#💬｜general-chat

1 messages · Page 83 of 1

fickle siren
#

Too deep for me to comment on right now, after severa lbeers 😄

pale latch
#

so right now, you put an input into a Digital neural network, and it infers that thorugh the weights. Billions Millions of numbered parameters that multiply against each other. After that calculation, it''s off. It's no longer thinking.

#

so if self awareness is achieved, it would only exist in those moments of execution

#

would we even know?

#

is something waking up and existing for the first time whenever we run inference?

#

deep? we've barely scratched the surface

prime loom
#

Any good and cheap cloud service on which I can fine tune stable diffusion?

pale latch
#

runpod has always looked half decent to me

humble ravine
#

hru all

young ivy
#

are there any open source dictation/voice typing AI trained softwares? (free) That can learn my voice so it doesn't look like I'm speaking seventeen languages like windows 11's built in does?

#

or just free software doing the same

plain crane
#

Yo , what is better for stable diffusion, rtx 3090 on the cloud or colab pro? It seems like colab cheap tpus or gpus are worse than an rtx 3090.

pulsar venture
#

If it could be possible, replace text embedding llm bert with gpt4 will give better output?

clever vault
#

Anyone here with a rtx 4070 and can share Performance?

soft moss
#

how many options are there for "live image to image" ? i have tested the Krita plugin, and it works, but i kinda hate Krita XD. is there any other method you guys know ? or app? i prefer gimp but there doesnt seem to be anything

hasty hornet
#

Anyone has a good way to fill "empty" parts of an image?
Without masking it by hand, it'll get messy, too many details and without changing image itself
I don't remember how I was doing it before thinking

#

I mean on png, where you have no background

weak kettle
#

anyone with good kohya config to make face model?

rancid hedge
#

Error occurred when executing CheckpointLoaderSimpleWithNoiseSelect:

'ModelSamplingConfig' object has no attribute 'sampling_settings'

File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\execution.py", line 153, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\execution.py", line 83, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\execution.py", line 76, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\custom_nodes\ComfyUI-AnimateDiff-Evolved\animatediff\nodes_extras.py", line 49, in load_checkpoint
out[0].model.model_sampling = BetaSchedules.to_model_sampling(beta_schedule, out[0])
File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\custom_nodes\ComfyUI-AnimateDiff-Evolved\animatediff\model_utils.py", line 59, in to_model_sampling
ms_obj = model_sampling(cls.to_config(alias), model_type=model.model.model_type)
File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\comfy\model_base.py", line 28, in model_sampling
return ModelSampling(model_config)
File "E:\StabilityMatrix-win-x64\Data\Packages\ComfyUI\comfy\model_sampling.py", line 27, in init
beta_schedule = model_config.sampling_settings.get("beta_schedule", beta_schedule)

soft moss
rancid hedge
#

Error occurred when executing KSampler:

forward_timestep_embed() got an unexpected keyword argument 'time_context'

hasty hornet
soft moss
#

then i dont understand what you need help with

hasty hornet
soft moss
#

image2image inpaint

hasty hornet
# soft moss image2image inpaint

this isn't going anywhere unless you read the question...
Without masking it by hand, it'll get messy, too many details and without changing image itself

soft moss
#

there is an extension called "segment anything"

#

maybe thats what you are looking for ?

#

oh sorry, "inpaint anything" its better

hasty hornet
#

yea that might be an idea

soft moss
#

inpaint anything even has options to do the inapinting right there from the extension, or to export the mask

zealous maple
#

I'm sorry for asking this, but could someone please tell me which version of stable diffusion I should be using?
There's 1.5, 2.1, and XL

hasty hornet
zealous maple
#

Thanks! Is 8GB VRAM enough to run SD?

hasty hornet
#

even 4gb is enough

#

I think 8 should be fine for sdxl too

zealous maple
#

There is a big quality difference between 1.5 and XL?

hasty hornet
#

but people don't really use or compare base models

true field
#

i know gpu + gpu vram is the #1 bottleneck for stable diffusion. what is number #2? is it system RAM or is it CPU?

deft summit
#

hello is it possible to run confyui remotley but not from the same network

zealous maple
soft moss
true field
trim magnet
#

the gpu mem bandwidth

autumn obsidian
#

Noob question here! When using Textural Inversion (Embedding), is it exactly like LORA/LyCoris? I just have to click the embedding in the correct positive/negative prompt box, and use the activation words? Or are embeddings always on? Thanks!

