🏞｜general-with-images

[stoic pivot]

Sorry for this, there been so many good images here now and I will ruin the flow with a bad armature meme.

[stoic pivot]

Wow, that was cool @glossy herald

[glossy herald]

got some nice batches

[fickle gorge]

I was a bit disappointed before with my low resolution poorly detailed results, but after that day there was a discussion about upscaling, I learned some things about SD and started getting some good results and started looking this channel more often. I already can notice improvements on my SD generations

[stoic pivot]

3 word prompt: Doomsday, dystopia, ashes

[light gazelle]

Soviet Mcdonalds Propaganda Poster

[stoic pivot]

AI do AI: "You are a survivor of a global catastrophe that wiped out most of humanity and civilization. The world is now a bleak and barren wasteland, covered in ashes and ruins. You have to scavenge for resources, avoid dangers and find other survivors. How do you cope with this doomsday scenario? What are your goals and motivations? How do you interact with other survivors and factions? What are the challenges and opportunities you face in this dystopian world?"

[wispy nest]

oh that's cool

[glossy herald]

I was trying for that rainbow salamander

[wispy nest]

That's nice transition too

[glossy herald]

xD

what was that x)

got another one !

[stoic pivot]

Why do I feel so pessimistic today, "a doomsday scenario of a global catastrophe that wiped a civilization, a damaged robot is looking at a bleak and barren wasteland covered in ashes and ruins, an dystopian world." I need to add clorful and trending on artstation.

[wispy nest]

I'd call this "result of human stupidity" 😄

[summer viper]

don't need artstation

[stoic pivot]

I wrote "a colorful and happy doomsday scenario of a global catastrophe that wiped a clown civilization, a damaged bunny robot is looking at a bleak and barren wasteland covered in candy and glitter, an dystopian world".
And my AI was like: I DO NOT UNDERSTAND YOU, what do you want from me... take this and shut up!

[glossy herald]

that was enough for party cat, let's move on

[wispy nest]

SDXL + this image +"result of human stupidity prompt"

[light gazelle]

suggestion, inpaint the smoke here to be a volcano with the smoke billowing out of it, think that would be really cool

[wispy nest]

more SDXL

[summer viper]

very low res

[stoic pivot]

Cool

[glossy herald]

xD

[wispy nest]

challenge accepted.

challenge instantly failed.

[glossy herald]

xD

[wispy nest]

I don't have enough credits lol

Wanted to go 1024x1024

okaaaay back to firefox relay...

[stoic pivot]

75 % BS based prompt, "best quality, realistic depiction of human stupidity in the form of a suspicious beagle wearing glasses, UHD"

[stoic pivot]

Optimized to "a suspicious beagle wearing glasses"

[stoic pivot]

And it is not AI until it is a bit silly, "beagle dog riding a moped in Amsterdam"

[wispy nest]

Really? You can't even generate 1024x1024 on dreamstudio?

[summer viper]

try 1152x768

[wispy nest]

1024max

[stoic pivot]

@wispy nest But, it generated the text very well.

[wispy nest]

what is 262144 pixels, how do I math it out

[long trench]

Hi again

[summer viper]

this is 512x512 max

[wispy nest]

yea it's just squire root it seems

[long trench]

I finished writing my book, decided to make a chapter book

[wispy nest]

well, that's lame

[long trench]

Any ideas on what I can do for the cover, we can discuss it in dms

[wispy nest]

well , since I made another account anyway...anyone wants to test anything on SDXL?

[glossy herald]

going out with my cow friend
or getting cowfished

[stoic pivot]

"Kulturhuset" and it understood Swedish?

[long trench]

Character design

[wispy nest]

okay I'm disappointed in dreamstudio

[glossy herald]

That's a strange creature

[long trench]

same

[wispy nest]

maybe it's just free limit, idk

[long trench]

Exactly what i need, except the girl in the bg ):<

[summer viper]

Sergey Bodrov

[stoic pivot]

Is this just luck or do my AI understand Swedish? "fika med kanelbulle en fin sommardag".

[summer viper]

ahahaah

[glossy herald]

well time for bed, see you everyone

[long trench]

meh

[stoic pivot]

fika is a relaxing time with coffee, then it is a lovely summer day.

[long trench]

FINALLY

https://tenor.com/view/dace-gif-25608746

[stoic pivot]

Intresting, "en katt dricker kaffe" translate to A cat drinking coffee.

[fickle gorge]

I got results so awful back then I gave up on my native language prompts. Is this running locally?

[summer viper]

russian mafia

[stoic pivot]

Ha ha. Meatball in Swedish is "köttbulle" it can be translated to "Meat-bun", and guess what the AI did give me. A meat bun 😄

[fickle gorge]

Hmmm... it sort of knows something other than English, it seems

[long trench]

why is anything v3 turning into dalle real quick

brown eyes

[stoic pivot]

This is the last for this test, "Kungen dricker öl på Slottet", translate to The king drinking beer on the castle, and it look like Stable Diffusion understand some Swedish words.

[long trench]

whyyyyyyyyy

[stoic pivot]

o.o

[smoky oak]

My goals to use stable diffusion to generate a reference for my VR character were a huge success, and now that I have my VR headset working, my progress on him is flying, and I think I might be able to make money off of RV chat model commissions. That's what my uncle does for a living haha

[long trench]

It generated the exact same thing

[fickle gorge]

Positive prompt: foto de um carro verde estacionado, em uma cidade do início do século 20, carro do início do século 20, ano 1900
Negative prompt: fora do quadro, horrível, feio
photo of a parked green card, in a early 20th century city, early 20th century car, year 1900
out of frame, horrible, ugly
It works in Portuguese too, but it is more frustrating to use as it seems to just ignore a sizable portion of the prompt

I'm noticing the language barrier is being a blocker, I notice other people using SD and getting bad results in their native languages, then they use a translator and get what they want

GPT-4 is doing a CLIP-like thing now of understanding images, but not generating?

[magic stump]

👍

[summer viper]

where are the eyes?

[cold berry]

what generators do you guys usually use?

[ripe cedar]

First time using blender to generate a depth map. Used it but it still doesn't quite follow the depth. Need to practice and learn more I suppose. I'd love to try using a rigged 3d character in a 3d setting to generate stuff to use with SD

[magic stump]

could any one help me create a kitchen drawing with the requirement of this Create a futuristic, science-fiction-inspired kitchen design. The kitchen should be an open-concept design connected to a living room, measuring a total of 20 square meters (4m x 5m). The kitchen features a 4m x 3m wall of white and light champagne-colored cabinets, as well as a 2m island. The overall design should be inspired by the Oppein Modern Kitchen PLCC21325, which can be found at this link: https://www.oppeinhome.com/product/kitchen-cabinet-plcc21325.

My midjourney can not work today

[ripe cedar]

You could trace over it and feed it into the Controlnet scribble model, maybe

[magic stump]

oh

[stoic pivot]

"black red white BRUTAL art", as a test what RAW did for a prompt, so I used brutal as a synonym and got interesting results with FaeTastic v1.0

[ripe cedar]

Or use img2img to make it into something similar

[stoic pivot]

black red white BRUTAL hare art

"black red white BRUTAL pony art" with the AnythingPony v1.0 model.

