#💬｜general-chat

1 messages · Page 65 of 1

distant swift
#

yes, i use image browser favorites tab, then press send to txt2img and boom, all that's left it to change seed to -1 and your done =]

waxen frost
errant oyster
#

Is there a way for ADetailer to avoid all faces and just do one like its detailing the guy and girl

frank spruce
#

I feel like the inpainting workflow needs a better GUI

#

controlnet global harmonious should be on by default imho

potent phoenix
#

Guys

#

SD is taking my cpu instead of my GPU how can i fix that?

#

please :( im asking since yesterday

#

i haven't been able to do much since its so slow with the cpu

frigid pelican
#

This can't be posted because it contains content blocked by this server. This may also be viewed by server owners.

#

how do i fix this?

#

i cant type the dream command

karmic brook
frigid pelican
#

yes

#

can't actually type it ANYWHERE

#

its odd af

karmic brook
#

check your settings for slash commands

frigid pelican
#

ah, turn it off?

#

ill try that

karmic brook
#

it should remain off

#

if it is off, restart your client and try again
if the command does not appear, some users reported they could use it once rejoining the server

frigid pelican
#

nope that didnt work

#

okay ill try and restart

#

thanks for your help btw

karmic brook
frigid pelican
#

that did it!

#

thanks fruit!!

karmic brook
#

awesome ^^ happy dreaming!

frigid pelican
#

lol ty

frank spruce
#

inpaint should ask for confirmation if nothing is masked at all

vernal ore
#

hiu

timber lodge
#

Greetings. I'm new here, but.. i'm trying to figure out how to get help with a ui issue. I'm trying to find out if there's a channel i'd ask my question in?

open night
#

Back when stable was pretty new there was a popular guide to installing it with automatic111, the voldy guide or whatever. Is there a up to date popular tutorial for the current stuff that kinda includes all of what you'd want in a easy to digest format?

#

I got a new pc with 12gb vram so i'm looking forward to trying out some training which weren't really possible on my old card

#

If there's no longer a popular tutorial for installing stable diffusion like back then i'll probably check out some aiprepreneur video I guess, yeah this will probably do may aswell start now

open night
#

ok that was easier than i remembered lol

craggy sundial
#

Hello, there are different models in the web ui. I want to know how can I use a specific model to generate images here in discord like in web ui?? As discord makes it easier for me to do the work. Can anyone help?

shy pumice
#

you can't, far as I know

#

the discord bot just uses the SD XL model

#

there's probably some way to set up your own backend for a discord bot, I know stability was working on something like that before

#

but I'm really not up to date

craggy sundial
#

I really wanted to use chilloutmix and chilled_re-generic model 🥺

unique hamlet
#

I heard there’s a discord room where SDXL is being tested currently. Which room exactly are people talking about?

vast thunder
#

Stalker lol 😂

hasty hornet
sudden ruin
prime fiber
#

Good evening, can anyone tell me if it is possible to add the StableDreamer On a server or does it only work here?

sudden ruin
prime fiber
#

thank you 🐧🫸🫷

fervent thunder
#

Hello there! Can someone answer me some quick question? I dunno much about this ai generated art but what about copyrights, when I generate a image whines the owner? Can several people use or copyright an image generated by others?

#

If there’s already a discussion about this topic in here, someone please redirect me 😌

young ivy
#

i think the law says everything generated purely by AI without external modification etc has no copyright to anyone

sudden ruin
#

The law depends on where you are located

midnight vessel
#

heya got a stupid question. Does it matter if I have heaps of models/should I delete the ones I do not use?

#

or does it not matter

fervent thunder
#

why does kohya download so many gbs of files?

sudden ruin
fervent thunder
#

Does anyone know how long it takes for an Apple M1 to generate images?

fervent thunder
#

I was asking about copyright because some generated images come with a signature on the bottom … anyone knows something about those signatures?

candid flume
#

Wasn't there a website for this ai before ?

wild steppe
candid flume
stable pawn
#

Help me find a model.
I want to do "(teddy bear) in a pink onesie pyjamas, solo" but I often get a human, so I added "nsfw, human, homosapien, female, man, male, girl, child, kid, baby, anime, manga, cartoon, drawing, sketch, painting, 3d, render", but it still give me humans and anime. SD1.5 give me a teddy bear but often out of frame, other popular models that I downloaded tend to go towards females pose sexy and the word "pyjamas" is like it was connected to "child" or "baby" so that is what I get.
So what model shall I download to get (semi)realistic teddy bear in a pyjamas and to avoid kinky poses?

pale latch
#

use dreamshaper. use negative prompts.

frank spruce
magic citrus
plucky shoal
#

probably more accent on teddy bear () is not much try ((())) @stable pawn

pale latch
#

copyright is a right. Rights are inherent to people. They can't be redefined by some clerk.

#

it's dumb to think that somebody wasn't there intending the creation. AI is just a tool.

#

don't be dumb

sudden ruin
pale latch
fervent thunder
#

Sorry, I’m kinda newbie 😌

pale latch
sudden ruin
#

Yeah I never said that, but you're right

pale latch
#

yeah its not like the ai has the trained images stored anywhere. they're long gone. it just has knowledge about how to predict denoising based around those images

sudden ruin
#

Especially if you seek deeper knowledge of the process

pale latch
sudden ruin
#

I think I watched that already, I got to the point where I think I understand it but it's still hard to explain it to others for me dead

pale latch
#

i rewatch it a lot because i like his accent

sudden ruin
#

Next week I gotta present some people stuff about AI and it will be a disaster kek

plucky shoal
#

@stable pawn
Teddy-bear in pink pyjamas
and realistic vision v 20 <- model

pale latch
#

RV would likely put a naked lady in there waering a teddy lingerie

#

it's got strong biases

#

all those realism models are trained on hardcore pornography

sudden ruin
#

Not all of them, but a lot

plucky shoal
#

i think - glue it for clip better understanding.
It working for cats, others not, and it working on teddy-bear, so i cant complain

stable pawn
pale latch
#

analog is a good one but it's stylized to film

hot lagoon
#

hello 👉 👌

hot lagoon
rain egret
#

Hello. Is someone here experienced in training Loras for SD? Its not a character. Its not directly NSFW but might lead there during the process. I want to train at home. And help is highly appreciated.

chilly echo
#

Was making a batch of images and half way through, it started making pure black pics. What happened?

plucky shoal
#

can you post screen of console? cmd i mean?

chilly echo
#

Posted there

young ivy
#

right now, that's what we have to go by

#

until challenged in the court of law

pale latch
#

lol link storm. Those all likely link to the same incident of one clerk saying it didn't count for reasons they made up

#

clerks don't define your rights.

#

imagine if you go to mcdonalds, and the clerk there decides that since they don't like your face then you don't get your human rights. lololol

young ivy
#

It was a decision made by a government agency, so yes whatever you want to call them, they do decide that

pale latch
#

you can argue all you want. You're being dumb though. "the government" is not one clerk

#

also, that was regarding registration only

young ivy
#

you keep saying one clerk but I'm not seeing evidence of it being one clerk

pale latch
#

Well yeah. You'll cherry pick your evidence because you're being dumb. Here's something to get hung up on. It's copyright. A right.

#

oh i see you linked the zendaya comic. Yeah. That wasn't given copyright because it was using zendaya. whatever her name is.

#

Dude changed it and got copyright. wow. cherry picked evidence lol

young ivy
#

You can keep trying to put me down to somehow invalidate what I'm saying but that's not gonna work. You're just showing how immature you are

chilly echo
#

Is there a way to add a list of things it can take as "or"? Whenever I make batches, it gives a lot of similar pictures which is handy at times, but sometimes I want a variety of backgrounds, outfit styles, etc, and wanna just have it make 20 pics or something.
Eg:
Eye color (blue|red|brown), outfit (hoodie|spacesuit|tshirt)

pale latch
#

Well i mean, when it comes to invalidating human rights, i'll treat that person as dumb as they're being. Rights. They can't just be decided by some clerk.

#

Copyright is a right that's given when you create something. simple as that.

#

Link dumping is by far one of the most immature argument approaches, so before you start bragging about how your maturity levels are over 9000, take a breather and reflect

young ivy
#

I don't know why you have this very hostile and insulting attitude towards me...But you don't seem to understand that copyright isn't something guaranteed. There are rules and sometimes they add new rules with the changing landscape, especially now with AI. You're not guaranteed a copyright just because you made something. If it falls within the rules, then yeah. Otherwise, no

#

Damn you have an attitude problem, can't you communicate like a normal person? Are you like 15 or something?

#

I'm done, all you want to do is insult me

pale latch
#

Copyright is implicit upon creation. It's a right. Don't make this about you. you're being dumb again.