zealous maple
true field
#

pcie bandwith between ram and GPU?

trim magnet
#

no,the 4060 ti bandwidth

#

500$ for 128bit bandwidth

true field
#

i dont understand

#

i am budget limited so i cant getanything better than the 4060 ti, i know it is cripled on purpose by nvidia

zealous maple
# true field i dont understand

He means the memory bandwidth on the GPU itself. GPU chip to GPU memory. I was talking about the PCIE bandwidth, which is the CPU+Ram <-> GPU communication bus

trim magnet
true field
trim magnet
#

or you could also go with a new intel a770 16gb for 300$

true field
true field
finite cloak
finite cloak
trim magnet
#

yea also pcie 4.0 vs a 5.0 doesnt matter on a 4060 ti unless u have a 4090 or an a100

true field
#

like 5.0 is overkill for a 4060 ti?

#

ah

trim magnet
#

yea unless u plan to upgrade later then its better to buy the newest motherboard with pcie 5.0

finite cloak
#

PCIE 5 won't be used by any current cards - RTX 40xx uses gen 4

true field
#

ah

trim magnet
#

yea i guess u right,unless u wanna go with an a100 maybe it would be useful

stark garden
#

what do fps_id, motion_bucket_id, cond_aug, and num_steps parameters do in Stable Video Diffusion?

finite cloak
stark garden
#

in example code they set steps to 25 for small model and to 30 for xt one

finite cloak
#

means the same wherever

zealous maple
#

What UI do you guys use to finetune SDXL? There are so many options now, I'm at analysis paralysis

fossil ivy
#

hi guys

dire kestrel
#

Auto1111 with dreambooth, if you're asking how to train a model.

zealous maple
#

i.e LORA finetuning

dire kestrel
zealous maple
#

Not in the machine learning sense, but sure 🤷‍♀️

#

What UI should I use for that

dire kestrel
zealous maple
#

Yes, exactly, thank you!

dire kestrel
#

I may be wrong but in the sense of finetuning and training I think in Generative systems like this finetuning means taking a model that exists and changing it.

And training is just training off the data.

zealous maple
#

Your understanding is correct. Training a model typically means training from scratch. While finetuning (whether using LORA or not) means updating the weights of an existing model to be better at something new. I think thats what Dreambooth does. I'm reading up on it now

#

There are just so many UIs for it. There's automatic-1111, StableSwarmUI, ComfiUI, and a couple of other ones

dire kestrel
#

That's what I've used to make new models 6 months ago, things may have gotten better since then,

#

I don't think you can train using comfy, interested to know if youcan.

zealous maple
#

I'm not sure tbh

wraith sand
#

Any news on the 3D API from stability? I've registered my interest but not heard anything back yet 😦

desert meteor
#

Finally cracked Ai image generation , can fool hive now 🙂

zealous maple
#

no idea. I know about ML in general, but not about SD

plain crane
serene depot
#

Hi guys, can you do video to video with 16gb ram 6gb vram?

bleak matrix
#

Good morning, everyone! How are we all today?

pastel maple
azure surge
#

Hi all

true field
#

would you guys suggest a 4070 12gb or a 4060 ti 16gb? at first i thought the more vram was better but now im doubting myself

#

with the sales going on the 4070 costs $100 more and has 25% less vram. and i figure in the future the models may want us to have more vram so even if 12gb is fine now, maybe in future 12gb will be considered a bit low.

#

bbut also i guess somehow 4070 processes images faster

#

thats what makes me think maybe im out of touch

lavish lake
true field
#

ah okay, and when you say ram, you mean the vram right?

