, let it work.

cobalt crag
#

I cannot install ComfyUI Impact Pack

lusty beacon
#

what is the difference between stable diffusion web ui and dream studio

fervent thunder
#

Hello, is it possible to buy Credits for a company? We are trying to get an invoice for a company but we have failed to find invoice section

lusty beacon
fervent thunder
#

Thanks

#

Can someone help me out

#

With a

#

Problem

#

After rebooting in windows 11

devout totem
#

i have a question for mods

fallow pagoda
lusty beacon
fallow pagoda
lusty beacon
fallow pagoda
#

Its probably if not the best, one of the bets options

lusty beacon
fallow pagoda
#

From what I am seeing now seems like a clean and easy ui

drowsy robin
#

Hm.

#

How good is StableLM 3B compared to other models?

karmic brook
devout totem
drowsy robin
#

convenient to staff

#

and make a ticket

karmic brook
cunning crescent
#

hello

#

anybody know how to create sdxl model?

wintry hill
#

Hi, was the controlnet removed?

lavish stag
#

Hello, first of all, I’m a BTC maximalist and I haven’t dealt with other projects since 2016, but why am I here with you? Maybe someone will be offended when I tell them that PoS is bullshit, but thanks to this, there is a community that will agree with my opinion that blind faith in PoW is a dangerous path and blind faith that game theory will protect us from attacks on the blockchain. So if anyone wants to help me prove that PoW is not the holy grail, DM me. Thank you for taking the time to read this far.

visual lagoon
#

do people reckon bing's image creator is just SDXL?

#

just wondering if it's coincidence that they're also doing 1024x1024 or if it means they are just training on top of SDXL

covert sand
#

what is a .part file? does it have a pickle?

visual lagoon
#

bro got kenshiro pfp

#

he thinks hes me

devout yew
#

hello

#

does anyone know what checkpoint or model this stable diffusion discord bot is using? Thank you!

wary belfry
boreal kernel
wary belfry
vast pollen
#

The Bing AI makes good images. Is there a checkpoint or something that is as good as the Bing AI?

boreal kernel
#

Which Model is Good To use on GTX 960 4gb card?

wary belfry
boreal kernel
#

Idk much about these Models and Bots, But I Installed 1.5 model by watching Yt tutorial but that doesn't Satisfy me As this Channel bots do

boreal kernel
vast pollen
#

idk either, I just found Bing's new image AI and SD doesn't make stuff as good as that, so I'm looking for a model or checkpoint that is as good as that.

boreal kernel
wary belfry
vast pollen
vast pollen
boreal kernel
#

No way! Could be Prompt Mistake. I used both but This Bots are On Next Level

jovial ore
#

is there a bot that we can use on other discord channels ?

drowsy robin
pale latch
#

*subjective

#

thing with coherency is it's a zero score whenever it rejects a prompt. you've got to count that in the average

#

sdxl could be set up to generate images and automatically reject a huge selection of them for a variety of reasons too

trim magnet
#

Steven Seagal

fervent thunder
#

hello stable diffusion channel!

vast pollen
#

I'm guessing the model for that will be released for SD at some point?

vast pollen
pale latch
#

they're committed to "safety"

vast pollen
#

oh... unfortunate

vast pollen
pale latch
#

part of that is making sure nobody can further train their models too

vast pollen
#

That's so unfortunate because I like running it locally

steep knoll
#

Who is the best LoRa creator here? I want to create children’s books and have LoRas for multiple characters to be able to create books at scale.

Anyone have experience doing something like this?

sonic pumice
#

I missed out on the events...

#

They seemed like a lot of fun

sonic pumice
#

I personally would hire an artist and take a more traditional approach if I wanted to make, and sell, children's books (honestly, doesn't sound like a bad idea)

grand glen
#

how does SD compare to whatever version of dalle bing has rn

#

i really like it but i would like to have the flexablity of SD

merry sentinel
#

Someone using any free procedural prompt generator ?

sonic pumice
#

I just use Dynamic Prompt's Magic Prompt

#

It's great

sonic pumice
fervent thunder
#

so Dalle-3 and Midjourney will know more, while SD remains local and you can fine tune it in order to get specific outputs

sonic pumice
#

I prefer SD for complete control, both in terms of generation, image manipulation and privacy

warm scroll
#

hi

#

whos owner

fervent thunder
sonic pumice
#

^^

#

It's also a hell lot more creative and allows for pure self expression

#

The generation process is tailored to the individual

#

It also means we have to work a lot harder to get the results we wamt, so there's a kind of sense of accomplishment involved too

trim magnet
shrewd fiber
#

Hello, i want to create a text like bagel with a texture of a bagel on this letters. How can i do that ?

spiral holly
#

Hello I want to ask something

#

Is there anyone online?