#

It's not about you

pale latch
#

your example is actually really close to their syntax. you'd type {blue|red|brown} instead of ( ) and it generates prompts for all the varietys

chilly echo
#

Tyvm. I'll check it out.

pale latch
#

heads up though. When you write a big long dynamic prompt, save it somewhere. The images that are created will have the baked prompt without all the dynamic syntax

sudden ruin
pale latch
#

When people are saying rights can be taken away by a clerk, and i explain why that is dumb, i'm not **just **calling people dumb.

"thanks" << Don't be so passive aggressive. Be more sincere and honest. It'll improve you as a person.

sudden ruin
pale latch
#

I don't know how to respect people who are agents of eroding rights.

#

It's the same idea of tolerating the intollerant. I just don't know how to do that.

karmic brook
#

If you'd like continue the convo, feel free to do it within DMs with those who want to continue
for here, let's stop the convo and move on as it seems to be effecting everyone involved poorly

#

no bad blood or hard feelings melodyhappy

pale latch
#

Definately. Coming into a creative space and arguing that people don't own their work and that copyright doesn't cover artists working with AI, is a very negative energy.

If i engage someone on this topic further i'll certainly get banned from all discord.

karmic brook
#

Nope, it's just that sometimes it can get heated and the topic at hand can spiral

pale latch
#

These trolls know how to push buttons too. Link dumps for example. Okay. Done. I got to head out anyways. Be creative. Live long. Prosper. All that positive stuff to go out on.

karmic brook
#

Have a good day!

young ivy
#

Since I've been publicly called out as a troll and positioning me as someone that argues that people don't own their work, or that Im an agent of eroding rights. I have to publicly defend myself cause I'm not going to stand for being seen and called something I'm not. All I did was provide information to an issue that's currently taking place regarding the copyrights and ownership of AI created content. I didn't advocate for anything. At all. I certainly didn't argue people don't own their own work and that copyright shouldn't cover artists working with AI. I think to some degree it should, but it's a very complex web of legalities and it's going to be argued in courts and such for a long long time yet.

I'm not in favor of taking away copyright from artists using AI in their work. I never claimed to be. I'm also not spreading any negativity or intolerance and I'm not a troll. I've been here long enough that at least some people will know that I'm just another guy trying to come up with new ways to use AI in my workflow. I'm not here to attack, I'm not here to insult and I won't stand for being insulted either. That is all, I have no hard feelings but I needed to defend myself against that. I'm not going to continue this into any conversation beyond this message so @karmic brook don't worry about that

karmic brook
#

All good @young ivy
I think we're all just trying to get better in our own ways
It's easy to sometimes offend or come across extreme when online tone etc and mixing things this community is passionate about.

#

Thanks for stopping here and have a great day as well

iron pendant
hidden pond
#

has anybody tried InvokeAi?

royal pumice
#

If I want to create a lora for a face, should my source images be closeup face photos tagged as "closeup, face shot", etc?

#

Something you could use to tailor general photos to a particular face

raven bronze
#

How to train Models or Lora using Google Colab?

hushed flower
#

hello! i have a problem when i try to generate an img2img using open pose, the result is very distorted and straight up bad, but only when i try this openpose proces to do animation, wham am i missing?

prisma rivet
#

Yo

royal pumice
#

i was asking about the point in multiple controlnets the other day. I tinkered around and discovered using an openpose controlnet on top of a depth or canny controlnet really helps nail head position and hands

hushed flower
royal pumice
#

my controlnet supports four

hushed flower
#

oh yea, now i've found the option, i tried with canny on top of openpose but still a bit off, trying now with depth and hope i won't run out of vram (6gbvram)

hushed flower
#

ok i think i got some decent results in terms of face, takes a lot of playing around xd

hasty bane
#

someone who uses google colab, can you teach me how to download the ''output'' folder that contains the images produced with the SD model?

halcyon granite
#

is it normal for dreambooth concepts to bleed into each other.? They seem to merge in my model

wet oyster
#

might by a silly question, but when I run diffusion locally, is the url also private or can anyone that has the url able to use it too?

#

also does the url share or reveal any personal info? total noob here

broken smelt
#

the url is a local ip

#

everyone enters the same one into their browser

#

its like going into 192.168.0.1 for your network

frank spruce
#

It looks like an external IP but it's one of the ones reserved for internal use or something

#

the advantage of connecting the browser to the python via fake networking is it helps it be platform independent

lost pond
frank spruce
#

imagine asking the devs to write three separate gui applications for mac, windows, and linux, with divergent functionality for touchscreens and mouse keyboard, and expecting them to regularly update them

patent nebula
# wet oyster might by a silly question, but when I run diffusion locally, is the url also pri...

192.168.x.x are not internet routed. So some other person via the internet can't ask to be connected to that 192.168.x.x address and somehow reach you just by using that. It won't traverse the internet/WAN. If somehow you set up your own router though to route requests coming into your external (internet/WAN) IP to that specific LAN ip (like via DMZ) or just particularly incoming to that exact TCP port [:7860?] (via port forwarding) to your LAN ip, then there might become an issue theoretically. but I highly doubt that would happen. though honestly, it should default bind to 127.0.0.1 or ipv6 equivalent to make it even more harder to reach, and then force users to bind to the LAN ip if it needs to be reached within the LAN, but then some people who separate where the inferencing is running vs where the browser is running in the same LAN, might have an extra step or few.

#

can't wait for the fine tuned models to come out for SDXL. Of course that depends on SDXL being released 😛

steep rampart
obtuse night
#

probably been asked a million times, but does anybody have an idea of how much VRAM will be needed to run XL once the weights are released?

elder cradle
#

In fact, better than AUTOMATIC in terms of usability

#

But keep in mind that the default installation is very chunky, or at least, the last time I tried InvokeAI

hasty hornet
frank spruce
#

I just install openpose editor for first time and i cant live without it already xD

hollow kiln
#

any benchmark of titan v volta exists? that thing is like a ghost i can't find any iterations a second of it 🤣

hollow kiln
unique hamlet
#

Is the @radiant meadow bot planned to be a permanent feature of this discord? Or it’s just for testing

teal pagoda
#

Yo! Any idea about when SDXL will be released?

broken smelt
#

soon is all we know

teal pagoda
#

Someone was saying that it will be released this week

broken smelt
#

they must know a lot more than everyone else then lol

#

at best, emad is counting down for the end of this week on twitter for something

#

nothing else has been said with certainty

teal pagoda
#

I see

#

It's still not good at fingers and hands from what I tried

#

:)))))

sudden ruin
#

Thing is you can use xl as a base and train from there for better hands dogsmile

hollow crow
#

Anyone who is capable of creating different versions of animated/cartoonish faces of people/dogs/cats if given pictures of the subject. In the future I have consistent work load for creating these pictures. Searching for people who are motivated and consistent. DM me for more info

unique hamlet
#

How can I do a face swap maintaining the rest of the body intact in A11111?
My idea right now is to do a photo bash in Photoshop and then improve the result with low denoise, but I feel like there's a better way to achieve this

rough barn
unique hamlet
crystal patrol
#

Hey guys and gals, I'm using SD with Automatic1111 for the first time on Google colab. Does anyone know where i can find some testing image packs to train my model? Since I'm experimenting I'm looking for a fast solution. It can be anything but would prefer person images. Thank you!

hardy swift
#

What is your workflow when upscaling and providing more details to images?

inland plank
#

вы умные?

real kernel
#

I'm so hyped for XL release - I've heard that last time SD 2 got released Emad also counted down. Is that correct?

true locust
#

Hello, what is the best version of stable diffusion, 1.4, 1.5, 2.0 or 2.1, because I heard someone said the most of the stable diffusion arts was made by 1.4 but I don't know if that is true, and when I search comparaison on internet I don't really know

hasty hornet
hasty hornet
prime spindle
#

Hi ! I was wondering if there is a way with ai / stable diffusion, to merge Hdr pictures from the exact same shot, just different exposures. And an ai does the cleanup in a certain way and makes it look good? This would be from an image that I would provide to the ai. Mostly architectural photography would be used

true locust
hasty hornet
true locust
tall yew
#

XL release news ? heard it is releasing today 🙂

real kernel
#

So today or tomorrow

hasty bane
#

Guys, I create many prompts a day using chatgpt. But at one point the idea of putting the initial words hangs and it takes a long time for me to finish doing all the prompts I need. Any tips?

tawny needle
#

Hello - a question. I don't see any information on payment, subscription or paid service. Is the usage of these bots in discord completely free? How many images are we allowed to generate?
The reason I ask is, that I'll be giving a guest lecture on AI to kids next week, and I would like to offer them a chance to experiment with AI image generation. If there is a limit on the number of images I'm allowed to generate, I would like to avoid hitting that limit during the lecture.

earnest badge
#

From my understanding it's entirely FOSS

dawn mulch
#

but while the bot exists, you can generate infinite images for free with it

magic bough
#

can anyone help me fix a eroor

earnest atlas
#

any aprroximate release date for stable diffusion XL meanwhile ? alpha beta ...

real kernel
real kernel
#

are you sure?

finite cloak
tawny needle
#

Thank you very much for that information.

real kernel
#

Yes I think so! Switch over to #✨｜sdxl, there are more people talking about the release date

dawn mulch
#

for a 2d game?

kind vessel
#

I've been using stable diffusion for images and video. But I keep seeing Images that just look astonishingly real from midjourney. But SD seems to only give me these derpy, half way realistic images. Is midjourney just better, or do I just suck at knowing how to use stable diffusion properly??

hushed flower
#

guys this happens when i try to batch img2img for an animation

#

A tensor with all NaNs was produced in VAE. This could be because there's not enough precision to represent the picture. Try adding --no-half-vae commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.