#

also do you think i should get an i3 or an i5? will it make any difference at all?

lavish lake
#

don’t cheap out on the cpu, it will be your bottleneck down the road, i5 or a Ryzen 7 setup

#

And yes more vram better

true field
#

ah thank you! hmmm

fervent thunder
#

Is replicate the only place to run video diffusion

waxen geyser
#

Hey Does someone know if there are any siginificant downsides about the 128bit bus speed of the 4060 ti 16gb?
i could get this card really cheap today

hot plank
#

which ai can i use to generate images?

lavish lake
hot plank
#

ok

lavish lake
#

Just go there and prompt with slash dream

vital sphinx
#

Does anyone know why

  • fine tuning
  • Inpainting
  • Chat UI
  • Out painting
    is no longer included in Stable diffusion api?
full linden
#

Yo, I have a 4Gb rtx 3050 grahpics card on my laptop

#

is that good enough to run stable diffusion

full lark
#

I have a 1060 3gb and I can generate up to 1024x1024

full linden
#

But I am getting the OutofMemory error

full lark
#

You have to use --medvram command

#

Are you using automatic1111?

full linden
#

The resolution is not too high and the sample size is low too

full linden
full lark
#

Add this to your webui.bat @full linden set COMMANDLINE_ARGS= --xformers --medvram

full linden
#

Oh ok

#

I was a the text - no module 'xformers'. Processing without...
no module 'xformers'. Processing without...
No module 'xformers'. Proceeding without it.

#

Is this suuposed to be a fix for that?

full lark
#

Mmm that´s strange, idk why it doesn´t recognize xformers

fervent thunder
#

is there anywhere online to use stable diffusion video

#

that isnt really expensive

#

or takes eons to generate

gleaming ferry
#

hey guys. need some help. how do i keep sdxl from applying color filters on the finished product? it's a real bugger

rough oracle
#

I'm told that Stable Diffusion is the finest AI image generator on the web today? Or was that Dall-E 2?

full lark
#

The best right now its Dalle-3 @rough oracle the only bad thing is that is not local and censored (and doesn´t support things like loras), but it has the best results (for now)

rough oracle
full lark
#

Yeah no nsfw or guns, things like that, but all the other images are cool, it also allows some variety, like realism, cartoon or anime style, 2d, 3d,etc

rough oracle
#

And Stable Diffusion is basically the only one to allow NSFW? Also, what do I need to run Stable Diffusion on my computer? My potato was built in 2018.

full lark
#

What gpu do you have?

rough oracle
#

Device name DESKTOP-MA1DIB4
Processor Intel(R) Core(TM) i5-8500 CPU @ 3.00GHz 3.00 GHz
Installed RAM 16.0 GB
Device ID D487DEC4-AF79-4565-9D1A-45C301A66FA3
Product ID 00326-10000-00000-AA882
System type 64-bit operating system, x64-based processor
Pen and touch No pen or touch input is available for this display

#

So, yeah. That's my PC. If I try to run Stable Diffusion, will my house burn down?

full lark
#

Yeah it can run, you have an integrated gpu right? I mean no nvidia or amd one? @rough oracle

#

I have the same CPU and with the integrated gpu I could run it some months ago

rough oracle
full lark
rough oracle
#

A version of Stable Diffusion?

full lark
#

Yes

rough oracle
golden dust
#

Anyone know a cheap 12gb+ gpu ?

humble ravine
#

im testing a new model on nnear max settings

#

also ever ran an AMD x NIVIDIA dual GPU setup for AI

#

I did and boy its an freaking beast

golden dust
#

what da hell

#

how do you even

humble ravine
#

idk lenovo did it on my ideapad pro

golden dust
#

Does AMD work well with stable diffusion?

wheat hazel
#

I was thinking at something, maybe stupid... Regarding the pipeline: prompt->tokens->embedding->unet->etc... Correct me if I am wrong: the clipencoder and the unet have been trained together at the same time. So, when you do a merge on various checkpoints and then try to learn an embedding, it fails, because the mixed clipencoder and unet are no longer necessarily related to each other.
If that's right, do you think it could be possible to reestablish this relation? what would be the effect on the overall model?

river venture
#

I'm thinking of upgrading my PC. (assume NVIDIA)
Does SD work fine work fine under Win-11? Are there any bugs waiting to be ironed out?

forest trout
#

I have auto1111 running on windows 11 and it works for me.

#

Seems like it can handle the extra 11's...

river venture
fervent thunder
#

I have an amd gpu, is there a way to install stable diffusion on my PC?
I saw some tutorials on yt and google but is some of them worst than the other? which one is the latest/best

earnest prairie
#

Why do the bots put a video format at the end of my prompt when I don't want to ?

fallen mason
#

is wifi needed to use stable diffusion, to generate pics and stuff

silk gulch
#

how do I download sd 1.6. Also, why is no one talking about 1.6???

jade tangle
#

Excuse me, I'm wondering if it would be possible to request community members create AI modifications of existing images?