#

Launching Web UI with arguments: --lowvram --opt-split-attention --xformers --disable-nan-check --no-half
*** Error loading script: controlnet.py
Traceback (most recent call last):
File "E:\automstic1111\stable-diffusion-webui-1.6.0\modules\scripts.py", line 382, in load_scripts
script_module = script_loading.load_module(scriptfile.path)
File "E:\automstic1111\stable-diffusion-webui-1.6.0\modules\script_loading.py", line 10, in load_module
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 883, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "E:\automstic1111\stable-diffusion-webui-1.6.0\scripts\controlnet.py", line 21, in <module>
from ..configuration_utils import ConfigMixin, register_to_config
ImportError: attempted relative import beyond top-level package

#

How to fix it?

covert sand
#

anyone else experience their gpu making a weird buzzing at the end of a generation?

lofty elk
#

Does anyone know when is impainting and outpainting going to be enabled via API?

sonic pumice
#

I think all these questions go there to avoid bloating this channel

devout totem
#

breed me, gru

#

from Despicable Me (2010)

covert dirge
#

Why do faces and hands in photographic mode are often strange, when i try to generate an image? For privacy problems or because it's so difficult training the AI for these parts?

normal hemlock
#

Can anyone here invite me to a 3d printing discord server?

elfin storm
#

Hi. How is it possible to use the infinite zoom on discord bots?

vague galleon
#

Could anyone tell me how to make those goofy AI generated sceneries that look like images?

signal wagon
#

no

quiet sapphire
#

Can anyone recommend some models/tools (stabilityAI or other) to generate variations of an image based on a text prompt? Clipdrop is nice for generating variations but it's not possible to guide it in a specific direction

#

And ideally I'd like something that remains quite faithful to the original image (e.g. I want to give it an image of a room and ask it to change the style and some details, but the room should remain completely recognizable)

vestal dew
#

Dale 3 is still working on an image I prompted 2 hours ago

#

Also, it's not locally hosted...MS controls it, sees all your work, has copies of your work, can change the terms, features at any time. Also, you have limited credits, and even with credits it can take a long time to generate an image.

hasty hornet
grand glen
#

why does img 2 img not work for me

#

i put my image and type a prompt but i swear it has nothing to do with my input image

#

ok i just did a test, it does take the image as an input but it's so bad

covert sand
#

can you run controlnet confortably w 12gb vram?

hasty hornet
#

oh yea and bigger models probably need more vram, but 12 gb is good anyway

grand glen
#

How do you make those landscape memes where if you look closely you actually see something else

grand glen
#

I'm not having a good time tuning a model

#

For some reason nothing wants to work, so many problems with either torch or numpy. Everytime I fix one I get anotehr

idle needle
#

I can’t even train sdxl kekw

#

One I just get a nan error for it or 2 it just refuses to even start

#

Iv went back to 2.1 for now

#

@grand glen

#

I think there’s a bug with sdxl

#

I shouldn’t be using 23Gb for 1 batch size

grand glen
#

I haven't even gotten that far

#

Every tutorial is different

#

I've downloaded so much stuff already

#

And the downloads are like 1mbps it's so painful

#

I want to train it with my cat

#

My cat was a stay and has lots of scars and stuff so he is very unique

inner lintel
#

Any plans for Text-to-video AI? (similar to Gen-2 by RunwayML)

distant swift
#

hey guys, I finally figured out why OpenAI is called 'OpenAI' while never having any open project! the Open in OpenAI stands for "Open your wallet"

#

this makes so much more sense

dense cypress
#

can't put an image ton prompt from ?

leaden mural
#

Anyone got a good colab alternative for free SD generations?

broken gull
grand glen
#

ill give it a try

torn mist
warm junco
torn mist
#

Thanks @warm junco

gray wolf
#

Can stable diff do text in images like Dall e 3?

wise aspen
#

ok guys, where is style selector in last verison? i cant find it after last patches

#

it was at right under genearte (automatic1111)

#

and now its where?

covert sand
#

do you need as high of vram to create a given resolution if you're just resizing as if you are generating an image?

lavish lake
idle needle
#

im training a SDXL model

#

its going slow but well

dusky sentinel
#

Guys, how to do img2img here?

thin abyss
#

What happen to being able to use a picture for reference? The prompt for its gone?

onyx hedge
#

bonjour, gros probleme avec pitorch apparement

#

je ne suis plus à traduire les choses ou faire l'effort de parler anglais. vous faite un boulot formidable

#

Plein de bonne vibe pour la continuation du developpement ♥

#

If i find something around openpose, i'll communicate you , cheer !