#

what do

#

i tried to add these argmunets in the notepad from the stablediffusion.bat but it won't launch anymore

forest trout
#

Mainly prompt manipulation.

#

Also midjourney can't make ugly people.

kind vessel
forest trout
#

Yeah. Basically. You trade good looking images for much less control.

#

You can't do controlnet with midjourney, Deforum isn't possible, custom models and Lora's aren't a thing...

#

If you are wanting good looking images, SDXL will eventually release.

young ivy
#

I don't have a hard time making photorealistic images that don't look like Cronenbery went apeshit on them, you need the right models, the right vae, possibly the right lora, the right prompt and if you want to control pose, then you need controlnet

kind vessel
young ivy
#

Midjourney isn't so much better at it, it's better at giving you what you're prompting for. Or a facsimile to be more accurate. Cause it goes for the best possible language conversion for the prompt and toward the generation

#

yeah switched to safetensors a long time ago cause as the name implies, they're safer

#

you can't put any hidden code to be executed in them, at least not very easily

kind vessel
#

is there a way to make safetensor models work on the notebook?

forest trout
#

Here's something else to consider: midjourney is a one shot program. If you don't get it right the first time, too bad.

young ivy
#

Yeah it's incredibly hard to navigate your way towards what you want on a discord bot

kind vessel
#

Not discord bot like the google collab notebook for deforum

young ivy
#

I tried MJ for a month back in August/September, when they activated their first beta which accidentally made everything quite lewd for a short time

forest trout
young ivy
#

midjourney is just on discord, unless I've missed something

forest trout
#

Yep it is only on discord right now.

young ivy
#

I don't think they have any plans to move it away from there

#

they'll just call it a family friendly image generating community

kind vessel
#

I really need to reset my PC. I really want stable diffusion installed locally but it won't work because I switched to an AMD graphics card and the software is still detecting nvidia drivers even though i used display driver uninstaller twice

light crystal
kind vessel
light crystal
forest trout
#

@sudden ruin

light crystal
forest trout
#

@bleak matrix

young ivy
#

@kind vessel I'll show you something I made 6 months ago in SD with photorealism in the general-with-images channel.

kind vessel
#

can u stop spamming and shut uo

young ivy
#

just block him

fervent thunder
#

i swear theres no active mods here and if you try and contact them you cant smdh

forest trout
#

If I was still mod here he would have been banned already.

#

Yep. Too bad.

steady bridge
#

Wtf he is doing

young ivy
#

something dumb

steady bridge
#

Banned

#

Finally

#

I g

forest trout
#

Nothing of value was lost.

bleak matrix
#

Sigh

steady bridge
#

Can I ask questions regarding stable AI here?

forest trout
#

It's general chat so probably.

bleak matrix
#

So, what's up, everyone?

#

How's your day going?

forest trout
#

Aiyabot hasn't been updated for weeks and it's been broken for me. I haven't been able to generate images on my build for close to a month. Sigh.

steady bridge
#

So how do I make a character from other anime/manga wear joker make-up?

bleak matrix
#

RIP. I just got home from physical therapy myself.

young ivy
#

try that

bleak matrix
#

There's a lot of ways you could do it. Prompt it, use a certain model/lora. A combo of it. Inpainting.

forest trout
#

There's probably a joker Lora. I wouldn't be surprised.

young ivy
#

there is

bleak matrix
#

I would try prompting first-- just try a bunch of things til you find something you like.

#

Can always use the bot, too.

steep mortar
#

does anyone know the name of a tool that allows you to do all sorts of image editing via a grid. you drag and drop objects onto a grid and it has these wires you connect to the objects to cause it to do all sorts of image effects?

bleak matrix
#

You mean... like... comfyui or like image editing?

forest trout
#

There's that dragGan program but I've never tried it.

#

Oh wait that sounds like comfyui

steep mortar
forest trout
#

It's comfyui

bleak matrix
#

There's Invoke, too, though, to be fair, I haven't looked in ages

steep mortar
#

thnx, yea I think it was comfyui

#

yup this is it, thnx

bleak matrix
#

NP

sudden ruin
young ivy
#

oh god, I just read the worst Star Trek news...They're actually making a "Star Trek Academy" series...Anyone that watches TV nowadays knows how this well end

forest trout
#

My spammers at Harmonai post when I'm sleeping lol.

bleak matrix
sudden ruin
bleak matrix
#

It's only when you don't look, you know.

steep mortar
# bleak matrix NP

wait, this isnt it. it wasnt meant for stable diffusion, it was a stand alone image editor. but it has a grid just like this. I remember using it for upscaling

young ivy
sudden ruin
#

I liked discovery too PU_Sadge

#

Also Std sounds really wrong

forest trout
#

Star Trek Disease?

sudden ruin
#

Star trek discovery

bleak matrix
#

I don't want to think about this.

forest trout
#

I know. It's a joke.

sudden ruin
#

I like the old and the new ones the same

bleak matrix
steep mortar
young ivy
#

I loathe the new attitude they have towards Star Trek. "FRANCHISE IT!!!" and make a dozen shows devoid of the original spirit that Star Trek had. But I guess I'm not their demographic 😛

sudden ruin
#

I mean you have to adjust to modern times to survive

#

But it's fair, not everyone has to like it

steep mortar
#

Found it , its chaiNNer

#

woohoo

sudden ruin
#

Also og star trek was a canceled show if I'm not mistaken, not a wave I'd ride today kek

young ivy
#

it was cancelled because it abandoned the original Star Trek "feel"

bleak matrix
#

was it something more along the lines of krita /Photoshop?

young ivy
#

hence why Enterprise did so poorly

sudden ruin
#

I loved enterprise omegalul
Guess I'm just weird

young ivy
#

I guess so 😛

bleak matrix
#

I loved Enterprise, too

forest trout
#

I just can't get into star trek / star wars.

sudden ruin
#

I only got into star trek the past years

forest trout
#

Like philosophically I like star treks concepts but it's just not for me.

sudden ruin
#

But always liked scifi and star wars

#

I'm sure you like other stuff waow

bleak matrix
#

I watched it growing up, so

#

lol

young ivy
#

Starfleet is kind of hypocritical about borders and stuff though. If anyone violates Federation space, they're hands on. If they violate anyone elses space, they just say "we're explorers, we're gonna explore. You have no right to stop us"

#

then get into a conflict

bleak matrix
#

Yeah, I can't say there aren't plot holes

#

But I like a lot of the characters, energy, and character development

young ivy
#

DS9 does illustrate the issue with borders and the hypocracy of Starfleet (by intending to) in a lot of stories

bleak matrix
#

which actually makes me think, on a sidenote, about using AI to craft and create new genres

young ivy
#

I doubt you're gonna find new genres after humanity has spent 10,000 years crafting stories

#

at least ones we can "read"

bleak matrix
#

I think there is a lot of potential to create a lot of amazing new stories/ worlds / build better stories

young ivy
#

I don't think AI is going to do that though. You'll create a scenario similar to ep 1 of s6 of black mirror 😛

#

It would be awful

bleak matrix
#

Not just AI, though, lol

#

Just watch me do it, lol. I'm up for a challenge! 🙂

young ivy
#

An AI generating realistic content that you can't tell apart from reality and just parsing it is as tailor made content. Thousands of new shows a day

#

That would be horrible

sudden ruin
#

There are already Youtube channels that make their scripts with AI

young ivy
#

people would tire of it so fast, or worse, they would become so consumed by it they're zombies

sudden ruin
#

They also let it read by AI and done

bleak matrix
#

I mean more as a tool, part of the creative process, etc.

young ivy
#

yeah but those channels are obvious

sudden ruin
#

Right now yes, wait a few years

young ivy
#

yeah which is what I'm saying

sudden ruin
#

And even if obvious, people still watch it and generate income

bleak matrix
#

All of the creepypastas will now be accompanied by fantastic art.

steep mortar
#

do you guys have a fav background removal tool?

young ivy
#

deprived of all originality, every movie will have a different ending for each person just to make them happy and continue to be a customer. Every possible angle and plot device and so on will be created. It will become mechanical and stale

sudden ruin
bleak matrix
#

Or maybe they'll keep the black screen thing and the rain.

steep mortar
sudden ruin
# steep mortar lol

I like my stuff cleanly edited, no tool can achieve that currently from my knowledge

bleak matrix
#

Background removal? Hmm. TBH, I use selection tools/ Photoshop/ the eraser tool, masking, etc.