#

I'm an artist looking to use AI to enhance my work

#

But I'd rather not install Stable Diffusion to my own PC

#

Full credit will be given if anyone is happy to help me

trim magnet
#

how much in money 💸

jade tangle
#

How much would you be asking?

trim magnet
#

idk what kind of stuff u looking for,u mean like running your images thru ai to make them look better or using an existing video and turning it into anime or stuff like that?

jade tangle
#

Running my art through AI

trim magnet
#

but they are drawings or photos?

jade tangle
#

Kinda both

gaunt meadow
#

Getting back into the AI game a little bit

#

Because it's cold in here mostly

#

And I recently switched to a small capacity SSD instead of my big hard drive, and now I need to figure out how to store my models on a secondary drive because this 256gb HDD is not big enough for my OS, programs, and AI models lol

#

Is there a good way to put the models (and only the models) somewhere else on a1111?

#

I know you can on comfy but I'm too stupid to use comfy

teal flume
#

hi guys, should there be a better rundiffusion-com ?

pastel canopy
#

I recently trained a character LoRA based off of ~300 pictures with a diverse range of expressive facial expressions. These expressions weren't tagged. However, when I try to prompt any expressions, even expressions in the dataset, the effect is barely noticeable (e.g. "smile" only yields a small closed-mouth smile), and the face distorts so that it doesn't really look like said character smiling, but rather the character with a different person's smile seamlessly overlaid. Would reg images fix this, or is my LoRA just underbaked? Grasping at straws here, as I am a newb.

stoic apex
#

I wanted to talk to an expert about the training data for this type of application. Video that is. Possibly image as well. But video. and preferrably with audio (if it has been done yet).

I potentially have a great way of generating high quality "labeled/descriptive" data. Crowd sourced and verified by a very thorough system.

I wanted to know if my method was practical cost wise.

solid thicket
#

yo can i ask somone a favor rq?

#

anyone?

#

hello?

craggy dune
#

Hii

azure surge
#

Hi all

#

Does someone tried Stable Diffusion Locally with an Asus Rog Ally ?

#

If you answer to my question, The Rog Ally is a portable console using Windows 11. I didn't buy it yet but plan to do it and just curious if that work.

#

((Personally I didn't use Comfy UI dislike nodes system (giggles ))

plain crane
azure surge
plain crane
#

Lol his message was deleted for some reason

azure surge
#

ok

crisp sable
#

Hi, I am a beginner and I have a question. Is there a tool that would show training settings of loras? I launch the tool, drop a lora into it and the training settings are revealed. Or may be there is a plugin for Stabile-Diffusion-Webui? I know that I can view the settings by looking directly into file using a text viewer, but honestly it is a pain 😉
Although I was quite successful in training my loras in 1.5 (or hypernetworks, they were awesome, the best results), I am struggling with SDXL. The same dataset does not work - I train faces and in SDXL they are just hard to recognize, while in 1.5 they are perfect. I must be doing something wrong. I've tested about 20 settings sets for Kohya_ss and I have not found anything that would work.

random ingot
#

hi how do I edit prompt from an existing generated one?

wild iron
#

Can I add the bot to my own server and is unlimited?

warm junco
warm junco
wild iron
#

So it's possible 🤔

bleak matrix
#

Good morning, good morning to you! 🎶 How's everyone?

warm junco
visual burrow
#

can someone share an SVD workflow for comfy?

carmine whale
#

is there a way to forbid it from deciding to do videos?

torn juniper
#

Anyone know if there's a good (easy) way to merge a set of TI, prompts, neg prompts, etc into a model so they're no longer needed but get the same results? I currently have a template using specific prompts and negative prompts combined with negative textual inversion plus hires fix. This refined selection of things produces the images I want consistently but if I want to use something that, for example, doesn't use hires fix then the images don't look the same. So is there a way to merge everything into a new model or do I just have to try to train a new model with images produced with my current method?

fervent thunder
#

what could be the reason that a lora model is not visible in the webui?

full lark
#

I mean like <lora:"insert name that doesn´t show up":1>

fervent thunder
#

found the problem owo

#

thanks

gloomy pebble
#

Help! I am trying to create an image with six concentric circles on SD, but all my prompts have failed terribly. I want to use it to talk about varying gravitational pull in proportion to siz.
Any help will be very much appreciated.

warm edge
#

Guys, sorry if this isn't where I can post my question, if not, tell me which channel I can ask it. I'm still using SD 1.5 and controlnet was working fine, except that two days ago I updated "Git Bash" (because it always insisted that it had a newer version and I never clicked to update), when I did that the controlnet is there, but it no longer works. Does anyone know what happened?