#

Pytorch x D and dependancy ♥

onyx hedge
#

pytorch ....

#

how can i downgrad pytorch

#

that all i think i'm a noob Tittle is i'm a noob, how to i can scale stable difusion to 1.5.2, and how to keep day line on all dependency from all dependancy

#

x D

#

Have a nice night i'm close to give away, my knowledgies is small have fun .

#

tirred...

rancid spindle
#

whats the best free upscaler rn?

nova oxide
silk coral
#

@alpine ingot

#

YO

upbeat tide
#

Is anyone here good with music and notes? I have a sound I want to play so it 'matches' (harmonises?) with a map's music in-game. If you're happy to help, please DM me. Thanks. 🙂

fervent thunder
#

uh?? Why I can't access deviant art when bing image creator is opened? For some reason I have to close image creator, else deviant is blocked

forest sierra
#

hi there ! i have a problem with my diffusion

#

can you guys help me

#

It says ''RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check''

warm junco
fervent thunder
#

can someone Interrogate CLIP for me please?

#

mine doesnt stop counting

#

I just need an image

forest sierra
forest sierra
warm junco
#

And make sure to edit the webui-user.bat in this line:
set COMMANDLINE_ARGS=--medvram --no-half-vae --opt-sub-quad-attention --opt-split-attention-v1

fervent thunder
#

/vote a

iron willow
#

quick question, is it possible to generate the rest of the scenery from first picture?
i.e. from a face-only image to full body, or from full body to include surrounding scenery.

wispy jasper
#

Does anyone have any idea on Dalle-3 Tech? I Mean why iss it SOOOO effing good.

plucky shoal
#

@iron willow i think only with outpainting

green void
#

How many average people mess with stable diffusion is the majority graphics designers? I notice theres alot of features not complaining its lit.

rain fern
# green void How many average people mess with stable diffusion is the majority graphics desi...

I'm an average person trying to learn it, but yes its not easy to start. I have downloaded a ton of models which I'm working through to see what I like the best. I'm finding some Lora's work well with one model but terrible with another. Also drying to master inpainting but I'm thinking I may be better to photoshop the image then put it back in the image to image to get a better result.

grand glen
#

how do you make those squint your eyes pictures

carmine spoke
#

anyone else have an issue with controlnet ruining background? When off, my images generate with the correct background as i describe, but when trying to incorperate a pose using openpose (or any form for that matter) it completely disregards my scene description and just does like grey backgrounds

grand glen
#

someone told me it's like 15 billion params

#

and SD is less than 1 billion

obsidian geode
#

Any controlnet for making your img2img more black and white?

inner lintel
#

Dalle3 may work better at 1024x1024 resolution, but with that censorship it wont allow to generate cute anime-style images, which is strong side of finetuned SD checkpoints

#

What's the point of that proprietary software?

safe crater
#

Hi, am new member, wanna ask, How to generate person with name tag and its lanyard with realistic checkpoint, i use Automatic1111, is there any lora for it? Appreciate the answer 🙏

upbeat tide
potent spire
#

First of all ChatGPT plus is already a "must have" for some

#

DALL E works seamlessly with it

#

And you dont need to make some tedious setups

#

Tedious for some people

#

And you dont need a good VRAM for that

bleak matrix
#

Good morning, everyone!

upbeat tide
#

What word/phrase to get a topdown view? I tried (topdown view:2) and I get a regular side-on view,.. any ideas please?

deft summit
#

Anyone knows if animatediff works with stable diffusion xl?

bleak matrix
upbeat tide
merry sentinel
#

"RuntimeError: Not enough memory, use lower resolution (max approx. 704x704). Need: 0.5GB free, Have:0.2GB free"

How can I increase the threshold to generate high quality image?

lavish lake
#

try 640 x 640 or 512 x 512

#

then you can 4x upscale with no problems

runic breach
#

How do i get less people in my images, when i want a 16:9 style format it always gives me like 2 or more poeple, very Noob question sorry

plucky shoal
#

you want just one person @runic breach
you can try 1woman, or 1man
Or learning Regional prompter, which is powerful extension which can help you control image very. 🙂

obsidian geode
upbeat tide
#

Hm,.. thinks Maybe add some fake colour in post then reimport it back into SD and go from there?

obsidian geode
obsidian geode
upbeat tide
#

Tbh, I don't know enough myself,.. I'm still pretty new to it and still getting a grasp on it.

eager tinsel
#

Hello

#

How do i install SD for mac?

lyric geode
#

Anyone know how to get a proper top-down image? I'm trying to get some minimalist icons of ships, but it seems incapable of it.