#

Color selection, and a wide variety of techniques helps.

steep mortar
#

hmmm ok

bleak matrix
#

If you're trying to do background removal, and you're doing it manually, with a mouse, you can use a program like Krita.

fervent thunder
#

depending on the image, then there are loras, and prompts that can remove a background

bleak matrix
#

You can prompt for solid backgrounds, etc.

young ivy
#

Here's a creepypasta for you, or an scp, you decide. The Knocker. It begins with a knocking on the wall, or the floor or the ceiling. They escalate and become louder and more violent as you struggle to find the source of the knocking. You tear up the floorboards and there's nothing there. Now it knocks from underneath the concrete below. Then it will knock on the front door. If you open up. The next time it knocks will be from inside the house and it won't stop knocking. It will wait until you're in the bathroom, or somewhere isolated. So there's no other way out. If you open now, you'll become trapped in an infinite corridor with white doors. They're all locked. So you begin to knock

steep mortar
bleak matrix
#

Well, I wouldn't tear apart my house, nor would I have the energy to do so. However, noises can be obnoxious, and that's why I'd like to talk about the sponsor of my comment-- RAID Canceller! That's right! with this unique, one of a kind product, you will never, ever have to hear your door bell again! Just shove these two delicately handcraft, airbrushed aitisan $2 lumps of hard tack in your ears til you can't hear anything and call it a day. If that doesn't work-- you're right!!! If you can't beat 'em, join them by starting YOUR own band of tintinitus inspired, yet heavenly acquired, set of drums and don't wear any noise cancelling headphones. Just shove your head right next to the high hat and you'll be fine.

drifting ocean
#

hi everyone, I am trying to get into stable diffusion. I am somewhat confused by things like invoke ai and automatic1111. where can i get more information on this?

bleak matrix
sudden ruin
bleak matrix
#

Ifyou need and help otherwise, please do feel free to ask. There's a lot of channels, so I suggest just checking around and you'll find a lot of g reat stuff.

fervent thunder
#

is it out or nah

bleak matrix
# steep mortar thnx, I have affinity photo

Awesome! Then I would just use masking/selection, etc. It really helps, when you're selecting/erasing to put a layer that is darker underneath, like a dark gray, to check for little edges /issues as you're working on your project.

fervent thunder
#

what model is the bot using

steep mortar
bleak matrix
#

For current information on that, please see the bot channels.

bleak matrix
wintry stream
karmic brook
raven bronze
#

Does anyone have millie bobby brown model?

drifting ocean
#

sorry guys am not sure where to ask this, I ran the webui-user bat file like an hour ago and it seems stuck at one of the steps. Am I good to close and run it again?(SOLVED)

wintry stream
young ivy
#

you could do ctrl-c in it if you want to be extra sure @drifting ocean

fervent thunder
#

hey there, I haven't followed news from art AI for a while, is there a reason why there's no new version since the 2.1? maybe an article/video that could explain it in detail?

wintry stream
fervent thunder
fierce fox
#

Hey, I’m sharing a new way to experiment & combine generative AI models - kind of like a Jupyter notebook specifically for generative AI. You can use ChatGPT to generate Stable Diffusion prompts and then produce Stable Diffusion images all in one place.

AI Workbook: https://lastmileai.dev/workbooks/clj2y933l000mr0avd2ck42s9

unborn perch
#

Hey guys, do we have website where you can check images generated along the prompts?

broken smelt
#

If it's just the prompts you need, I read that you could just open the PNG image with notepad and it should have the prompt there

frank spruce
#

My browser freezes when I input the directory text for batch img2img

#

like all the dropdowns and controls stop working

#

and i have to close the tab

#

is this a bug?

unborn perch
ocean locust
#

I'm looking to create a "filter" or prompt/workflow that could upscale existing images and give them a "palette knife" painted look. Anyone have a resource or tutorial for something similar that I could modify?

broken smelt
#

The bot for sdxl in this discord should also have pictures with prompts

unborn perch
sharp lynx
#

hi, excuse me

#

im having problem with realistic images

#

while image is generating looks good but final, looks like it has a filter

raw dagger
sharp lynx
#

Was a vae problem, now is working perfect

unique hamlet
#

Is there a waitlist for early access to SDXL? (not 0.9)

hushed quarry
true sequoia
#

hi

thick dragon
#

Can i try sdxl 0.9 here in discord?

hushed quarry
thick dragon
#

Thanks. but how do I invoke this new version in my prompt?

opaque coral
#

hi

drifting ocean
#

random question, how long does an image take to generate on your device?

sudden ruin
drifting ocean
sudden ruin
#

I can do a batch of 8 images Euler A 20 Steps in 10-11s

#

512x512

true sequoia
#

hi

kindred inlet
#

Hi

true sequoia
#

good u

upper minnow
#

I didnt save an image and now i cant find it

#

I searched for it

true sequoia
#

nothing much

upper minnow
#

From @radiant meadow Mentions @upper minnow

#

I cant find anything

#

Do the images get deleted

true sequoia
#

hi @upper minnow

upper minnow
#

Ello Stranger

true sequoia
#

hi

upper minnow
#

meh

hushed quarry
#

Are you just here to try and rile people up or are you planning on actually chatting or contributing? (Well that person is gone now berk)

upper minnow
#

Tis a Sad day

hushed quarry
# upper minnow ACTUALLY I LOST 2 REALLY GOOD IMAGES

Oh no! Well the bot should be sticking around for awhile, you can totally keep trying to get more good ones! Still sorry bout the others, the images only stick around a few hours unless they win the pantheon (ofc you can copy or dm them to yourself for keeping)

upper minnow
#

oh no you’re a mod

hushed quarry
#

Hah, technically not a mod but I do work here (and ofc have the mighty ban hammer for mean people like that last dude)

bleak matrix
#

Good morning, everyone!

hushed quarry
#

Mornin sunny!

bleak matrix
#

How are youuu?

hushed quarry
#

Good, technically its my goodnight haha

#

I did just notice how close these role colors are..

#

Like a little hotter pink maybe

bleak matrix
#

Night, day...close enough

#

Yes, yes, we're now closer to Bubblegum fever, dude! 👏

#

I see something always happens when I am not paying attention lmao. But I am glad you are good!

hushed quarry
#

Nah you didnt miss anything 🙂 yea hope you are good as well!

bleak matrix
#

Not anyone important, ahahahaha! Pretty good! Just working on some textures---couldn't sleep, so I gave up on that idea.

sudden ruin
#

Very pink in here Susge

hushed quarry
#

It did get very pink all of a sudden

#

Hahah

feral finch
#

how can i get lincense ?

#

hi! How can i get a licença

bleak matrix
#

What do you mean?

feral finch
#

The use for stable difusion how can i pau

#

pau

#

pay

bleak matrix
feral finch
#

thank vou

bleak matrix
#

You can also go to one of the bot channels, check the pinned comments to learn more information, and test the bot for yourself there for free 🙂

warm junco
#

Or you can use Dreamstudio

feral finch
#

is free?

warm junco
#

The Bot here is free, and if you install it localy too, but Dreamstudio is payd

feral finch
#

the dresmstudio?

#

dresmstudio

bleak matrix
bright sand
#

heyoo

#

anyone knows how can I make a gif that transition smoothly from a real life photo into an AI photo I made based on the same photo that is cartoonized?

#

I found one website that makes a gif like that but the quality of colors gets really bad

#

also I would prefer to do it locally if anyones knows a way

#

oh it seems like a simple editing in sony vegas does what I wanted

#

now I see the ai pic is not wide enough but I already done all the post editing in photoshop so I'm not going back there this time xD

#

lesson learned for next time

sly willow
#

guys how are you making these cool AI animations?

bleak matrix
unique hamlet
plucky shoal
#

i believe put it in embeddings.

unique hamlet
#

what's the equivalent of PNGinfo in invokeAI?

violet sedge
haughty saddle
#

is there a way to know the lora/lycoris trigger words without opening the civit.ai every single time?

broken smelt
#

it automatically keeps track of the triggers

#

no need to manually input them

haughty saddle
peak portal
iron wren
#

Tuna?

karmic brook
#

shrimp

cerulean knoll
#

Which model is best to create anime messey rooms like those available on pinterest, if anyone know pls tell me

#

Messy***

forest lily
#

I eated the stable diffusion

#

Yumy :)

foggy spear
#

hello everyone

sudden ruin
visual oxide
#

Is there any plans to add a seamless tile feature?

outer crescent
#

because you can make repeatabile tiles in a1111

thorny root
#

I am having trouble keeping the same character between images(also, there like, shape...they be melting randomly?)

visual oxide
outer crescent
#

oh its something that you run locally. or possibly on colab. theres a lot of tutorials on youtube, its too in depth to talk you through here 🙂

plucky shoal
#

for sure best use official because easier.
@visual oxide what is your gpu?

visual oxide
loud turtle
#

Hello. I'm new to SD an just got it put on my machine. My goal is to make fantasy art work for a book cover i am working on. I was wondering if there were any models or add ons that are recommended for this? Stylistic I was thinking that fine fantasy art, like you'd see on the side of vans from the 80's, if you know what i mean.

broken smelt
#

head to civit.ai and browse around. we could all throw names at you but since its art, only your judgment on what looks good or not matters

loud turtle
#

Thank you. I appreciate the pointer. There is just SO much to all this its overwhelming. Even the 'start here' make me swoon a bit.

broken smelt
#

best learning process imo is to just load it up and generate a few pictures to see what each slider does and get a feel for the knobs you can turn

loud turtle
#

I been doing that, just havent been able to get what i want.