#

I use AUTOMATIC1111- stable-diffusion-webui

fervent thunder
#

The #🔧｜finetune channel isn't very active; are there any dedicated servers to finetuning?

pulsar venture
#

hi, what is gfpganv1.4.pth?

abstract plank
#

Can I install stable Diffusion 1.5 on Intel Core i5 generation with graphics?

silent gazelle
#

⚰️

earnest mist
#

Guys how do I get the bot to just do a picture? Not video

nimble slate
#

Why the hell do my images go grey

#

whenever i try to load a prompt it has a glitchey green screen then it turns grey

hasty hornet
#

So...it's been some time since I used invoke web ui, opened it to play with out paint a bit, but for some reason it doesn't seem to even see most of my models and those that it can't see - it can't load anymore, even tho they were working properly on older version of invoke...what happenned?

I'll probably just use painthua instead, but this is just weird

nimble slate
#

this issue COSTED ME 3 HOURS 😭

#

i keep getting grey images im losing it

#

please SOMEONE help me!! 😭

hasty hornet
#

I have no idea, I'd joke about glitchy eye, but for the first time in my life I'm not gonna commit to a dumb joke

nimble slate
#

bruh like there is nothing on youtube that could help

hasty hornet
#

We can go through your settings and stuff but if image completely grey, that's ...idk

nimble slate
#

i tried videos and it didnt help

#

I TRIED EVERYTHING

#

i uninstalled and reinstalled

#

and used OTHER VERSIONS

#

and what did it get me

#

ZERO PROGRESS

hasty hornet
#

I'd try to check if model working properly...and vae too, try other things

nimble slate
#

i waisted 3 DAMN HOURS FIXING THIS CRAP

hasty hornet
#

go with simple propmt too, just to see if it works, no weights , nothing fancy, just few words

nimble slate
#

…

#

Oh my gooooOOOOOOOOD

#

PLEASE just give me something!

earnest mist
#

boys

#

how do i make an image instead of a video

strong folio
#

hi guys, I'm curious, and a little lost, in the Panorama tools, say I got the image wrapped but I'd like to show it off outside the browser, is there something special I have to do with the panoramic tools to export it so it is encapsulated and 'self-playable' for the other user? Thank you in advance

strong folio
#

are you running 3.12? perhaps?

#

also there are hardware and other software requirements for the video stuff

strong folio
# nimble slate i waisted 3 DAMN HOURS FIXING THIS CRAP

it's not crap if you are still on the learning curve, trick is go from base and do one thing at a time, and not pile it on, launch, test, if it doesn't work after confirming all the requirements are met, then mosdef reach out

strong folio
earnest mist
#

WHY DOES IT KEEP MAKING A FUCKING VIDEO

#

everyone in the chat with me is generating images and i cannot

#

cant be the model

robust fjord
#

Hi everyone, I would love to ask, if someone can help me with outpainting one image v.v

lavish lake
#

you just have to learn how to do it 😜

fervent thunder
#

is the bot different from sdxl 1.0

rich kestrel
#

That 5% though, ooh-la-la.

fervent thunder
#

hey guys what is your fav. model?

rich kestrel
fervent thunder
#

realvisxl2

sharp flame
#

how do u use the stable diffusion image maker thingy

#

ive not used it before

#

idk how it works

fervent thunder
#

Well there are lots of tutorials on Youtube.