#

even using "satellite imagery" will only get a side view

sullen walrus
#

anyone here know how to use controlnet?

upbeat tide
lyric geode
upbeat tide
#

Hm,.. first, get your subject's shape/form right,.. then add/use Lora,..

lyric geode
#

If I don't use the Lora it's either not top-down, or definately not a ship

upbeat tide
#

Change CFG,. and sample steps.

lyric geode
#

Doesn't seem to matter what I change it to, the models seem incapible of displaying ships from the top down.

sweet kiln
#

has anyone had success with SD and making QR iamges that actually scan instead of looking like QR codes?

covert igloo
#

Hi

#

Need some help

#

I am working on a custom photo based QR code

#

And I am not able to get desired output

#

Any ideas?

sweet kiln
#

same

blazing quest
potent spire
#

or vice versa

forest trout
sullen granite
#

Speaking of Dall-e 3, I made a survey on whether people can find the difference between AI generated images and human made ones, I'd invite anyone here to give it a shot: https://forms.gle/7Bu3Tyi5fufmY8Vc8

#

Spoilers: Of the ~900 participants so far, only 14 people got a score of 15/20 or higher. Nobody can tell anymore, basically.

unborn basin
#

Hey hope this isn't strange to ask but what chat do I use to ask for tech support in stable diffusion?

pale latch
#

people can't tell when shit's fake already. its no surprise they can't spot synthetic imagery on average

#

staged pranks, instagram filters, obviously scripted reality shows. people all falling for it

#

nothing to do with how good ai is. people just be dumb mostly

carmine stone
#

this is great, thanks. We just published this article about using the new ChatGPT-4V multi modal to help improve and accelerate captioning datasets for SDXL training https://t.co/3kmvS8T9jd

magic citrus
lusty beacon
#

is there a model which is close to dall e

deft sentinel
#

hlo

stable bough
#

Someone can tell if the arc 770 16gb is better then rtx 3060 12gb?

plucky shoal
#

check benchmark in general with images @stable bough

stable bough
#

Is it worth to have one, or is complicated? It will have support for it for next stable diffusion updates? things like that

plucky shoal
#

ppl with arc and 3060 personal experience... very few people if any. Nvidia future is way brighter. On other hand arc has 16GB, but check amd with their VRAM and its unusability in SD

stable bough
#

Do you mean Radeon RX 6600 like cards?

stable bough
#

From arc

#

And the price is very cheaper than others 16gb cards

fervent thunder
#

who wants to work with me I pay 500 usd a month

fossil phoenix
#

Что у тебя не так? Ты что, душевнобольной? Не говори так. BruhWtf

pale latch
#

looks like a crypto scam

untold lava
#

Quick curiosity question; how many sample drawings would it require to train the AI on a specific style?
Asking on behalf of a discussion in another chat ^^

flint jewel
#

hello all

carmine stone
lofty girder
#

does sd-webui work on rocm5.6

swift swan
#

How can I edit my photosadcat

raven token
#

Question. Is there a place for the rejects? When SD generates ugly? That is something I appreciate about SD is when it makes an ugly image. Is there a gallery for ugly SD?

trail wing
grizzled spire
#

In many places AI art is getting banned and now Steam is banning AI generated games which is a good thing.

winged trench
#

Sorry if this was asked before (can't really find a 100% clear answer) , can you run stable diffusion if you have a good cpu but old gpu ? I've heard you can prioritize cpu usage but idk at what extent

#

(also I'm probably in the wrong channel now that I scrolled the list a bit)

young ivy
#

Quest 3 is so tempting, for gaming, mixed reality (entertainment and work application like sculpting life sized models and getting in there

vocal vector
#

What happened to the COW channels?

young ivy
#

they were killed to prepare for some "new ideas", the same thing happened to POW half a year ago I think?

vocal vector
#

Was there an announcement about that?

young ivy
#

not in public afaik but there was in the stable society channel

#

where us black names hang out

#

to make it clear, there was no announcement

#

just a comment about it

#

so nothing official or unofficial and they're probably revamping the competition space entirely

vocal vector
#

Ah ok, thanks for the info!

young ivy
#

if there is an official announcement and I missed it, then that'd be my bad for sure

#

but I hope that when it comes back, it comes back with some really cool contests

timid kiln
#

is there a tutorial somewhere to just get stable diffusion?

grizzled harness
timid kiln
grizzled harness
timid kiln
#

Local

grizzled harness
#

what went wrong?

timid kiln
#

I forget its been a while. Ill try it again tho and see

zenith prawn
#

Hello, is ist possible to make on the same image a character both from face and profile angle ?

vast ingot
zenith prawn
ebon flower
#

epic realism only gives 1 arm on the model?