[ripe cedar]

You mean model?

[cold berry]

no I meant among local installations or online services

trying to collect a few images from less commonly used services

[summer viper]

Stable diffusion (deliberate, faetastic, Illuminati Diffusion)

[ripe cedar]

Just SD for me thus far

[summer viper]

Automatic11111

[stoic pivot]

I start to love "BRUTAL", this is " red black BRUTAL samurai art"

"red black BRUTAL samurai art ukiyoestyle-1500", ukiyoestyle-1500 is an embedding I found on Civitai.

[stoic pivot]

Now I call it a day and goes to bed.

[long trench]

eh

thoughts?

[wispy nest]

🧐 What are you trying to do

[long trench]

making a book cover

and no the book dosen't have sexual themes

the images were anoying

[wispy nest]

if you want text on it, probably wait for sdxl, it's better at doing text...or generate just image and make your own text withou AI

nsfw into negative?

[long trench]

I wrote the text in photoshop

[wispy nest]

multiple view too, if you don't want repeatnig image...idk what you want

[long trench]

did do, some images bypassed

[long trench]

just want feedback on the cover

[wispy nest]

highres fix?

oh

eeeeh...idk

[long trench]

its bad

[wispy nest]

i queued a few images...had to edit the xyz script to let me even start it. probably should have realized it was gonna take a month to do that many lol -_-

[ripe cedar]

Is there an Ottoman and also Egyptian hyroglyph model out there?

[tardy sigil]

kkop idol

[crystal delta]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[grizzled sage]

experimenting with thie cuttoff extension. whatever it's doing is trained on anime to control it, but it's not half bad elsewise. Neon red Lamborghini , Man posed in a Blue firesuit

[heady night]

cute cat in basket

[bright mortar]

hi all. any suggestions on my automatic1111 and stable diffusion? I feel like my negative prompts don't do anything.

i used a model that said it could generate sticker art to get the left image. but i couldn't get any stickers to generate that weren't all mangled. So i thought, to switch to SD 2.1 feed in the best sticker I had and see if SD would return a more usable photo. to no luck 🥲

any thoughts?

[summer viper]

[summer viper]

try experimenting with other models

[bright mortar]

i've been. +/- success. can you help me understand why regular SD2.1 isn't cutting it? I guess I was expecting more variety out of stock SD. I've got good photos that look realistic, and paintings and such. but specific photos like stickers, imaginary stuff not so much.

[summer viper]

try faetastic and oldjourney

models in Civitai

[bright mortar]

woh... faetastic is something. eye opening. i need to study up on this. thanks for suggestions.

oldjourney too. so far I understand the attempt oldjourney is doing. faetastic, i need more sleep. lol

what's picklesensor mean?

[timid solar]

pickle tensor was the original file type used to store models

before safetensors were made

pickle tensors can contain malicious code
and so are not prefered

[summer viper]

"oldjourney" trained in the pictures of "midjourney"

[summer viper]

don't ask how it happened

[keen cipher]

A group of first-grade elementary students went hiking on a sunny day. They were lively and walked in groups of 4-5 along the sidewalk, which was shaded by trees on both sides.

[crystal delta]

@keen cipher

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[glossy herald]

hey and welcome around Rony

we don't have a picture bot, sorry

[crystal delta]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[shadow summit]

Sooo sorry my bad😅

[glossy herald]

no problem 🙂 lots of people try it 😉

cow

[glossy herald]

Dark souls boss

[viscid island]

Minimalist modern online kids toys seller logo

[mossy jolt]

Apple

[glossy herald]

sorry we don't have a picture bot here @viscid island @mossy jolt

[crystal delta]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[smoky oak]

Howdy, people

i am messaging from VR

It is a bit messy, but oh well

@glossy heraldhow are you doing on this fine day/

*?

[glossy herald]

eating right now 🙂 but good, and you ?

[grizzled sage]

sometimes, negative prompts can be detrimental to the goal of your prompt. i notice you're doing drawn artwork and using "badly drawn" and i instantly don't think that this is going to be very effective. you might want to run some tests with xyz grid and an s/r prompt on the negative field.

A lot of people really prefer to put tons of negatives copy pasted from good generations, while i'm on the side of just using simpler ones, and sometimes negative embeds.

[summer viper]

526mix

[onyx scroll]

i love this

[ivory oxide]

https://i.imgur.com/XRv9Fzg.png

any ideas would be greatly appreciated

[glossy herald]

training a landscape model

[hidden mortar]

which model are you using?

[hidden mortar]

WTF

what are those settings bro

I can't achive images like that

[hidden mortar]

h-howw

[glossy herald]

the model name is "realbiter". and yeah, those were great.
Let me give the settings for some

[hidden mortar]

I would really appreciate it

[glossy herald]

anime (8k, best quality, masterpiece:1.2),(best quality:1.0), (ultra highres:1.0), modelshoot style, (extremely detailed CG unity 8k wallpaper), full shot body photo of the most beautiful artwork in the world, beautiful young woman, by agnes cecile, half body portrait, extremely luminous bright design, pastel colors, (ink:1.3), autumn lights, 1girl,anime lines, a surrealist sculpture by Michael Deforge , [:anime line art:0.2], Ice and Hex and solar and Cold themed, epic splash art , excessivism, , realistic
Negative prompt: canvas frame, cartoon, 3d, ((disfigured)), ((bad art)), ((deformed)),((extra limbs)),((close up)),((b&w)), wierd colors, blurry, (((duplicate))), ((morbid)), ((mutilated)), [out of frame], extra fingers, mutated hands, ((poorly drawn hands)), ((poorly drawn face)), (((mutation))), (((deformed))), ((ugly)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))), out of frame, ugly, extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), Photoshop, video game, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, cross-eye, body out of frame, blurry, bad art, bad anatomy, 3d render
Steps: 50, Sampler: UniPC, CFG scale: 5, Seed: 1497396713, Size: 960x544, Model hash: c4947fa2c4, Model: DB_mix_realbiter_v10

anime (8k, best quality, masterpiece:1.2),(best quality:1.0), (ultra highres:1.0), modelshoot style, (extremely detailed CG unity 8k wallpaper), full shot body photo of the most beautiful artwork in the world, beautiful young woman, by agnes cecile, half body portrait, extremely luminous bright design, pastel colors, (ink:1.3), autumn lights, 1girl,anime lines, professional majestic impressionism oil painting by Waterhouse, John Constable, Ed Blinkey, Atey Ghailan, Studio Ghibli, by Jeremy Mann, Greg Manchess, Antonio Moro, intricate, High Detail, dramatic, makoto shinkai kyoto, animation key art, a bronze sculpture by Annie Swynnerton , [:anime line art:0.2], Ice and divine and Poison and Blood themed, epic splash art , figurativism, , abstract
Negative prompt: plastic doll, plastic skin, (easynegative), monochrome, (low quality, worst quality:1.4), cg, 3d, render, (multiples views:1.2), wrong hands, deformed fingers, author signature, text, watermark
Steps: 50, Sampler: UniPC, CFG scale: 5, Seed: 2239559250, Size: 960x544, Model hash: c4947fa2c4, Model: DB_mix_realbiter_v10

anime cow, anime lines, a computer rendering by Unkoku Togan , [:anime line art:0.2], Time and Hex and (fire:0.6) themed, epic splash art , les automatistes, , surrealism
Negative prompt: sketch,nsfw
Steps: 60, Sampler: UniPC, CFG scale: 5, Seed: 33397389, Face restoration: CodeFormer, Size: 960x544, Model hash: c4947fa2c4, Model: DB_mix_realbiter_v10