#

I learned there.

merry escarp
#

Hey, does anybody know if it’s possible to use stability Ai to create a version of it where you can upload a picture of yourself or other people and in the output picture and for example everybody is wearing the same hat or jeans ? Sorry for my bad English. I’m not a native speaker

lavish lake
warm junco
warm junco
deep obsidian
deep obsidian
median torrent
forest reef
#

Does anybody know how to use my generated cartoon character in different pictures. For example this character is in the city , then it is somewhere in the forest and so on. Different pictures but the same character (not human)

jade citrus
#

i need help. some one help me for best SD colab notebook plss

somber jackal
#

Why don't I see the Create subarea button？

cloud gyro
#

yo guys im new here

#

hello

somber jackal
#

hello, im new too

lavish lake
forest reef
halcyon crown
#

Hello, i want to convert me and my cat pic into a cute cartoon style so which model should i use in img to img?

somber jackal
#

@forest reef i need it too, but it seems like it can't use seed here

silent flare
#

does anyone know how do i achieve Magnific AI using A1111 stable diffusion?

warm edge
somber jackal
#

does anyone know how to add a seed

#

ohh i find it

still gust
#

.

somber jackal
#

how to use the seed?

merry escarp
#

Hey, is it necessary to install the gRPC Client via Type Script or is there another way?

pulsar venture
#

how many vrgb i need to be high in sdxl?

lavish lake
#

depends on where are you working, here on the bots seeds are random, if you are on automatic1111 there's a seed parameter accesible and you can put any number there and have it fixed

#

no, they are randomized

limber birch
#

Can I use SDV on colab free version ?

lavish lake
floral umbra
#

Has there been any diffusers lately that is somewhat similar to luma? The text to 3d model generator. where it can make a rough sketch/model of how you want the model, then once you like a result, click on the model you like and it will then process a much mroe detailed version?

radiant grail
#

i need help building a pc and im currently thinking about using one of these gpu: rtx 3060 12gb / rx 7600 8gb

lavish lake
radiant grail
jade tangle
#

You can't add texture to existing images on any website can you

rough oracle
#

I'm on Black Magic. Is that a sort of extension of Stable Diffusion?

#

What's a tematic?

full lark
#

Wait wrong word, my english sucks ahah

unkempt grotto
#

Is there any way we can replace an object in an image with an image of another design object, for example I take a picture of a living room and then I select that I need to replace the Table.in it with the Image of the Table that I already have, Is there a way to make this possible, Thanks in advance

fossil phoenix
copper crystal
#

i googled creative society. radical anarchist movement that wants to abolish all power structures and private property

#

hard to decide whats more evil, the extreme left or the extreme right. both are so insidious

ornate bramble
#

anyone using SD with a mac

#

i currently have a mac m2

#

wondering if it would go well with SD

rancid belfry
#

Hey there, want to get into SD, may I request some guides/ youtubers that you might recommend checking out to learn it. Also is my old getx 1080 able to do this without melting?

warm junco
warm junco
#

It will work good with SD

rancid belfry
#

Thanks for the quick info on that

safe jewel
full lark
hasty hornet
#

my cooler is about to die tho thinking

weak tusk
jade tangle
#

Nothing about AI art

hasty hornet
#

That might be bad news for open source thinking

trim magnet
#

yea github shutting down tomorrow agony 😵 🗣️ 🗣️ 🗣️ 💸

hasty hornet
#

I meant open source Ai things, like a1111, oogoabooga, sd, whateverelse is there

#

we might not be able to do whatever we want anymore thinking

copper crystal
somber jackal
#

@weak tusk i heard AGI is almost here, is it ture?

copper crystal
#

copyright law is the framework that allows open source licences to exist the way they do in the first place. regulation isn't inherently a bad thing

copper crystal
somber jackal
#

@copper crystal how come?

copper crystal
#

there'll be generalized models that are good at doing one thing, but there isn't anything like general intelligence that shows any indication of being soon.

#

you throw gpt6 into a robot and put it in a survival scenario, it will die. it's just dumb is why. it doesn't understand things. current models are just prediction algorithms

#

some things. generalized models people can used for some things

somber jackal
#

thanks for replying, but just when ai can update by itself, it will be faster and faster i think

#

just need a short time it can do this

copper crystal
#

can it? is there any real model made developed with this biologic approach?