#

My poor laptop wants to melt. heh.

warm junco
outer crescent
loud turtle
#

THAT is the one i was looking at.

#

I'm trying to figure it out. Hell I don't even know how to install or add it yet.

outer crescent
#

someone was using it here, lemme check

#

quite a few people actually. Anyway, it doesnt work like normal models as i say. personally i like the v2 model. but check each out.

#

for the van mural style you could maybe try 1980's airbrush art? not sure

#

"airbrush fantasy art" seems to work well on search engines so even if zoya's awesome model doesnt get it right first time, you can always move it over into another model as a low-mid denoise image to image. or maybe the inpainting versions of their model

loud turtle
#

Should I be using something other than Easy Diffusion?

frank river
#

Automatic1111 or ComfyUI or InvokeAI

loud turtle
#

Uhm,,,,

visual oxide
loud turtle
#

I'm trying automatic. It seems to be installing.

#

I installed it in a dumb location. is this going to be happy if I just drag the thing elsewhere after install?

ashen estuary
#

Hi there, new here

#

I just installed stable diffusion locally, and ran a few hundred images. Is there any guides I should read to get the most out of this?

primal current
#

where are the vae.safetensors files stored? when I saved them in VAE it generated an error, I saved them in models and it was solved, but logically I can't use them as VAE

rancid basalt
#

guys can i download the latest version of stable diffuision on easy diffusion and if i can where can i download it from ?

opal path
opal path
outer crescent
#

you're essentially taking the image to layers but instead of making them flat scrolling/moving like scooby down running down a corridor its on a bit of a z axis

#

if you're looking it up, "2.5D" is the term the cool kids are usin @opal path

#

come to think of it, lama inpaint or segment everything would be amazing for this kind of stuff if you run a1111 or comfy

plucky shoal
#

paralax can be made well in opentoonz. There is more tutorials.
opentoonz is free and handy animating software. So complex that there is probably not person on earth know everything it can make 🙂

hushed flower
#

ey, can i use like 2 multiple controlnet for an openpose+ a reference on img2img to change completely how the animation looks but without fucking up the pose? messing around a bit and im not sure it works, has anyone got it to work somehow?

broken smelt
#

this vid was pretty popular a few months back

outer crescent
#

oh nice, i was trying to make my own anaglyphs the other week

#

this repo hasnt been really touched in 5 months tho

broken smelt
#

for general use purposes, i found it just fine without needing to create any PRs

outer crescent
#

maybe they just did all they needed to do with the repo, if it aint broke etc

#

there was one of the geneology websites did this with old photos

#

might be a cool thing to do for family if there are any old black and white old timey photos in the house

#

anyway, @opal path rootbeer's link is probably a better one to play with

hardy swift
#

does img2img, interrogate CLIP take a long time for you guys?

desert yoke
#

Hello everyone!!

twilit osprey
#

Hiya 👋

tough vine
#

Hello people, do you know how to increase the size of an image when generating it?

steep star
tough vine
#

someone had made the image large when generating the image

plucky shoal
#

@hardy swift first time take very long

open night
#

training my first test model woaaah

#

seems the eta settled around 50 minutes cool

#

really curious how it'll turn out seeing as I have no clue what i'm doing lol

sudden ruin
#

My first model was over trained and pretty much generated the same thing every time agony

open night
#

lol

#

I settled on the standard settings so i have no clue what to expect

#

I only included like 17 reference pics so..

#

copium it'll be fine

wind kelp
open night
#

well

#

idk how fast it is

wind kelp
#

tyy

hushed quarry
wind kelp
#

ty!!

main kite
#

Hi all, as some of you might have seen, Epic has released Metahuman Animator which has a great feature that is using the iPhones depth sensor and a single photo together to generate a metahuman face. I was wondering if it’s somehow possible to inject an AI generated image and one of the Controlnet depth preprocessors depth map and compile it all together to a format that Metahuman Animator in unreal engine 5 would take. I’ve seen on this Apple developer page that there is an API for it but I’m not a programmer so it’s hard for me to tell. https://developer.apple.com/documentation/avfoundation/additional_data_capture/capturing_photos_with_depth

#

This could be way to turn our stable diffusion characters into fully rigged 3D characters with zero hassle and manual labour.

open night
#

welp that couldn't have gone worse

#

NansException: A tensor with all NaNs was produced in Unet

thorn stirrup
#

heyo question how do i set an aspect ratio

quasi sky
lime herald
#

Однажды по дороге в школу мне захотелось срать. Переборов в себе порывы, я решил добежать до школьного толкана и там извергнуть свое говно. Но каков был мой ужас, когда я обнаружил, что толкан не работает! А срать хотелось все больше.

fading wraith
#

hey guys. have any new models / checkpoints been released since april that are now considered state of the art for photorealistic images? and is automatic1111 automatically updating, have there been updates, and if yes do i need to install it newly from github?

plucky shoal
#

git pull should be enough, if you have git. In your stable diffusion folder, as when you installed it. right click, git bash -> git pull

finite cloak
tender nebula
#

@humble elk hey, could you accept my friend request? thanks

violet sedge
#

Someone know a website that good with creating bedrooms?

floral geode
#

Guys I have installed automatic1111 buy extentions loading showing error

#

Is there anyone getting that?

hasty hornet
#

Probably extention isn't up to date with a1111 version

shut karma
#

bot doesn't work

bleak matrix
#

Good morning, everyone! How are we all this morning?

bleak matrix
thorn pilot
bleak matrix
#

I was thinking of doing something beach related.

thorn pilot
hushed flower
#

is there an option to save previous instances of stable diffuison so i wouldnt have to do all the sliders/prompt again?

#

when i reopen it

sudden ruin
bleak matrix
#

I love those sculpted ice cream things

hushed flower
#

but i cannot find it anywhere anymore...

sudden ruin
#

Really? How have i never found that hmmm

hushed flower
#

yea, i followed some guy's video on some settings, and checked every setting he had, and then my stable diffusion was like that, but he also had some settings i didn't like so i have no ideea what setting to look for, there is nothign similar on that guy's settings xd

#

and i don

#

and i don't want to check every setting and then look for what to uncheck

#

thought you guys knnow something about it

bleak matrix
#

Do you mean like saving the model in RAM?

sudden ruin
#

No like saving all the settings when closing

bleak matrix
#

You mean default settings

hushed flower
#

let's say, you already have a propmt+ whatever control net you have active and whatever sliders you have set up, if you close stable diffusion and then reopen it it will be reset and you have to set everyhtign again

#

but there was a setting where every option you have before closing it (controlnet sliders, CFG, Width-Height, and everything tbh) will save adn remain the same after you open it

reef epoch
#

is there any info about the stability VideoLDM method? similar to nvidias? like could we implement it into any SD txt2img model we would like?

#

@wise stratus 🧐

bleak matrix
#

Did they have an extension installed?

#

Hey Maki! Sup?

reef epoch
#

wassup sunny

bleak matrix
#

I was just thinking about you yesterday

hushed flower
#

don't think so, it was just a setting, i had that setting before but idk what it was since i blindly copied that guy's settings

reef epoch
#

prove it

bleak matrix
#

I did buy that sandwich

#

Not as big as I wanted but it was good

reef epoch
#

damn you still didnt buy it

#

its been awhile

bleak matrix
#

Just chilling and doing...this...bone therapy thing. You?