[hidden mortar]

Thank you so much

[glossy herald]

no problemo, happy you like those.
I don't upscale everything, takes too much time, but lots of those could use some hihger res

[hidden mortar]

I'm feeling lost in SD, it is so hard to use. I mean MJ is so unskilled.

I mean you do, beautiful magic tree and you get world masterpeace

and btw what Sampler do you use? I see that sampler matters a lot

[glossy herald]

yeah. SD is harder to use than MJ, that is for real.
MJ does a lot of things under the scene, and it's hard to replicate on SD, but like you see, you can achieve great things in it too
It's a lot more modular too, like you can really decide of your workflow in SD, whereas MJ doesn't give a choice.
But the learning curve sure is steep

UniPC, EulerA, DPM++ Karas something

[hidden mortar]

do you use automatic?

[glossy herald]

for those, yes, but not only

you can do lots of fun things inpainting/outpainting in invokeAI for example

but recently, I've felt in love with comfyUI

even if it's even more technical than Automatic 🙂

[hidden mortar]

Is UniPC built-in automatic?

I don't have it in Invoke

[glossy herald]

this is the picture I made for the "PoW' contest here this week (check #⭐｜pow-info for more) using comfyUI

and this is what the UI looks like ^^

[hidden mortar]

POV: you spent 8h creating an Image

[glossy herald]

totally

[hidden mortar]

Btw thanks a lot for everything of it

if you have other tips, I would really appreciate them. I really struggle with creating cool, creative abstract images.

[glossy herald]

I spent two days on it almost x) I'm happy you liked it

well, I've been experimenting with each and every example in comfyUI, and for now, I'm trying to get how to really work with noisy latent composing. I don't get the results I want for now. So I'm spending another 8h on a nice workflow

https://comfyanonymous.github.io/ComfyUI_examples/noisy_latent_composition/

[hidden mortar]

Waitt... I just realised that this isn't 4 images combo, this is one image!

[glossy herald]

yes

it's one image, with 5 different zones for prompt

I prompt a flower on the total picture, even use controlnet on it

and on each quarter, I describe an element

then highresfix

[hidden mortar]

https://civitai.com/models/16592/realbiter is this the model you're using?

[glossy herald]

right now I'm working on a workflow where :
1/ I prepare prompts in a somewhat easy format for the bachground and each character
2/ I pre render for just 4 steps each character and the background independantly
3/ I sum the prerendered "pictures", positionning the characters on the bacgkround
4/ I finish to render the scene, composing the prompt
5/ highres fix for 1920x1080 rendering in the end

I'm trying to work on 4/ to keep conditioning the picture correctly, pushing the prompts at just the right place in the final picture for each character

[hidden mortar]

this is magic for me

[glossy herald]

if you like this sort of things, it's really a fantastic UI

it can be modified quite easily too, I'm loving setting those up

[hidden mortar]

Images are really cool, but I don't have that much time to do it

[sleek ingot]

how do I make images

I went to start here but it doesn't say 😦

[crystal delta]

@sleek ingot

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[glossy herald]

and welcome around 🙂

we don't have a bot for it sorry

[glossy herald]

this was automatic

[sleek ingot]

ahh got it, ty Guiz!

[hidden mortar]

oh I see. I like the style. I was talking about that I don't have time for nodes

[glossy herald]

and if you download my pics, and drop them in comfy UI, it should load the workflow and the seed !

(when made through comfy UI of course, not that cow)

[hidden mortar]

wtf

metadata in image for a hekkin ai config

[glossy herald]

? not sure what that means

it's a config of nodes, never heard of hekkin personally

[hidden mortar]

sorry, I was super unclear

I mean just data stored in image

[glossy herald]

yeah there is a lot in there

[hidden mortar]

data - config for SD

that's impressive

[glossy herald]

it's a full json describing the full node organization I did

and like you correctly infered, I spend hours on this so lots of nodes

but it's the best UI if you like those sort of things

you just get "in the zone" making your workflows

[magic geode]

what on earth is that? looks neat (and confusing)

[hidden mortar]

what embeddings do you use?

[glossy herald]

this is a lot more complex than the base setup :

[glossy herald]

none, this was pure models

[hidden mortar]

nodebased generation

[magic geode]

what's the point of using that over a normal SD UI?

[glossy herald]

they do have a tutorial that is also a visual nove if you want

[sleek ingot]

my first cow, yayyy

im using stable diffusion tho on huggingface

[hidden mortar]

more control, but you spend more time on creating that image

[glossy herald]

a lot more possibilities in how you pipe things together. it's like writting an extension especially for your needs each time

[magic geode]

oh nice

can I run it on colab?

my pc is terrible

nvm, found it

[glossy herald]

I think so yeah, it runs on low things as long as you don't push the sizes to big things

[sleek ingot]

so thats in dream studio @glossy herald ?

[glossy herald]

no

that is on my computer

[idle snow]

/blend image

[sleek ingot]

how do I put it on my computer

oh its blend

?

[glossy herald]

#1080946152318443610 is where you'll find the most tools to use it localy, if your computer can handle it
#🤝｜tech-support is where you'll find help setting it up
#1011228442399883294 is where you can find more help running it in the cloud if you didn't have the computer for it

[sleek ingot]

is it free?

ty

[magic geode]

what is your GPU?

[crystal delta]

@idle snow

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[sleek ingot]

NVIDIA RTX A5000 24GB GDDR6 Graphics Card

[glossy herald]

to install, and run yourself, yes

[magic geode]

then yes

[glossy herald]

clearly you run it lol

[sleek ingot]

what if I had a Voodo 3?

[hidden mortar]

do you use some time of repairing, fixing on your images?

Mine looks super blury (it's the same image)

and colors are dull

not vibrant

[glossy herald]

not on the ones I posted there. Sometimes I add nodes to add post processing effects, like reducing luminosity and bloom, but not on those

[sleek ingot]

okay im ready to roll, gonna download now

thanks for the help

[idle snow]

I'm unable to blend images

[sleek ingot]

I can do 3d stuff with this one

[glossy herald]

this has no link to the card you were talking about lol

nope, can't run on that

[sleek ingot]

😄

[magic geode]

troll lol

[glossy herald]

4GB VRAM and NVIDIA are the 2 keys to make it easy

[sleek ingot]

im not, I just thought of that graphics card when you asked so I had to post it lol

[magic geode]

actually kinda funny, unlike most trolls

[glossy herald]

it can work on other things than NVIDIA with more work

[sleek ingot]

thank ya kindly

[hidden mortar]

super weird. You didn't do any photoshop color correction or you have some settings changed in automatic?