#

any fix

torn wharf
#

hello

azure bobcat
#

hi, does anyone know how to edit your photo using this bot?

charred thistle
#

aah, general general

#

finally got stable diffusion running in GPU mode on an RX590

charred thistle
#

lmao, I was almost encouraged to buy a new GPU just for this

bleak matrix
#

Good morning, everyone! How are we all today?

manic lynx
#

Have a quick question. Am still learning. So I understand what a VAE is, and what inpainting is. But I noticed when browsing civitai that for example rev animated has both an inpainting model, and regular. I am guessing this means if I want to do any inpainting I should also download the inpainting model and use that in those cases? My other question is related to the epicrealism model. There is a regular one, ands then one that appends VAE to its name. Does that just mean it has a VAE baked into it already? Thanks in advance for help.

past frigate
#

so any instructions for newbie who got here because he saw Dalle 3 pre-censhorship ? have just 8GB so i will propably have to rely on online models until i can get a proper computer after my old one went kaput which will happen in year or so.

manic lynx
past frigate
#

thanks im gonna look into it.

manic lynx
cobalt garden
#

Is anyone having issues with SD on colab? (paid version) nonsensical outputs

charred thistle
#

well that's weird.. seeing an image get clearer and clearer until 2/3 in it starts going backwards

tight valve
#

Hi everyone, I've been trying to find a good method to help me achieve this kind of effect: https://www.youtube.com/watch?v=gZprUqsLOyQ&ab_channel=uisato
But can't seem to make it work... I am using stable diffusion with control net enabled, I've also tried img2img, it's been driving me crazy. I've tired 5 different models and some Lora's as well but my effects are nothing like the ones in the video. Does anyone have any idea on how to do this? Or any good models which would help me achieve a similar style etc.
Thanks.

feral eagle
#

no matter what i do controlnet and openpose wont work

simple swift
#

I have question on SDXL like Image generator on hugging face ? is this a GAN connectig with TextToVector convertor ? There is an other place for this question ? I try some projects on CNN but never on that

#

or a transformer ? a random forest algo ?

warm junco
feral eagle
# warm junco why?

No idea reinstalled everything , watched start guides…. It something up with template for runpod

grand sail
#

I have entered the domain

#

You cannot stop me

fluid imp
#

Why does the interface translation plugin not work, I see in comfyui can be translated

#

StableSwarmUI

zealous dew
#

Can you rotate building with stable diffusion using controllnet?

#

Does anyone know?

fervent thunder
#

https://youtube.com/shorts/dwwBvDE37L8?si=OLc5Q5ebo1qEXbqy

A friendly hello to my other ai enthusiasts here. I am
currently making a short film with stable diffusion, I
nave already made a small teaser, in which you get
glimpse of what it's about and already get to see
some of the protagonists
Let me know what you think about it.

Voiced by the ai voice of Morgan Freeman

grand sail
#

"Hello, my name is morgan freeman!"

#

"Welcome to the next step of evolution, A.I and A.I accessories! I am your guide into the world of A.I"

onyx birch
#

I'm a noob at all of this. How to get it to stop putitng out legs with hands or arms merged into legs. I put a negatice promt in for that. Maybe I'm doing sometihng wrong? Or is it someting I have to deal with?

warm junco
#

For 1.5 models its 512x512

zealous dew
#

I wish i can

#

With RX 6600 XT

onyx birch
warm junco
#

With highres fix or jmg2img upscale scripts

onyx birch
#

that's what I'm doing wrong. Thank You

spice plank
#

Upscaling kinda sucks tbh

onyx birch
#

The humorous part is that, sometiems you get some really nice "adstract" art when they come out all distorted.

patent geyser
#

yo, I am new to this and I ma having issues that nothing online is helping me with

#

anywhere on the server I can put my issue?

zealous dew
#

Dose anyone knows any extensions that will rotate on its axis?

peak frigate
#

Anyone know why my lora tab dont show on my 111 ui ?

placid comet
#

A

valid quartz
#

Hello everyone🤚

#

Can I ask one question?

#

I have an avatar and I want to improve it (add a realistic style) is it possible to do this with your bot?

#

Or do I need to install a1111 and do it through it?

gray palm
#

Hello
What is the current best promp generator?

vast pollen
#

How do I use safetensors, I click on it from the dropdown like a checkpoint, but it's as if it isn't selected

forest trout
#

What program are you using to run stable diffusion?

iron imp
#

o you were prolly replying to them and not just asking .. nvm

forest trout
#

Yeah I was trying to help them troubleshoot.

rotund wolf
#

1

dawn reef
#

reImagen ??