#

I DID AHAHAHAHA

reef epoch
#

wanting to figure out this VideoLDM shit

bleak matrix
#

I still have the photo, I think

#

Deforum, yo

reef epoch
#

i WANT txt2img model into a txt2video model

#

yeah yeah old shit

#

im bored of deforum

#

i want MORE

bright sand
#

any tips on how to remove something in the background of an image with inpaint? I want to change it to fit with the rest of the background which is sky and clouds at dawn and can't pull it off

#

it keeps using the unwanted thing beneath the inpainting and generating another unwanted thing

bleak matrix
#

They're always adding a lot of support for stuff

bright sand
#

oh I just got it while typing this with latent noise with random seed

#

I always kept the seed coz I thought it will keep the style similar

#

oh well 😮

bleak matrix
#

You can use a custom mask

#

Oh, well, lol

#

There you go

bleak matrix
bright sand
#

sometimes when I use inpaint and remove the inpaint to use it on another part of the image, it generates inpaint on the current and the deleted inpaint.
its a glitch obviously, any fix?

bleak matrix
#

Refresh (your browser)

plucky shoal
#

are you using inpainting models? @bright sand

bright sand
plucky shoal
#

there are special for inpainting. Containing inpaint in their name. Not sure about anime, but wait a mmnt.

#

deliberate2, realistic vision v20, dreamshaper, neverending dream. All those i have with inpainting models.

bright sand
#

can I use inpaint_only+lama methord in img2img after I edited the image in photoshop?

#

it doesnt seem to outpaint the image properly

plucky shoal
#

there is nice extension for inpainting and outpainting called openOutpaint.
I dont know whats lama method is 🙂 try what you want, at the end picture will be well finished.

hushed flower
#

so, im trying to doan img2img animation and everything is PERFECT but the fucking hands.....anyone got some tips and tricks on how to fis them noodle fingers?

quasi sky
#

Currently trying to learn more about finetuning to hopefully help work on it when SDXL open is released 😄

sly eagle
#

Hello, anybody as used Runpod? I want to use it but I don't know if it worth the money

willow coyote
#

What up what up

#

Cool

rustic sentinel
#

Where to ask on how to get the image I want using SD?

broken smelt
#

Worst case scenario you can always bash in some good hands from another picture and edit it in and if needed, img2img to blend them in

sinful talon
#

Hey for thoses that use Comfy UI, what can you do that you are unable to do in A1111 ? can you use extension in comfy or not at all ?

finite cloak
#

the main direct benefit is advanced users can quickly swap around and experiment with whole new tooling

#

vs having to wait for somebody to code it into the python and PR it to auto webui

hushed flower
sinful talon
# finite cloak comfy has extensions, yes

thanks for the awnser ! do you know how much space doest it take on drive and if i can use the models that are in my A1111 models folder or do i have to copy past them in comfy ui ?

buoyant panther
sinful talon
floral geode
#

Guys anyone with the automatic1111 webui?

#

Can anybody help?

#

I am having some problem with loading extentions

bright sand
#

mmm after using ultimate upscaler, it leaves tiles marks. any idea how to fix this?

buoyant panther
# sinful talon Thanks :D

oh, to clarify, you'd write a .bat file that looks something like this, edit the path to wherever your webui is located:

C:\stable-diffusion\a1-sd-webui\venv\Scripts && ^
python main.py
pause
willow coyote
#

yo

#

lets redesign spacex' starship, yall down?

#

one sec, going to #laion to focus on the engine

#

dont sleep! it's gonna look a lot better than starship

bright sand
#

how do i check my lora's info? like png info for images?

#

(made with kohya ss)

stable pawn
#

Random idea for any programmer; extension that read the prompt and after that select model that been downloaded after model description, maybe also suggest models.

neon mural
#

what is the best model for outpainting using an imported photo?

analog granite
#

Hello, i was wondering if i can run stable diff with AUTOMATIC1111 on a rx 5500

#

locally

#

6gb vram

pearl fog
#

Automatic is kinda iffy with AMD cards.

analog granite
#

any other suggestions?

broken smelt
sand adder
#

how do i update stable diffusion?

analog granite
#

git pull

#

enter that line in the webui-user.bat

#

obv only if u got it through git

#

on linux do basically the same

kind raft
#

Up until today I was able to generate fast images and animations in SD, but today I changed my driver from game ready to studio driver and what used to take about 15 minutes to do an animation is estimating 2+ hours.

The only thing I changed recently was the Nvidia Game Ready driver to Studio Driver.

Could this be the issue? I can switch back to the other driver but I just wondered if anyone had run into a similar issue. Thank you!

#

I think I found the solution after reading through others' similar experience. I think I need to install the CUDA driver package for my GPU which I am doing now. Hopefully this works. crosses fingers

spice plank
#

Is there a way I can get info on how a checkpoint was merged/trained? I I think they added a lora to it but I am unsure and I'd sort of like to take it apart to get more information

high lichen
#

Has anyone found the best prompts for fingers?

floral geode
#

has anyone able to load the available extentions?

unkempt hatch
#

why can't we merge negative embeds into a model?

sudden ruin
sudden ruin
sudden ruin
sudden ruin
sudden ruin
floral geode
#

No arguments added

sudden ruin
#

hmmm thats weird for sure

warm junco
#

Or changing DNS at first

sudden ruin
#

Did it work before reinstalling?

heady flax
#

Is the model available yet for use in stable diffusion?

timid bloom
#

em, clipdrop (pro) have limits on sd 0.9 image usage? 1500 a day?

#

because i getting Thank you for using ClipDrop

You reached the maximum Pro quota.

#

😂

sharp burrow
#

hey, trying to place a living healthy man in front of a zombie apocalypse. But he keeps getting zombie features as well. Do you know of any way of making the ai to undertand that he is not part of the zombie-apocalypse?

high lichen
sudden ruin
sudden ruin
spark hill
#

Is SDXL 0.9 currently being the model running through this server's bots or is this server using an earlier model for image generations?

hushed quarry
peak portal
fervent thunder
#

hey guys

#

I'm trying to make a discord bot like the stable dreamer but I can't understand how to make it as it's my first time(don't get this in a wrong way)

#

I've made a multipurpose bot before

finite cloak
#

StableDreamer here is overcomplicated cause we need it to handle a trillion requests a minute and have the whole fancy showdown stuff, but making your own smaller image gen bot isn't too hard if you have experience with making discord bots

peak portal
# fervent thunder I'm trying to make a discord bot like the stable dreamer but I can't understand ...

i have two discord SD bot projects on github you can learn from, that do everything on their own, without any external APIs that mcmonkey suggests you use.

https://github.com/bghira/discord-tron-master < this one is very complex, it has a master/slave worker network architecture, and lets you distribute jobs over many systems.

https://github.com/bghira/diffusion-python-bot < this one is very simple, it is more just a proof of concept and can help you a lot. though it doesn't use the best layout for a Discord bot, so, anything you pick up there i'd definitely take a moment to find out if there's a better way. namely the way the Cogs are organised is problematic.

peak portal
#

i imagine a few of the issues it's faced could have been resolved through a community pull request, even

fervent thunder
#

ohk tysm for the advice!

finite cloak
#

not right now, need to change up how it handles the backend of things before it can be shared publicly

peak portal
turbid kraken
#

Dear all, can you give me a recommendation how to start?

  1. Which UI ist the best and have most options? I do not want to work on command line level.
  2. which stable diffusion model is the best? 2.1?
#

Thanks in advance!

stable pawn
#

I seen so many recommend "Euler-a" over "Euler", when I test it look like Euler-a (and dpm2 a) give different looking result than than most other sampling methods, self do I often lean towards DPM++ 2M Karras and DPM++ SDE Karras for variation, mostly for I see it as they give a tad cleaner images, so can some tell me what they see as Euler-a strong side?

trail dew
#

is there a way to save prompts? cause i dont really like typing the negative prompts every time i load sd up

tulip yoke
#

what was the lyco model folder called

#

i dont have it but i need it

stable pawn
#

If you use A1111 there is a small icon looking like a old disk under the Generate button, enter the negative prompt you often use, klick it and save @trail dew

trail dew
stable pawn
#

The "Save" symbol is a disk from the ancient times, like a USB is today but could only store 1/2 JPG file

trail dew
merry solar
#

Hello, SD community! I’ve been closely following the developments of both MidJourney and Stable Diffusion in the image generation space. I think its safe to say SD produces images of comparable image quality and resolution, and capable of even higher resolution images with tons of detail, I’ve noticed that MidJourney still seems to have a certain quality that sets it apart. I find it difficult to pinpoint exactly what that is—maybe it has to do with MJs methods of layering or some ability to express inventive and poetic concepts. I’d love to hear your insights on this. What do you think contributes to the unique qualities of MidJourney’s image generation? How does it differ from Stable Diffusion in terms of the underlying techniques or approaches? Any perspectives or thoughts would be greatly appreciated!

tulip yoke
#

what

stable pawn
#

I an a bit dyslectic so I can have misunderstood @merry solar question, but I think MidJourney create high quality images, but it have a style that is a bit boring and the term "MidJourney style" or "quality" is often used to describe that style. I think MJ is trained on curated and selected digital art so there is not the big problem with getting "copyright" text in the images, and I think the original StsbleDiffusion just looted all stock sites and then automated the process so we get a higher chance to get "bad" quality, but we can compensate that with other models, Loras and negative prompts and so on, and now even ControlNet.