I don't have great quality as you do

[glossy herald]

never photoshop, no. the model helps, the VAE too

most I do on top is inpaint

[hidden mortar]

I see, I must use VAE

[glossy herald]

I use the one linked on that realbiter model https://civitai.com/models/16592/realbiter

[magic geode]

is there a guide for it?

[glossy herald]

full example list : https://comfyanonymous.github.io/ComfyUI_examples/
visual novel guide : https://comfyanonymous.github.io/ComfyUI_tutorial_vn/
video first use guide/install guide : https://www.youtube.com/watch?v=vUTV85D51yk

yes, visual novel guide

[sleek ingot]

thats sweet

[fiery carbon]

Try & error

[sleek ingot]

trial & error

[long trench]

Book cover

Advice?

?

[wispy nest]

make sure you're using right vae for a model.
Read through model description where you downloaded it, if none mentioned - use

vae-ft-ema-560000-ema-pruned.ckpt
or
vae-ft-mse-840000-ema-pruned.ckpt

[crystal delta]

FAQ: How do I use the Stable Diffusion website & what do all the settings do?

Check out these videos for a great overview of the website and how to use it! https://youtu.be/014J2Yo1aGI ｜ https://youtu.be/M3jCa6qTHpQ

FAQ: I'm new here! How do I get started?

Welcome! Start by heading over to #1072220168534642768 to get yourself situated and help find the channels you are looking for! Please make sure you review our #✍🏼｜rules-and-tos and feel free to assign yourself some #👥｜roles as well! Answer any questions your may have at our #1072229020520947753. There are many ways of accessing Stable Diffusion, take a look at #1080946152318443610 to start your journey!

[wispy nest]

just learned of the artist "ryohei hase" and faetastic model can mimic that art pretty strongly. not a fan of the dark/horror element, but it's kind of fun to immediately turn everyone into an anthro rabbit lol

[umbral dagger]

Can I share my art here?

[kindred ore]

Where do I go to create my first image?

[crystal delta]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

FAQ: What are the server rules?

The server rules can be viewed at any time in #✍🏼｜rules-and-tos

[long trench]

Fixed it

[wispy nest]

@static tusk

this is my install process. I simply find control net and click install, then reload the page after applying everything. is this incorrect>

[static tusk]

the config is missing the folder. you must have deleted that

[wispy nest]

I haven't deleted or even selected anything

That screenshot is the result of my 4th reinstall

[static tusk]

like C:\stable-diffusion-webui\extensions\sd-webui-controlnet\models\cldm_v15.yaml, just change the file name at the end

[wispy nest]

What is the "config" and can I just add it now?

[wispy nest]

I did change it to cldm_v21.yaml I thought

Like the huggingface page commanded

[static tusk]

your config only has the file name, it needs the whole folder structure and file name,

[wispy nest]

Oh ok thanks. I wonder then why it defaults to that

Am I ignoring the Adapter models field?

[static tusk]

both config fields you changed. change both

[wispy nest]

I haven't change anything but the ending. This is how they appear by default

What is the cldm_v21 model? I'd like to use the model I have currently loaded up in the models folder. Is this impossible?

[wispy nest]

And what am I changing it to? Sorry. I can send you some cash for your troubles

[static tusk]

open the folder in file explorer

[wispy nest]

Which folder

I'm in the explorer

under models/stablediffusion

[static tusk]

wherever you have your controlnet extension

[wispy nest]

Not sure where that is

[static tusk]

where you have stable diffusion saved, similar to this

C:\stable-diffusion-webui\extensions\sd-webui-controlnet\models\

[wispy nest]

I've never had to install anything via the explorer save for model files

I'm running this file explorer remotely

I'm paying a host to use their hardware

And everything is in the browser

[static tusk]

ok you never mentioned that

[wispy nest]

Web ui I thought meant browser

My bad

[static tusk]

there is local instal and online

[wispy nest]

right

this one is online

this is how I access everything

[static tusk]

so can you browse to the controlnet folder and see the 2.1 controlnet models?

[wispy nest]

this is what's in my "control net" subfolder

under models > controlnet

[static tusk]

not sure sorry. you might want to check in #🤝｜tech-support . Im more familiar with local install

[wispy nest]

thanks again

[static tusk]

in any case there should be list of models in the controlnet models drops down. if nothing is there then the models arent in the folder

[wispy nest]

@static tusk

hey sorry to bother again

but I'm going to try and download the control net models manually and adding them in the appropriate file browser via the webui

https://github.com/Mikubill/sd-webui-controlnet

this is where the models are that I need to download

but how do I donwload them?

I clicked on "models"

Then clicked on "go to file"

[static tusk]

that github link is just the extension. the models are on huggingface

[wispy nest]

https://huggingface.co/thibaud/controlnet-sd21

is this the "official" hugging face page?

[static tusk]

yes

[wispy nest]

Thanks a bunch

Which one? (sorry)

and do they all go in the same file?

[static tusk]

safetensor files, 700mb files

[wispy nest]

or folder rather

youve been so nice and helpful thank you friend

[static tusk]

they go in the same folder structure I posted earlier

[wispy nest]

and then I have to copy that extension text into the extention settings in the webui tab?

//stable-diffusion-webui/models/

[static tusk]

for online Im not sure

[wispy nest]

dang

is there a code that I can use to download this directly to the ui?

for example, google drive file download very quickly by using gdown --fuzzy "host download"

It's a difference of waiting 3 minutes or 5 hours to upload this safetensors file

command is the word I was looking for

git lfs clone https://huggingface.co/sberbank-ai/ruT5-base ?

guess not

not working either

!git clone doesnt work either

git clone <url> doesnt work either

!git clone <url> also not working

straight from the horses mouth

[ripe cedar]

Should a 3060 12gb be able to make a 768x768 with high res fix at 30 steps with Euler A? Mine says cuda out of memory when I try.

[wispy nest]

idk how tf everyone else can figure this out

[static tusk]

you will get more assistance in #🤝｜tech-support

[wispy nest]

Ok I will piss off sorry

[static tusk]

yeah should be able to. I can go to 1800x1200 with 10gb

[ripe cedar]

I see, thanks. I'll try to figure out the problem.

[static tusk]

do you have --xformers ?

[ripe cedar]

Yes. It said it allocated like 6gb to pytorch though when it failed. I'll have to check to ensure the xformers is applied although from what I remember, it was

[wispy nest]

Thanks again @static tusk for the kindness ❤️

[smoky oak]

Whats up, stinkers?