#

anyone ever use it?

prisma cobalt
#

hi

jade citrus
#

Any ideas why webui crashes my computer , started last night when trying to generate, it's never done it before , comfy ui is working still

fervent thunder
signal dome
#

hii

#

havent been in a while whats new with SD ?

fervent thunder
#

Pls🙏🙏

plucky shoal
#

i have 3070

grand sail
#

I have awakened kiddos

#

I have come to remind you all to be a homie to the cause

#

Screencap the page pls

warm junco
#

stop asking for this. Temu never gifts anyone free stuff. Its a scammy system to get more User Data.

grand sail
#

I don't trust no embedded images

#

There goes my meme time

grand sail
vale mirage
#

Where make picture ia ?

grand sail
#

Wat

mystic tiger
#

Hi guys, does anyone knows what happened to the controlnet option? Why can't we upload pics anymore?

grand sail
#

Uhm, id assume abuse

deft summit
#

hello is there a way to change ConfyUI folder for the models so i dont have to copy them again

pearl jungle
#

Hello, does anyone could recommend a good AI solution to transform my book into audiobook ? ideally that could run locally

fervent thunder
#

AI has to be trained to generate images. To be trained you must use photos or illustrations and most of the time all this data is copyrighted. Is AI generating images by randomly choosing elements from its database, meaning that it's "plagiarizing" art?

fervent thunder
#

I got this into the cyberpunk contest

#

there's music in it, and I also left my resources in the description for the post you just gotta hit view post, my deforum settings file is linked too

#

would love some positive reactions! 🫂

vocal vector
runic merlin
#

Can you recommend me a good model that was trained with only landscape and urban scenes or stufffff

#

im tired I cant avoid having people or "people" in my animations and I add all the possible negative prompts

#

and no luck

hasty hornet
#

go for higher priority (or whatever those numbers are called) on negative, like (man, woman, girl, boy, human:1.3) or something, it worked for me before

#

but there are some models on civi for landscapes too, won't recommend any since not using them ¯_(ツ)_/¯

runic merlin
#

I did the priority stuff and anyway nothing xd

frosty pasture
#

Any advice for making fantasy battle images?

#

Like, knights, dragons, etc? 😮

fervent thunder
#

images made by AI, do they qualify as brand new?

crimson belfry
#

i feel bad for people using AI that bans them from making political memes

fervent thunder
#

it's more or less the problem with comedians I think. How often are comedians charged in the law for making jokes of politicians?

latent gull
#

What are some good tips to make the AI actually listen to what I want it to do.

fervent thunder
#

Hello. Where did the button to save the current prompt as a style disappear to?

fervent thunder
marble stump
#

Guys, yesterday in discord I was generating an image of 3000 pixels, today only 1024. What’s the matter?)

potent spire
remote umbra
#

Has anyone tried duel gpu's

rain widget
#

Hi there

spare spade
#

Hi. Why is it no longer possible to generate based on some kind of image?

limber birch
#

Guys , I have 2 GPUs , and when I run webui it only uses one of them, how can I make it use both ?

#

Will it be worth it, or generation speed will be same

plucky shoal
#

you can use it in parallel each gpu on its own. Somehow using #🐝｜swarm-ui , try ask there.

charred thistle
#

hey peeps

#

hey.. do y'all think it might be worth upgrading to a newer GPU for this?.. cause like, it starts to chug after a half dozen renderings in SD1.5 based models when I add in a LoRA or two

warm junco
charred thistle
#

RX590

bleak matrix
#

Good morning, everyone! How are we all this beautiful fall day?

flat belfry
#

Well I finally had a working version of SD 1.6.0 with 1.5 and SD 1.0 XL models working just great until this morning. Until I tried to make a pic with a SD 1.0 XL LORA inside the prompt. Now I'm getting the teaser pics of my images then when finished was getting peach colored squares. So I restarted SD 1.6 again and I'm getting grey squares when I try to create any 1.5 models, but 1.0 XL seems to be working fine. Does anyone know how to resolve this. Should I try removing that Lora That did work with XL? TIA, James

warm junco
# charred thistle RX590

The directml Version on windows isn't as good as if you where using auto1111 on Linux.
Hopefully you use the recommended Commandline args for best performance.
But yea it might be worth an upgrade

charred thistle
#

I'm using auto1111 on Linux, but using ROCm

warm junco
#

7700xt or 7800xt is worth a look maybe

charred thistle
#

but, the RX590, and related GPUs do require a hacked version of torch in order to work with ROCm, since AMD dropped official ROCm support for it. I hear RustiCL actually outperforms ROCm and works with a much broader range of GPUs as well, but I don't think stable diffusion utilizes that, although, supporting open source drivers seems like a good idea for open source software