hasty hornet
hasty hornet
broken smelt
stable pawn
hasty hornet
broken smelt
#

Yeah purposes are also a factor to keep in mind

#

Some people like generating thousands of 10 step pictures and some samplers are faster and create better looking pictures for that than others

zenith citrus
#

hello is there any way to try SDXL ?

shut hare
zenith citrus
#

well wrong question sorry. Is there any way offline?

sudden ruin
#

Not yet

storm arrow
#

Can I set a custom seed when i use the dream cmd?

merry solar
wise stratus
reef epoch
#

based on that math, it means.... soontm

wise stratus
#

its done when its done

#

tbh felt a bit uncertain about saying mid july for SDXL 1.0 but team working hard

#

I think once released just keep releasing and improving but first model is always hard

wind kelp
turbid kraken
reef epoch
wise stratus
#

it seems a reasonable timeline unlike distilled which kept breaking

#

its been running well on the bot and now clipdrop, will get more data from api etc tmrw

fervent thunder
#

model looks incredible to me (xl)

reef epoch
#

the distilled model is what ive been waiting for my mac

#

we need speed, reeeal speed

wind kelp
spice plank
#

I saw that 8gb was the minimum for sdxl, is there a chance we will get a lowvram or other options to run it on 6gb or a pruned checkpoint?

#

Or can we run it at 512x512 without screwing up the images?

finite cloak
#

on a 2070 (8GiB VRAM) + ComfyUI in default mode, SDXL took 6.5 GiB of VRAM and ran for ~20 seconds to produce a 1024 image.

#

I sure hope lowvram settings could trim off that last 0.5 lol

fervent thunder
#

can't wait for this bad boy to release

turbid kraken
spice plank
#

Thats similar to what I am getting in sd1.5 with medvram. I am getting 5.8gb at 1024x1024

hasty hornet
#

waow rip my 1050ti

spice plank
#

There is a defunct checkpoint model I like. The author is lost in the ether, is there a way to essentially update the checkpoint for sdxl? I was thinking someone could technically generate a bunch of pictures at 1024x1024, inpaint, clean, etc and update it that way?

#

Sorry if thats a dumb question, I dont know much about checkpoints

rough citrus
#

Who do I have to bribe around to get access to SDXL 0.9? I want to try fine-tune it.

waxen storm
#

open Source Releasing in July...its on Clipdrop now and Dreams tomorrow

static badge
#

hi a quick question
how to multi prompt? i cant find wildcards extension anymore
i need to multi prompt so i can make a lot of characters autoamtically

wispy jasper
#

So what ever happened to DeelFloyed, that project basically DOA?

livid cloak
#

I know posting NSFW is against rules, but may I ask someting about the program itself and NSFW?

spice plank
broken smelt
#

ew cooties talk, im closing my eyes

livid cloak
karmic brook
#

If you think you shouldn't talk abt something and have to ask an opinion of if you should ask or not- maaaaaaybe not haha

livid cloak
#

Good point

finite cloak
wind kelp
#

I've considered using cleaned up nice ai generated images to add training data tbh, but I wouldn't w/o a good non-ai base dataset

atomic atlas
#

Hi I would like to get into investing into some hardware for SD fun. I am primarily a macOS user and would like to hear advice from those that might have setup your own SD server, whether it’s on a Mac Studio or PC or via the cloud.

wind kelp
#

can sd even run on mac?

atomic atlas
#

Yes of course it can.

wind kelp
#

huh, neat

atomic atlas
#

Ideally I like to have my server running somewhere at home and I interact with it using screen sharing or via a UI I can access from the cloud. And ideally macOS lol just because I am familiar with writing automations and scripts / iCloud / etc etc

#

Doesn’t seem ideal at this point however…

frosty mist
#

anyone know a good midel to try with stable difusion for degining homes and cars and thier insides

thorn pilot
atomic atlas
thorn pilot
atomic atlas
#

Looks like this discord isn’t as active at this time.

thorn pilot
ionic forum
#

does anyone have a link to stability ai sponsorship program?
where they sponsor gpu compute. Can't seem to find it anywhere

bitter vector
#

how is doing everyone

pliant anchor
#

Hi 👋🏻. Has anyone got to use SDXL 0.9 api yet? Thanks!

sudden ruin
stray tangle
#

Can the bot be added to my server? Please and thanks

wind kelp
#

I've been wondering that but I'm pretty sure it's a no atm from what I've seen

finite cloak
#

we might do an open source/public version of the bot in the future, but not right now

wind kelp
#

OOO... a public version of the bot in the future..

#

I'd adore that someday, I have a friend I wanna show the bot that doesn't like being in public servers

pliant anchor
sudden ruin
#

4AM in the US

little dagger
#

hi

#

im new to stable diff and i was wondering if it could generate specific clothes on specific characters?

mint gulch
#

Hello, i faced a problem, My engine is stable-diffusion-xl-beta-v2-2-2， when I call the api with this prompt: This is an indoor scene picture with a large glass floor-to-ceiling window in the background, which shows high-rise buildings outside. The center of the picture is a mother and her child playing with Archimedes blocks on a crawling mat. The mother is sitting cross-legged on the mat, applauding her child with both hands, while the child is standing and placing the blocks in his hands. Please try to show as many details as possible and make the scene look vivid and interesting.
I got API Error: Invalid prompts detected, and when i use the same prompt in discord bot of this channel, it successfully created, and I want to know what's the invalid prompt in my prompt and why their have different behavior.agony

smoky thistle
mint gulch
#

OMG, do you recurrence this error with same prompt? API Error: Invalid prompts detected

smoky thistle
mint gulch
faint vine
#

Hi, I'm using nmkd and would like to convert a ckpt to onnx. When I do, it creates several folders, shouldn't I create a single file?

remote urchin
#

Hey guys, can somebody explain the application difference between a Lora and a Lycoris?

#

I mean, they both go into the >lora> folder

#

and even though I have the extension tabs for lycoris, they only show up in lora tab in A111

#

Unless I've been doing it wrong this entire time

raven bronze
#

How to write a prompt like 3 girls standing together based on embedding (Textual Inversion) i have.

Let say, Person1, Person2, Person3 are trigger word of Embedding.

I am writing like,

3girls, photo of Person1, Photo of Person2, Photo of Person3, standing, looking at viewer.

But random 3 girls photo showing or anyone one face from one embedding showing.

How to fix this?

thorn pilot
raven bronze
#

So, it not possible to do with Prompt alone?

thorn pilot
late blaze
#

I haven't really been following AI for the last 6 months. Is NovelAI keeping up with the progress/worth it?

hardy pollen
long flame
#

hi im new to discord but would love to talk about ai speech and chat algrithems

#

hello

plucky shoal
waxen storm
little dagger
turbid trench
#

it's 26 June and can't see it on Dreamstudio

pearl fog
stray tangle
sudden ruin
pallid flame
#

hi

#

can anyone explain DPM adaptive sampler to me ? is there anything noteable about it ?

karmic furnace
#

has anyone seen a discord bot or website that's good for img2img for interior design? I've tried homedesigns.ai and zmo.ai, and they are terrible (like, can't recolor a fireplace).

rigid harbor
#

wtf is wrong with dreamstudio...i've been trying to log in but it keeps giving error saying i cant log in. anyone else have this issue?

urban dome
#

@wise stratus any update on when the new SdXL will be released on the REST API?

karmic brook
wise stratus
#

doing final tests now cross your fingers

novel crag
#

Can you guys release a previous SDXL checkpoint ahead of time for testing compatibility? Even if it only generates noise, I want to be ready when it becomes available agony

light folio
#

Did someone record the whole conversation? I ended up getting only half of it... I was going to transcribe it on Whisper and post it on Reddit for people who didn't participate to understand everything that was said here.

wise stratus
#

thanks to @regal raven & team for a great stage, hopefully lots more soon

regal raven
rigid bough
rigid bough
tough vine
#

I didn't get to see the conversation, let me know when you have the video to watch it.

velvet fossil
#

this place used to be a help, but all I see it as it is now is just a burden

#

goodbye

bleak matrix
mint lily
#

what is a comfyui vae?

neon oriole
#

what's the context? Are you talking about the node in ComfyUI used to use the VAE?

mint lily
#

@neon oriole oh hallo! 🙂 In the installation instructions it says to put my vae in .... I've not been too much into SD for some months and even lora was new to me when I returned, so I'm not sure what the autoencoder is either..

neon oriole
#

stable diffusion generates "compressed" images called latents, the VAE is used to decode those latents into regular images

mint lily
#

ohh, I didn't know that. thought there was just some checkpoint thing and it handled it all. so our vae will bottleneck latent down then back up to RGB space? intressin'

hallow flare
#

can someone explain how negative embeddings work n how they influence ur generations?

urban dome
neon oriole
#

that's what the 3 outputs for the checkpoint loader in comfy are

#

but the unet is called "model" because it contains a bit of other stuff too

wise stratus
random crystal
#

I'm getting error code 1 and it's saying can't find pip eternal can somebody help?

hollow crow
#

Anyone who is capable of creating different versions of animated/cartoonish faces of people/dogs/cats if given pictures of the subject. In the future I have consistent work load for creating these pictures. Searching for people who are motivated and consistent. DM me for more info

hazy solstice
#

Pushing my code to GitHub takes forever and sucks my data in mb/secs. Am I doing something wrong?

fast lichen
#

I'm looking to run stable diffusion locally and I'm curious what kind of speeds I see using a GPU or just a CPU. Is it like night and day? I have an AMD 5900X with 64GBs of RAM. I also have an AMD Vega 56 cards (I think with 4GB of VRAM). Let's just say I'm making a 1024x1024 images, would a pass take minutes on this CPU compared to seconds on the GPU? Would there be potential image fidelity losses too?