Hope you are all having a nice day

[static tusk]

sup. yeah all good

[smoky oak]

Thats good to hear

I have been working in VR sculpting for the last few days with my uncle to bring my VR chat model to life. I actually generated the references I used for him in stable diffusion. It was a whole long ass process, but hes looking so damn good and I am happy :>

[static tusk]

how does vr chat and sculpting combine together? I havent looked at VR much at all

[smoky oak]

There is a free program called Adobe Medium (not actually made by the, just bought out)

and its insanely good

its sculpting, but instead of surface displace, its 3D voxel. So you can basically draw mass in 3D space

[static tusk]

ah right. Ive seen something like that. 3d modeling in VR

[smoky oak]

so that is what I am using. Its wayyy faster cause you just draw in 3D space rather than having to rotate a flat image around and around and around

[static tusk]

did you see emad's tweet about SD 3.0 coming very soon?

aka SDXL

[smoky oak]

Wait, wait

is it confirmed to be 3.0?

not 2.2?

cause if it is 3.0, then we have a better chance of it having a less trash TE

[static tusk]

https://twitter.com/EMostaque/status/1641795867740086272?s=20

Emad (@EMostaque)

Stable diffusion XL #SDXL is in testing with our partners & https://t.co/MrZUa23TpG, getting data to refine it before open source release.

It is a 2.3 billion parameter variant (original 900m) with a range of improvements that will feed into Stable Diffusion 3

🚀 ✨

Likes

887

Retweets

126

Twitter · Mar 31, 2023, 1:33 PM

[smoky oak]

it is 3.0

[static tusk]

yep

[smoky oak]

please for the love of GOD don't let it be another failure like 2.x

[static tusk]

the issue may be the file size and vram constraints

[smoky oak]

SDXL on its base is better at some things than the best trained 1.5 models, which is insanneee

cause base models are usually garbage

[static tusk]

yeah all base models arent anything to write home about

[stoic pivot]

Prompt was: "lora:epiNoiseoffset_v2:1.2 red-glow black yellow white detailed [brutal art|splatter art] profil art lava devil sheep"

[smoky oak]

@dense tapir Not sure if you say, but I guess SDXL has been confirmed to be Sd 3.0 and now 2.2.2 as was originally suggested in the beta previews

which means there very well could be a good chance it has a better text encoder :>

well I mean, it does, we know that based off of just how damn well it listens

but I mean we may actually be able to train it this time

[split rover]

XL is not 3.0 but much of what we are learning making XL will go into 3.0 proper though. Some cool new model tricks and scaling ofc 😄

[smoky oak]

[smoky oak]

Are you able to say anything on if 3.0 is gonna be another 2.x flop? A good base, but a nightmare to train/finetune?

[static tusk]

so 3.0 is a step up from 2.2.2 XL?

[stoic pivot]

Easter image

[smoky oak]

I want so bad for 3.0 to be trainable, cause 2.x really was such a let down

[split rover]

XL is still listed as Beta because we are still doing some work on it and training on it a good bit more too. It's got a few new things it in and we are seeing how we can push it to make it the best for everyone before pushing it out as the "finished" XL

that said I am really excited to see what everyone does with the full thing when you all get your hands on it

[static tusk]

awesome sauce

[spring kelp]

How does one try out these beta models yall talking about

[static tusk]

only online https://beta.dreamstudio.ai/generate

[smoky oak]

I hate this hahaha

thats rancid lol

[grizzled sage]

porn as other things is such adumb meme. low effort

i miss when trolling was impressive. like maddox era

[smoky oak]

I mean, I have been working (been paused for a while) on an extremely high quality LoRA of Ricardo Milos

[smoky oak]

taking more than an hour per data set image

[grizzled sage]

well making porn i can't criticize. thats something people make can a living off of and it is probably legit. Trying to post it as food though, thats just weird and low effort

[smoky oak]

Yeah, I get you

[grizzled sage]

now i've fired them up

the littlest adrenal gland that could

[spring kelp]

Ah oki ty

[rain gazelle]

this channel is weird 😄

[stoic pivot]

You Weird 😄

[smoky oak]

Mom, pick me up, I'm scared

lol

thats how I feel sometimes

[smoky oak]

Ah yes, the emo grinch bunny thing

[dry cosmos]

It's-a Me

[stoic pivot]

Peach butt cheeks!

You will soon get one more to play with.

The make it red n blue

[grizzled sage]

must be friday night thirst hour

[rain gazelle]

that was fast

do this one

[stoic pivot]

Prompt for my easter bunnies was "lora:epiNoiseoffset_v2:1.2 orange-glow black green white detailed [brutal art|splatter art] easter bunny devil hug".

[rain gazelle]

you must be on a 4090 or something. 😄

[dry cosmos]

3060

[rain gazelle]

oh nice, me too!

[dry cosmos]

Time taken: 11.26s

20 steps

[rain gazelle]

ah ok

[dry cosmos]

but 10 steps get good results too

[stoic pivot]

me 3090, it was a gift so I think it is good, but it run so hot when working.

[dry cosmos]

10 steps
Time taken: 5.75s

[rain gazelle]

yikes

[grizzled sage]

you sunsabitches. i'm in.

[grizzled sage]

ami doin it right?

[stoic pivot]

I like the fury modell I have, but every time I switch to it and forgot negative prompt or the word "sexy" is in the prompt I'm going WTFffff.

[rain gazelle]

left one is pretty good!

[dry cosmos]

boku no hero style

[rain gazelle]

like 75% of my generations are supergirl ... it always just goes back to her. I don't know why. 😄

[smoky oak]

Here we go

[strange jungle]

Arnolina Schwarzenuperman?

[smoky oak]

@strange jungle This is the poor result I got from AItrepreneurs video on LoRA's. This was after probably 15 training sessions at more than 30 minutes a pop. Its garbage

And this is what I got first try from my first proper LoRA training in 7 minutes flat

[smoky oak]

He gives reallyyy bad info lol

I don't trust hardly anything AItrepreneur says

[strange jungle]

it got the color right

[grizzled sage]

a lot of them just throw anything at the wall and if it "sticks" they just publish and farm links

[smoky oak]

right

[strange jungle]

I dont trust anyone with bad background music when they talk

[smoky oak]

blegh lol

[strange jungle]

wow that rentry starts out confusing me

it says install this script that will break you webui

[smoky oak]

This is a LoRA I trained on toothless

[grizzled sage]

i prefer reading because i can comb through the info and as i work it, determine if it's working or not. videos are like, so.. on rails. one directional. hard to scrub them. video tutorials have never been a favorite thing of mine. they're just an ad grab

[strange jungle]

download another powershell script - edit it for the right paths

[grizzled sage]

video classrooms are a different matter. lots of learning through video lectures ot be had. information comes in many forms. video tutorials are usually just rotten an rushed

[grizzled sage]

ah yeah it's hulk goku!

[strange jungle]

I am not installing something that will break the webuii

[grizzled sage]

lol yeah there's some advanced stuff in there so that you get cool symlink style references to folders

[static tusk]

its written in all caps and highlighted in red to not install in your webui..lol

[grizzled sage]

i just copy the model

[strange jungle]

it's beyond me from the very stary

start

[stoic pivot]

Supper Hero.