#

i'm pretty sure all DNN have training and inference seperately

hasty hornet
copper crystal
#

if we define general intelligence as human, agi is that. intelliigence that can develop artificial intelligence

#

literal deus ex machina (heh)

hasty hornet
#

well, in this case it can if we count simple ai programs that act by some kind of logic, but not by it's own will , by instruction*.

somber jackal
#

@hasty hornet even we don't know why we have self-consciousness, we also just can guess if the AI have

copper crystal
#

the team that are trying to recreatea fully simulated human brain may crack something accidentally. "that's odd" kind of moment

hasty hornet
#

Honestly I think we're kinda overestimate human brain in a way...
I feel like it's pretty similar to neural network AI as GPT's and stuff, just has ALOT more areas connected and interacting with each other, that had years, decades to learn and who knows how many generations to make some progress in the world

#

well + physical body, lots of organs...

copper crystal
#

it's only the most complex structure we've ever discovered in the universe

hasty hornet
#

I'm not saying GPT is as good, nah, it needs alot more things, good starting point would be long term memory and alot, ALOT higher token amount

copper crystal
#

DNN and BNN are fundamentally different though. What we deal with, convulution networks and such, are mathematical constructs and not intended to function in any kind of thought form. they're just layered networks and we're seeeing what happens with them

somber jackal
#

is here any extension using to change face of pitcure?

fresh stream
#

hi, how do i replicate multiple (3+) BREAK functions in comfyui?

#

theres a "combine conditioning" node but only allows for 2 inputs

#

i need like triple that

fervent thunder
#

My image isn't generating like wth

full lark
fresh stream
#

alright so i tried prompting BREAK in A1111 using a method I found on reddit for consistent outfits, turns out the gen is nothing like my prompt. can someone guide me how to use the BREAK function properly?

somber jackal
#

gus, i found the insightface. it can add to your server and replace the face of the person in the picture with the face you specify.

#

and another thing, i was banned from moonvalley

#

within a day of entering it

fossil phoenix
#

But I never have had it act any good way for me

simple comet
#

Why my generation is not working anymore ? He is taking the request but nothing happens

slate dust
#

Did you ask him nicely?

tough jasper
#

what is the easiest way to play with the new image2video capabilities?

humble ravine
#

I finally found a combination in invoke that beats leonardo's photoreal

languid helm
#

anybody know a prompt dataset to download for free ?

#

i would like to generate images from stories and the workflow i have thought is to put the prompts in a vector store and search for the nearest prompts thats similar to the sentences taken from the stories 🤔

simple comet
#

It worked

golden ridge
#

404 not found or no authorization. I cannot find anything about it needing fetched with the API

lavish lake
golden ridge
#

in my react app

lavish lake
golden ridge
#

Yeah. I tried at the tech-support channel, unfortunately no-one answered. I'll try again. thanks.

pastel pier
#

hello

jaunty osprey
#

sup. Is there a way to generate each variable on a wildcard TXT ? instead of randomly picking one of these? Maybe wildcards is not what i need to use then, but i wouldn't know another way to generate a bunch of parts of the prompt

#

btw i mean generating an image for each possible entry on the wildcard. Like.. automatically "ok there's 13 variables for the clothes on this portrait of a woman... generating a batch of 13 images" Not combining or randomizing anything

fossil phoenix
#

yes

#

well. sort of,

#

maybe not directly? but if you sort each prompt var, and use dynamic prompts, you can say for each {1|2|3|4|5|6|7|8|9|10} and {11|12|13) and turn on combinatorial generation

#

you will get prompts from this.

so like {red|white|streaked} {hair|clothes|sky} will run combinatorially as prompts: red hair, red clothes, red sky, white hair, white clothes, white sky, streaked hair... and so on, each set of those a prompt.

high sun
#

hey guys, which model is used to turn real people into sculptures ? just saw that on instagram and it looks pretty cool

jaunty osprey
storm hare
orchid jungle
#

Hi! Would anyone happen to know of a model that uses a Chinese ink art style?

storm hare
#

I think your best bet would be to use a word processor to write the Chinese characters and then run that image through stable diffusion

orchid jungle
storm hare
orchid jungle
#

I generated those with Dalle-3 and I’m trying to find a model that would replicate those the closest

#

Though the one you linked is pretty close

storm hare
peak tendon
#

anybody got a tutorial to train a character model? I don't mean text, I mean feeding it renders of a character so I can get him in my text to image stuff by mentioning him for example

#

I use stable diffusion

#

installed localy

fervent thunder
#

Does anyone have a model to make royal pet portraits?

quick marsh
#

idk

#

i just know i can make images and stuff

golden marsh
#

Hi anyone know with ComyUI what this means? "Requesting to load BaseModel"

I'm using an Img2Img Workflow and DreamshaperV8 as the model but it keeps doing the same thing (see above bolded text) each time I Queue a prompt.