#

And then another question, what would be the comparison jump between a Vega 56 and an Nvidia 3080? Massively improved speeds?

ionic galleon
#

What model version are the bots1-10 using?

neon oriole
#

13700k sampling 1024x1024 on SD1.5 is ~20 seconds per iteration

#

3090 TI is about 4.7 iterations per second so about 100 times faster

fast lichen
neon oriole
#

yeah CPU will actually be slightly higher quality since it samples in fp32 because regular CPUs have no fp16 instructions

obsidian meadow
#

Hi all, anyone know why automatic 1111, is Installing requirements, over and over again, and then a lot of:
Requirement already satisfied:...... ? I have updated all extension, but still doing it, also ControlNet is automatically enabling units?

mint lily
#

@neon oriole where do you put t2i adapters and controlnets in models/ ? (or, if it's just in (lowercase) controlnet/ how do I get comfy to reload, since the .pth is not listing in the controlnet loader)

#

it's my first time working with comfy.. beautiful project.

#

(I'm running in a docker build)

verbal osprey
#

Hi Everyone, quick question, where do lycoris models go? Do they go in the "/models/lora" folder in automatic1111?

mint lily
neon oriole
#

reloading the model file lists in the ui is what the refresh button does

mint lily
#

@neon oriole thanks so much. 🙂 (I am assuming you're the author)

neon oriole
#

I'm the main dev

mint lily
#

@neon oriole beautiful! it's working. thank you so much.

#

I wonder if I can get chatgpt to allow ^V to paste an image from the clipboard into the image loader

obsidian meadow
#

After removing it from the extensions folder, the problem has gone

median agate
#

So I just recently subscribed to Rundiffusion and had zero luck with obtianing high quality photos on the platform. But, here, they are near perfect!

wind kelp
#

I love rundiffusion so much

#

but

#

it's running on 1.5 and 2.1 rather than sdxl, so the quality'll be different

young ivy
#

I don't get why bloody or gore stuff isn't allowed anywhere...Even where hardcore nudity is

uneven fog
#

Hello, I wanted to ask if anybody can give me a tip on how to make the background Solid white. It keeps turning grey and I could use a help tip please

#

Im using text to IMG

remote urchin
#

im dumb, how do I disable a specific extentsion in Automatic1111?

#

Do I simply uncheck it?

broken smelt
#

ya uncheck and reload ui

wind kelp
#

at least to my memory

young ivy
#

I'm not talking about this server

remote urchin
#

Oh nvm, I gotta hit apply and restart, not just reload ui, my bad

knotty agate
#

hi, i couldn't attend yesterday's center stage session. Maybe anyone took shareable notes by any chance please? 🙏

fervent thunder
#

help me please, how to --cache-dir torch... because it collects it and just freezes my computer, I am installing stable diiff freshly

wind kelp
# young ivy I'm not talking about this server

well, in that case: nudity = pretty normal. every has naked bodies
hardcore gore: aint no one wanna see that. it's not pleasant to click onto a server and see someone getting stabbed to death or their head chopped off or something

coral kelp
#

can @untold mist tell me the promp of generating plzzz?

inner trellis
#

hi can someone please share a link to the AI medicine oriented branch of stable foundation ?

fervent thunder
#

Guys

#

Who have A BIG knowledge about stable diffusion there?

bleak matrix
#

Good morning, everyone

#

How are we today?

warm junco
lavish delta
#

My usual 2 minute SD Renders are now... well over 10 minutes plus..... is it me or did the last update break something

stark flint
#

What is the delay to do an other prompt?

bleak matrix
warm junco
#

Hey Sunny!

stark flint
bleak matrix
bleak matrix
warm junco
bleak matrix
#

Not too shabby! It's 67°F here; gonna be up in the 80's today. More storms, too

#

I love the rain, though

#

I like how the radar is predicting that NY will suddenly get rained on again outta nowhere

bleak matrix
#

Usually it just travels up the coast and whallops everyone, if it's big, etc.

#

This time it's back for revenge

reef epoch
#

is there a rerun of the yesterdays event?

#

summery or a recorded vid?

radiant bronze
bleak matrix
cosmic goblet
#

Anyone know of any good comprehensive models on Civitai that are confirmed to be free from viruses?

remote urchin
cosmic goblet
#

Yes, some have malicious code

remote urchin
#

Gee, well that would've been great to know before I downloaded 200Gbs worth of models, lauras, lycorishes, embeddings and extensions

surreal lance
#

Any update on rocm for windows?

cosmic goblet
#

Lauras are safe as far as I know. It’s just the models

remote urchin
#

but thanks for letting me know

cosmic goblet
#

We are talking about code that lets people hack your computer

#

So people might not be aware it is happening to them

pearl fog
#

Plus, Civitai and huggingface both scan the files for pickles. There will be a green checkmark if they're fine or red X if they're malicious.

#

If you see the red X with "unverified" written next to it on civit, then they've flagged it as malicious. When I've seen it before, it was on the VAE files.

pallid flame
#

what does clip skip do exactly ?

unique hamlet
#

As CivitAI exploration sucks. What is your favourite website for sourcing prompts?

wooden briar
#

If I'm running Automatic 1111 everything is offline local to my machine yeah? I'm mainly asking about img2img using IRL photos of friends/family without them being shared

warm junco
wooden briar
#

TY, I want to toonify some friends as gifts but obviously want to respect their privacy

warm junco
fading jay
#

Is anyone aware of a website that has images generated with various SD models including full prompt and settings that is searchable? The civitai gallery is not tagged well as NSFW with a lot of gross crap and isn't really searchable. Specifically i want to go somewhere and search "snake" and see lots of types of snake images and if one is close to the style I want to generate, i can get the model/lora/embedding and try similar prompts, etc.

pearl fog
#

There's also lexica

pearl fog
fading jay
#

thats great, thank you. The lexica UI is better, but only seems to show their own generated images with their own models. prompthero seems to have what I want, so I will start there

ornate flame
pale latch
#

not like they won't work in sd

serene swan
#

where do put the git pull command for auto1111 again?

plucky shoal
#

you mean perform it on each start?

broken smelt
#

is batch size actually supposed to be faster? i did 10 runs of bsize 1 averaging around 45s, then a bsize of 5 and that actually took longer than the expected 45*5=225s, totaled at 4 mins 1s (241s)

#

ive done various tests in past here and there and they all had the same results, im just not seeing the purpose of it

#

usually run SDE karras, 30 steps, 512x640/640x512

wooden briar
serene swan
tawdry forge
#

Hey guys, is it possible to give an image to the model as a base ?

wooden briar
fading jay
tawdry forge
wooden briar
#

Oh I don't know, sorry. I'm on 1.5 using A1111

tawdry forge
pale latch
dreamy kiln
#

Hello folks, how can I give my own promt to the model from where . How can I give promt to bots to produce to img

fading jay
pale latch
#

their search setup sucks. it takes some wrangling. always go back to the homepage to start over, else it searches within whatever you're looking at

wild steppe
#

SDXL 0.9 Stable Stage with Emad & AppliedML Team!

https://www.youtube.com/watch?v=DK6xAo3iCzU

YouTube
Stable Foundation
SDXL 0.9 with Emad, Applied ML Team & Amli on Discord!

This week's Stable Stage with Emad Mostaque, Joe Penna, Alex Goodwin, Yam Levi & Amli of @Stability_AI

🚀 Summary: SDXL 0.9 Release and Updates:

  • Releasing on API, Dream Studio, and Clip Drop
  • Research weights sent to selected researchers
  • Targeting full 1.0 release mid-July
  • Creative ML license like original SD
  • Continuous improvement ...
▶ Play video
wild steppe
fading jay
pale latch
#

you could scrape those images from all these sites, then set them all up indexed locally