[strange jungle]

Yes, that's what I mean. It says create a separate virtual environment....no idea what that means

[grizzled sage]

take the information you need from the guide. see.. following them step by step they're not meant

aren't you a tutor? learning is your profession

[strange jungle]

No it's not

teaching is my profession

[static tusk]

first step. read the instructions

[strange jungle]

I read it.

[grizzled sage]

teachers are there to faciliate learning. you should know how this works.. lol

[strange jungle]

Now I have to look up how to make a virtual environment\

[wispy nest]

what rentry are you talking about?

[grizzled sage]

like what even is the problem. you found one issue with file management, so you just threw the entire guide out the window? why even ask for help to begin with. you're just trolling again frank. stop it

[strange jungle]

https://rentry.org/59xed3

[stoic pivot]

Spider-bat

[grizzled sage]

complaining the whole time when you ask for help. it's so tenacious

[dry cosmos]

prompt: complaining the whole time when you ask for help. it's so tenacious + LORA

[grizzled sage]

I was actually going to explain some things to you, but this became mania fast

[grizzled sage]

an anime girl who doesn't look 12. awesome!

[dry cosmos]

1girl, old,white hair, wrinkle, 60 years old

[grizzled sage]

nope. 12 again

[dry cosmos]

same prompt

[grizzled sage]

1200 in the lore

[dry cosmos]

1girl, old,white hair, wrinkle, 60 years old, fat, chubby, short, elderly, warlock

without 1girl

[strange jungle]

try woman instead of girl

or "old lady"

[dry cosmos]

with woman

[grizzled sage]

1girl is what does it. not only is it an anime booru tag for image sets, diffusion makes all "girls" suuuuuuuuuuuper young. i avoid the word and use woman instead

yeah those are way better old people results

[strange jungle]

Ah, I cant see too well

didn't notice it was a tag for the lora

[rain gazelle]

neat

[strange jungle]

I get the opposite problem with males

I thought I could make images for people showing their kids in professional/championship sport moments

your 10 year old little league player hitting the winning HR in the World Series

[stoic pivot]

I shall mix two new characters and you have to guess.

[strange jungle]

it's not easy to keep it from making adults

[strange jungle]

that's a kid's head on a man's body

[strange jungle]

I found I could use images of people in action to controlnet an action pose, but then especially it always makes adult looking people. No good for "see your kid in the world series"

imagine your son or daughter hitting the winning touchdown in the world cup, or getting the 123 pin in the Stanley Cup

[stoic pivot]

Darth Man

[earnest solar]

The sunflower druid.

[wispy nest]

Which model is this? Those colors are nice

[grizzled sage]

makes it impossible for some workflows but like, i took a whisky shot an fired one out

WABAM

home run i'd say

[dry cosmos]

Elysium

[earnest solar]

I'm using an Akira Toriyama LORA, does it show?

[grizzled sage]

AM OUT . /micdrop

[strange jungle]

Right. The bat is awful, the stances awful. especially if I use images for good poses, it makes not kids

[grizzled sage]

diddums 😦 must be impossible /shrug

better give up now

[rain gazelle]

i made tater tots

[strange jungle]

I haven't given up

[strange jungle]

I have been working on it, Simply Do, for hours and days and weeks

[earnest solar]

Looks nice! What did you do?

[strange jungle]

Not sure which is better, tots or fries

[dry cosmos]

just img2img controlnet

[strange jungle]

controlnet what?

[earnest solar]

That's fair enough 😄

[strange jungle]

cany, scribble, depth, openpose

[rain gazelle]

both great, depends on mood

[dry cosmos]

ControlNet-0 Enabled: True, ControlNet-0 Module: canny, ControlNet-0 Model: control_sd15_canny [fef5e48e], ControlNet-0 Weight: 0.55, ControlNet-0 Guidance Start: 0, ControlNet-0 Guidance End: 1

[earnest solar]

May I ask why you chose my sunflower druid?

[strange jungle]

I usually try cany and depth and scribble to see which comes out best

[dry cosmos]

because you are special, blessed by the eternal father

[earnest solar]

😊

[stoic pivot]

Two characters mixed with art, can you guess who?

[earnest solar]

Okay, wow - the Toriyama really shows on this one, but next to perfect.

[dry cosmos]

eichiro oda

[earnest solar]

👀

[stoic pivot]

Elsa from Frozen and Darth Vader.

[wispy nest]

which version? base elysium or one of the anime ones?

[dry cosmos]

anime

[earnest solar]

Hmm, not seeing as much of the purported Ghibbles

[earnest solar]

ah yes, bitcrunch.

You ever just watch as it's making the picture, and the shape is coming through amazing, and then it "corrects" itself and destroys some details that looked good?

[long trench]

how

[dry cosmos]

pixel model

[long trench]

gimmie

[earnest solar]

There, slightly more ghibli

[dry cosmos]

here

[long trench]

dude pixeled my dude

[stoic pivot]

Two Papa Smurf

[earnest solar]

He goes from happy to bewildered

[long trench]

Do this MF

[grizzled sage]

clip skip i think fixes this on some models. i dont know what it should be set to though. effectively skips those last steps

[wispy nest]

mind telling me exactly which model it is? tried elysium anime v3, elysium anime v2, and elysium anime kuro and none of them come close to that. or is ityour prompt that's resulting in those colors?

[dry cosmos]

with lora

[wispy nest]

then mind telling me which lora?

[long trench]

I need a style recreated in ai, but idfk how to train my model without giving errors

[dry cosmos]

boku no hero

[earnest solar]

Not what I was expecting.

[wispy nest]

would be a hell of a lot easier if he/she would just store the prompt info in the image -_-

[dry cosmos]

stardew valley,farm, 1boy, full body, white background, pixel art, gaming
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 6, Seed: 1961998022, Size: 512x512, Model hash: e6b5608138, ENSD: 31337, Pixelization pixel size: 4

Time taken: 17.72sTorch active/reserved: 3844/4384 MiB, Sys VRAM: 6799/12288 MiB (55.33%)

[wispy nest]

appreciated, but i was referring to the anime stuff you were posting earlier lol

[stoic pivot]

Expanded hero universe we do not want, Peter Pan-man.

[dry cosmos]

1boy
Negative prompt: lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 6, Seed: 2874368427, Size: 384x384, Model hash: 50265e48df, Denoising strength: 0.5, ENSD: 31337, Hires upscale: 2, Hires steps: 10, Hires upscaler: R-ESRGAN AnimeVideo, Postprocess upscale by: 1, Postprocess upscaler: R-ESRGAN AnimeVideo

[stoic pivot]

With a 8-pack, darn that is so Chad.