#

i was considering the 6700xt for around the $300 price point

warm junco
charred thistle
#

RustiCL I believe is what Linux's Mesa drivers use, which do outperform AMDs official drivers for gaming already

warm junco
#

Better then 8 for sure

charred thistle
#

7700 is 12gb VRAM as well

warm junco
#

On Linux with rocm it should be really fast

charred thistle
#

the 6700?

warm junco
charred thistle
#

hmm, interesting

#

I'd have to wait until prices came down on them though

warm junco
#

All 7000 GPUs have the ai cores

#

But for now there isn't a use case for SD

#

I think

charred thistle
#

the issue I'm having now though seems to be that with every consecutive use of stable diffusion, my system RAM gets higher and higher to the point I have to close and restart it to prevent a complete lockup

warm junco
charred thistle
#

yes, 8gb

#

16gb of normal RAM

warm junco
#

If you have enough SSD space I would make it bigger

charred thistle
#

i'm using swap on an HDD

warm junco
#

Oh

charred thistle
#

but I have an NVMe I use for gaming

warm junco
#

Okay

#

Maybe you can put it on there

charred thistle
#

swap would result in a lot of read/write cycles

warm junco
#

But if you plan on upgrading I would also buy 16gb ram to get to 32

warm junco
charred thistle
#

i have an extra set i need to send in to have replaced since they're buggy, that'd give me 32, I might be able to stick them in for a while and see how much it improved before they start causing issues and have to be removed

charred thistle
#

ok, 32gb installed now

charred thistle
#

yeah, python alone is now using 20 gigs of RAM

twilit turret
#

Am I the only one who gets a good chuckle when people make fun of how creepy and uncanny AI art can turn out?

charred thistle
#

i made some absolutely horrific stuff with dall-e, would love to be able to do similar on SD

#

like i wanted a bunch of ghost children screaming at the viewer... for some reason it decided to remove their eyes and have blood coming out of the sockets... didn't ask for that, but will gladly accept it as a horror scene

somber mortar
#

What's the difference between generating images on here and locally with the other stable diffusion?

warm junco
halcyon crown
#

hello, how do i train a model on my frnds face so i cam make some epic images for him?

warm junco
somber mortar
#

it takes up 100% usage of my GPU and my whole computer just freezes

#

this one image was taking like so long to generate that I had to close everything and restart my PC just it was still at 100% gpu usage even tho I closed it

halcyon crown
warm junco
warm junco
halcyon crown
somber mortar
warm junco
charred thistle
#

whole PC freezes when RAM runs out, usually

somber mortar
#

seems like my memory is already at 55%

#

and im not doing anything big

#

weird

warm junco
cunning saddle
#

I have a question...Do you know how to have MEN using Stable Diffusion? Because it seems like all the checkpoint, LoRa etc... give me WOMEN by default even if I write 1man, MAN!!!, Male etc...in Positive Prompts and WOMAN, 1 girl etc... in the Negative prompt! Is there a secret???

fervent thunder
#

are images generated with AI considered to be new, from ground up? Not a copy of existing art?

shrewd coyote
#

maybe that's the model's way of saying "no homo"

idle coyote
charred thistle
#

yeah, I'm budgeting for a new GPU now

charred thistle
#

leaning towards RX7600

feral eagle
#

So is the intel arc a770 a viable solution now for SD / textgen / ml?

charred thistle
#

anything recent seems fairly viable to me

#

my only concern with intel, and I could be wrong because I haven't looked into them much, is that they're fairly new to the serious GPU market, so I'm wondering if their drivers, and supporting software are up to par

#

stuff like CUDA and ROCm have been brewing for a while

somber mortar
#

I just installed Foocus

#

Is it good?

mystic tiger
somber mortar
#

whats a really good model to use

#

?

mortal delta
charred thistle
#

is gfpgan better than codeformer?

#

because it seems like I tend to get faces that are perfect without it, or still messed up with it

thin ledge
#

Hi guys! I am a dirty AMD user and have been using SHARK AI to make SD run on my graphics car otherwise it takes forever. SOOOO.. How do I find the options to change styles and all those other options? I only have text to image and what I get is what I get, no image options other than that

#

Thank you so much in advance!

somber mortar
#

realistic mostly

mortal delta
# somber mortar any

i find Deliberate to be good if you you want a general model that does a mix of manly real with a little bit of cartoon, theres probably better ones but that is a start.

tropic urchin
#

Does Bing ai art steal from artists

charred thistle
#

I like models that do a really good semi-real art style, like a modern video game that isn't trying to look photo realistic

#

but I still have a ton to learn to get what I want out of them

fervent thunder
#

is stability working on an image model rn

wispy jasper
fervent thunder
forest trout
#

I'm sure they are working on the next model but I highly doubt they have any interesting progress updates to discuss right now.

wispy jasper
#

They worked for ever on SDXL< it just got released. There is no more tech beyond what was released. We are at hardware limits now.