[dry cosmos]

you can use the yiffymix to furry/avian/scalies

[earnest solar]

It delet'd her books 😮

[south grove]

#🏞️ red sun

[stoic pivot]

I think I took it to far now, so I will go to bed. Ni ni

[dry cosmos]

prompt: 🏞️

[wispy nest]

still can't seem to get those kind of colors. i tried the various anime models and set my settings to match yours, maybe i've got the wrong boku lora? i picked up the 288MB one from civitai

[dry cosmos]

send an image

if it is saturated, it is because of VAE

[wispy nest]

i've still got the wrong model as well, apparently. hash doesn't match.

[dry cosmos]

the model doesn't make that much difference

is a heavy LORA

using a realistic human model

will depend on the seed

[wispy nest]

here's what i'm getting with as close as i cna get with those settings

i changed the denoising to 0.6 just because it was super pixellated, but other than that i'm genuinely at a loss

[dry cosmos]

boku_no_hero_academia, midoriya_izuku, saburou_(326xxx1), 1boy, armpits, bed, bed_sheet, blush, breasts, brown_hair, cleavage, closed_eyes, collarbone, drooling, freckles, green_hair, hug, lying, messy_hair, mouth_drool, on_back, on_bed, saliva, short_hair, shorts, sleeping

[wispy nest]

v2 gave me better results, but still not nearly as clean

sorry, forgot to include the lora on that last test, trying again

got a lot closer, lora weight was too high.

[thorny coral]

Чупакабра

[crystal delta]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[dry cosmos]

https://tenor.com/view/volodymyr-zelenskyy-zelensky-володимир-зеленський-ukraine-gif-25721682

[dense tapir]

For 2.1 that is a fantastic loss.

[earnest solar]

Kelda Keet, Secretary Bird - The right-hand to the Guildmaster, this officious wizard seems to keep the guild running all but solo. Their blunt, stern manner serves to keep the guild supplicants in line. A parrot Aaracockra who wears very formal clothes, glasses, and carries many documents around with them.
Aaracockra, lv14 Wizard.

[earnest solar]

ngl, much better than what I'm getting outa the prompt.

[dry cosmos]

upscaled version

[flat pike]

my collueques at work on monday morning

[summer viper]

hahahahahahahahahahahahahahaa

[sterile kiln]

Les vachettes d'interville apres reconvertion

[glossy herald]

c'est ca, faut bien leur trouver du taf

[shy sky]

Flowers

#🏞｜general-with-images red cards

[wispy nest]

general-with-images red cards

/general-with-images red cards

[glossy herald]

hello @shy sky and @wispy nest . Sorry to disapoint, but we don't have a bot making picture here

[crystal delta]

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[wispy nest]

#🏞｜general-with-images Bikini Beauty

[glossy herald]

@wispy nest I just pinged you and said there wasn't any bot for that here

[wispy nest]

Sorry,I think you're a robot

[glossy herald]

lol

I'm a mod

I'm not a robot but thanks

[wispy nest]

I see. Thank you for your answer

[smoky oak]

You look like you would benefit from my Dwayne Johnson LoRA

[wispy nest]

test

[smoky oak]

My LoRA for Dwayne Johnson

its an old and bad one, but I can remake it

[glossy herald]

did you see Dwayne eating rocks ? https://www.reddit.com/r/StableDiffusion/comments/126lieg/dwayne_johnson_eating_rocks/

[wet bolt]

#1072229020520947753

[glossy herald]

need some help ?

New controlnet model ?!?

[smoky oak]

oh sick!

[glossy herald]

https://www.reddit.com/r/StableDiffusion/comments/1281iva/new_controlnet_face_model/

Although other ControlNet models can be used to position faces in a generated image, we found the existing models suffer from annotations that are either under-constrained (OpenPose), or over-constrained (Canny/HED/Depth). For example, we often want to control things such as the orientation of the face, whether the eyes/mouth are open/closed, and which direction the eyes are looking, which is lost in the OpenPose model, while also being agnostic about details like hair, detailed facial structure, and non-facial features that would get included in annotations like canny or depth maps. Achieving this intermediate level of control was the impetus for training this model.

available only for 2.X for now

[smoky oak]

DEATH

[rain gazelle]

looks like an older Ms. Marvel with her black costume mixed with Thor.

[severe spindle]

#1072220168534642768

[lilac pilot]

Very nice been away from this stuff for a while. there is controlnet for facial manipulation (your second to left image)?

I just saw your model post

[glossy herald]

it seems so yeah, it's not integrated by default yet and only for 2.1 model for now

[lilac pilot]

Alright thanks

[glossy herald]

what about my model post ? x)

[lilac pilot]

I mean the post linking the control net haha

[glossy herald]

ho, ok ^^

just in case of confusion, I did not train that controlnet, I was just relaying the info I found

and yeah, it's for face manipulation. I find their logic quite good. canny is too restrictive, while open pose isn't enough

[lilac pilot]

Blink of a eye that dev will have it ready for us

[glossy herald]

they are trying to get it merged into the controlnet extension, and are training the same for 1.5, so I hope they do keep on going, this seems like very good

[crystal delta]

FAQ: What is Stability AI?

Our vibrant communities consist of experts, leaders and partners across the globe. They are developing cutting-edge open AI models for Image, Language, Audio, Video, 3D and Biology.. AI by the people, for the people. Learn more here

[lilac pilot]

@river quiver Welcome!!!

[glossy herald]

I want to play some overwatch again... I managed to stop, I should resist the call

but that name

that gives me bad thoughts

welcome, just joking 😉

[sterile kiln]

https://tenor.com/view/eyebrow-raise-shrek-serious-gif-14262407

Looks shrek

[glossy herald]

Shrek is cool though

while doing party animals I got some wild ones

[fluid herald]

helo

[glossy herald]

Hey there

[stoic pivot]

AI: To confirm you're not a bot, identify images of "zebra crossing a street".

[sterile kiln]

Shrek is awesome, i know every word of the 1st

[flat pike]

cyberpunk cat-girl and cyberpunk cow-girl^^

[glossy herald]

cow is the new cat

[proud mauve]

/cat fly

[glossy herald]

/cat is a new one

[flat pike]

my gallery grows so fast ^^ https://prompthero.com/BalrogDx

[rain vine]

fashion shoots are dead - gandalf & legolas modelling for loro piana

[stoic pivot]

How many tries will it take for your AI to understand the prompt "zebra crossing road"?

[near mirage]

UFO

[dense swallow]

a rose flower

water pump

#1072229020520947753

#1072220168534642768

[wispy nest]

I like cow more

[tropic shell]

#💬｜general-chat message

[tropic shell]

You first! 🤣

[hardy tulip]

never had this one before, new one for collection!

[sleek ingot]

I want that minotaur

nice on the batman @dense tapir can you give me the files

[sleek ingot]

@dense tapir are you using dreamstation

[dense tapir]

nope

[sleek ingot]

what in gods name u using then son

I like that walking dead one

[dense tapir]

A lora I just trained as I have been testing but sadly time is up on colab

[sleek ingot]

niceeee

[dense tapir]

I have so much testing I need to do but I just can't

[sleek ingot]

what is a Lora?

[dense tapir]

think of it as a model

I do styles others do subjects

[sleek ingot]

nice General, thats pretty sick

holy cownards Cowzmus, I like that moocow

[glossy herald]

moooooo !

[dense tapir]

I wish I had a gfx card that could do them so I could do this testing

[glossy herald]

going for cats now though :p

[sleek ingot]

ahhh that was what I was thinking General, gonna ask you about that. What do you have, a Voodo 3 card or?

[sleek ingot]

nice @glossy herald so then with dreamstation you can just fine tune the heck out of each little vector (node?)

[dense tapir]

Might as well be with this stuff

[sleek ingot]

so like if I made a minotaur and I wanted to focus on its should pad on its left shoulder, I just can?

yeah im a programmer not an artist, worst artist in the world, but now i wont have to pay for art anymore... cause of all this

[dense tapir]

My card is a 1060 6gb so too slow and not enough mem

[sleek ingot]

ohhh yeah the famous 1060

[glossy herald]

dreamstation ? not sure what that one is. I use EveryDream to train personally

[dense tapir]

Nvidia is too crazy now to buy

[sleek ingot]

omg woops, im thinking of MidJourney

[long wind]

dreambooth is good?

[sleek ingot]

yeah it is, but @dense tapir did you hear that soon we wont even need GPUs cause Nvidia will host it all

[dense tapir]

I am so hoping they lower the prices but even then a 4080 is my min I need for this stuff

[sleek ingot]

wow taht is insane

[dense tapir]

Well, cool then I can finally toss my PC in the rubbish bin.

[sleek ingot]

hahah yes, maybe this year

[long wind]

hey guys someone can help me?

[sleek ingot]

prob not till 2024 spring

[sleek ingot]

whats the difference from this and MidJourney? they are both text to image?

[long wind]

why my dreambooth trained generates these odd things? eyes and teeth

[sleek ingot]

@long wind sweet pic tho

[long wind]

hahahaha thank you

[dense tapir]

Trythis when gen. Neg prompt - teeth

not joking

[long wind]

trying now

[dense tapir]

for 2.1 I have to do that all the time

[long wind]

i use 1.5

im learning, im noobie hahaha

[sleek ingot]

im such a nub. @glossy herald what is difference from Stable Diffusion and Midjourney?

GPT 4 doesnt know

[glossy herald]

MJ does more than just push your prompt into its model. it adds flavor to it, it adds a fitting style based on what you prompt, and I believe it also applies some post processing hypernetworks or something, it feels a little more than basic generation for sure. They also use a series of custom models in my opinion. they aren't open on their pipeline so this is mostly my speculation.
SD is what you make it to be, you build your pipeline, choose your models, add this and that. it's your own cauldron, and can fit a lot more situations

SD is just the inner brick, text to image. you make it "midjourney" yourself

[sleek ingot]

interesting.. @glossy herald

so if im a non artist, a complete nub with stick figures, should i start with Stable or Midjourney?

[glossy herald]

it will completly depend what you want to do.
if you want to just make some nice pics for fun, yes, MJ got you covered.
if you want to use this professionally in a specific way, you'll need to be able to train models. and that is not in anything MJ can provide.
After all, one of the main pieces of MJ currently is SD itself

[sleek ingot]

intersting indeed!! thank you! @glossy herald

last question for now, is this as hard to learn as say Adobe Flash CS3 (old school) or any Adobe Flash?

[sleek ingot]

or like making music ike hard to make for like 2 months until you get used to it

im guessing super hard 😄

[glossy herald]

I haven't learned flash.
The good thing is, you have different tools available to you depending on your profile
From the most artistic (invokeAI, top notch UI) to the most technical and versatile (comfyUI, node based UI), while not forgetting the most popular all around (automatic1111, nice UI, the most feature complete of the lot)
there are almost for any tastes (lots of other options too)

[sleek ingot]

@glossy herald wow, crazy stuff. That strawberry sushi looks high key bomb on god no cap tho. But really, yeah that looks like my style to just make some easy pics fast

legit though, wow someone should make straberry sushi irl

[glossy herald]

this is the same UI than that sushi, best video to sell it imo https://www.youtube.com/watch?v=WmOUl8Gab5U

it's a "speed run" use of one of the features, the canvas

[sleek ingot]

I love it, thank you for your help. This is huge...

do you think my radeon rx 580 will be okay? or do I need more juice?

[glossy herald]

AMD has some problems running some of those, let me check

this one is compatible with AMD, yes:

An AMD-based graphics card with 4 GB or more VRAM memory (Linux only)

[sleek ingot]

hooray!! 🙂

thank you for your knowledge and efforts

[glossy herald]

https://invoke-ai.github.io/InvokeAI/

no problemo, always happy to open the doors to Narnia to people

[sleek ingot]

okay, first thing ill make in invoke is some turkish delight in honor of you

[glossy herald]

:p as long as you come and share around for the next new person to get interested too, i'm happy you know

I just think this is something to be tested by the most possible people

[sleek ingot]

absolutely

[gilded gyro]

#🏞｜general-with-images ultra detailed artistic photography, midnight aura, night sky, detailed gorgeous face, dreamy, glowing, backlit, glamour, glimmer, shadows, oil on canvas, brush strokes, smooth, ultra high definition, 8k, unreal engine 5, ultra sharp focus, art by alberto seveso, artgerm, loish, sf, intricate artwork masterpiece, ominous, matte painting movie poster, golden ratio, trending on cgsociety, intricate, epic, trending on artstation, by artgerm, h. r. giger and beksinski, highly detailed, vibrant, production cinematic character render, ultra high quality model

[glossy herald]

sorry @gilded gyro we don't have a bot making pictures on this server

[crystal delta]

@gilded gyro

FAQ: How do I generate images? Is there a bot on the server?

Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝｜tech-support for any problems you encounter while installing it locally!

[glossy herald]

let me run it for you though

adding some of my personal flavor to it

[long wind]

not worked

[glossy herald]

and without flavor to it

it's missing cat djs though

[sleek ingot]

sheesh, such beautiful art. I wonder if DeviantArt will go away

[glossy herald]

I hope they don't

what would we train on if they did

x)

sorry

[sleek ingot]

thats where I used to get people to make art for me. Paid a ton of peeps from 2004 to now

[glossy herald]

had to make that one

[sleek ingot]

hahah

[glossy herald]

PARTY CAT

[sleek ingot]

hotness

[glossy herald]

when he does scratches on the disk, it's for good

lol there is one special in that batch

[sleek ingot]

God, its all too cool

[glossy herald]

I'm upscaling that botom left one

[sleek ingot]

lol

Blines

instead of bboy

[glossy herald]

let's say the cat goes to war next

[sleek ingot]

😂 👏 so sick

[sleek ingot]

sheesh

its brain overload

just too much, too excited

[glossy herald]

works so great lol

[sleek ingot]

cant believe all this is real

[glossy herald]

where would he go next ?

[sleek ingot]

to mars

[glossy herald]

astronaut cat

[sleek ingot]

yaasss