#

Dalle3 is amazing b/c of its insane paremeter count and doubly insane hardware that generates it.

forest trout
#

Improvements can be made but it's mainly clip related imo.

wispy jasper
#

It the next models had better textual performance like Dall-e3 that would be a welcome addition

forest trout
#

It's been a running issue with stable diffusion. DALLE has always been better at conceptualizing text into images but shrug

south veldt
#

anyone happen to know details about the issue with the clipdrop.co webapp site? free pricing plan is still active but any attempt to actually generate images on the website (selecting Clipdrop SDXL or Doodle) results in a popup promoting subscription

frosty pasture
#

guys

#

where can i ask for tech help?

#

i failed to install roop, because i put the visual studio files on a hard drive different than C, therefore when I did the pip command to install face insight, it goes "files are not ON PATH"

#

i have the error to send, but... where do i send it to, to ask for halpz

upper epoch
#

does anybody have any experiance with amd instinct cards? how slow/fast are they for training?

inner lintel
#

Stability has released first version on Stable Diffusion shortly after DALL-E 2, and they released XL when Midjourney had a comparable quality of output images.
So, we may see an opensource alternative to DALL-E 3 soon, as proprietary AI software is just one step ahead. It's mostly matters of text encoder that tokenizes user input.

kind kettle
#

Hey, I was wondering if anyone could help me figure out a fool proof prompt for coloring pages that i can print off, hopefully looking for a prompt i can copy and paste then add what i want the content to be

inner lintel
lost tundra
#

hi

charred thistle
#

Hmm, the Arc A770 is under $300? I thought that was $400+

#

I'm either going to wait for a 12gb 7600 XT, get that Arc, or a 3060 12gb

dusk scarab
#

Hello everyone, new here!

I'm having trouble finding information on how to achieve certain results, as most of the public information usually involves realistic results, or blender 3d-like characters. I am looking for some kind of cartoonish style with details, mostly splash art / cover art for games to basically nail down some character concept art to help create moodboards.

Anyone has experience with similar results?

sturdy mist
#

so im installing but idk which i should do 1111 or comfyUI, if anyone has input on which i should go for as a beginner pls lmk.

warm junco
charred thistle
#

i'd be down with trying comfyui, but it'd be a pain to get working with my GPU

warm junco
#

But idk if the rx590 needs some special stuff like you needed for auto1111 too

charred thistle
#

i'm using an RX590, no ROCm support since version 4. something besides a hacked version of torch which i can only seem to get running out of a docker container

warm junco
charred thistle
#

automatic1111 is separate from stable diffusion itself though right? I wonder if I cuold install comfyui to that container, and launch it accordingly, but that download I have has the separated torch version, i'm just not sure how to link it all up

#

screw it, once I'm done with dinner I'll give it a shot

sturdy mist
charred thistle
#

comfy ui is the one that uses visual scripting, right?

sturdy mist
#

yeah nodes is the term im more familiar with but yea visual scripting

charred thistle
#

is there any improvements or benefits to it though?

sturdy mist
#

like i said im legit really new to this i think maybe it had a few improvements but i think its more for the node aspect. im not sure tho so anyone can feel free to correct me.

charred thistle
#

all I know is that after a few minutes of using the webui, python3 is taking all of my RAM.. I even put in a couple more sticks to bring it up to 32gb, and it took up all of that

#

plus 16gb swap

sturdy mist
#

ah i see. but is that really all that big of an issue? genuine question

charred thistle
#

it's a serious issue.. I have to close down the application and reboot the web UI after half a dozen or so images

sturdy mist
#

ah

charred thistle
#

comfyui seems to be much easier and more stable with RAM... so I might try it if I can get it to run

mellow meteor
warm junco
#

Without downloading everything twice

sturdy mist
warm junco
#

Like you type blond and it gives you "blonde hair" as suggestion

sturdy mist
#

ah ok ty. does it matter if im installing multiple models and would i have to restart the ui after each time aswell?

warm junco
#

You can switch between them without reloading. But reload if you add new models into the folder

sturdy mist
#

also do i put every model into the models/stable-diffusion folder?