Each instance takes ~300s to finish which is much longer than normal.

quick marsh
#

uh

#

im not a dev, plus, idk how to even code

#

i think that's a error

orchid jungle
cosmic oyster
#

did you find anything? im also curious about this 😊

normal lance
#

where i can find stable diffusion nsfw?

#

@flint harbor maybe?

#

or @still glacier ?

#

someone send me link here and said that i could find it here

flint harbor
#

Rule #4
No NSFW Content
Excessive gore/blood/nudity or anything that exists to suggest sex including sexualized poses. No illicit imagery. This includes NSFW content involving or depicting underage characters such as schoolgirls, lolicon, etc.

normal lance
#

so, where i could find it?

#

do you know?

still glacier
#

sure let's discuss the thing we're not allowed to discuss

normal lance
#

oh true, sorry

#

nevermind then

still glacier
#

There is no nsfw filter on stable-diffusion itself but we try to keep this discord in compliance with rules.

normal lance
#

alr thx for info and sorry if i ask smth that i shouldn't

tribal crown
#

Hello there. Can anyone help real quick? My images do not look good at all. I am new at stable diffusion so please bear with me. I dabbled in it on TensorArt and SeaArt and I find it amazing. However I want to try the local Stable Diffusion via Automatic1111. I am always using the NeverEnding Dream model and suffice it to say that my generated images look like sad waterpaints compared to amazing images I create on SeaArt or TensorArt. What am I doing wrong here?

Thanks. Will post comparison with the same prompt if anything is unclear.

ornate bramble
#

anyone here using SDXL with a mac m2 base model?

warm junco
strong folio
#

I've a friend who tried to install SD, they have, "Nvidia GTX 1660" but I think they ran into trouble is there a metric too look up what is what that can run SD?

warm junco
strong folio
warm junco
strong folio
warm junco
strong folio
strong folio
#

I've shared the link you've sent, so they should be good when they wake and perky

warm junco
wheat hazel
#

any opinion on comfyui?

fossil phoenix
#

in a1111 you can use [prompt|prompt] to alternate between two between steps. but what happens if you insert a lora within those [] does it alternate between the two? or do both loras load and only the prompts with them alternate eg. [lora:A:0.5, keywordA|lora:B:0.5,keywordB], prompt.....

#

not sure how it works under the hood.

trim magnet
marsh echo
#

anyone know how to make it so the ai doesnt turn your prompt into a video?

alpine hearth
unkempt mica
#

hey i have kind of a simple question but idk how hard it would be to actually do. i got a really perfect image generated but now im looking to get the same image but with all the streetlights and everything turned off. is that possible?

storm hare
ivory nexus
#

Hello!

clear osprey
#

i'm going to install stable diffusion for a friend, he has a rtx 4060 (8GB of vram) should i put --medvram in his webui-user.bat?

slate hare
#

hello

warm junco
bleak matrix
#

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

remote pelican
#

when I type in a prompt it automatically adds "format:Video" at the end, how do I remove that ?

craggy dune
#

Hiii

thorny frost
#

Hello! When trying prompts, is there any way to save them (and other settings) for later within A1111?

hearty bronze
#

/create

foggy night
#

how can I download images from bots, please help.

warm current
#

Can you do image to video using the bots on this server?

small birch
#

Can anyone help me with this: Error loading embedding vaeFtMse840000Ema_v100.pt
Traceback (most recent call last):
File "C:\AI\stable-diffusion-webui\modules\textual_inversion\textual_inversion.py", line 236, in load_from_dir
self.load_from_file(fullfn, fn)
File "C:\AI\stable-diffusion-webui\modules\textual_inversion\textual_inversion.py", line 208, in load_from_file
raise Exception(f"Couldn't identify {filename} as neither textual inversion embedding nor diffuser concept.")
Exception: Couldn't identify vaeFtMse840000Ema_v100.pt as neither textual inversion embedding nor diffuser concept.

fickle holly
#

Hi all, does anyone know please. SVD, does this work with MAC?

vestal dew
#

OK, what am I missing? I'm trying to change only the hair color in an image. Inpainting - it either does nothing or changes the hair style.

shell venture
#

I tried using batch but it's just no consitant

#

And when I use animatediff it just doesn't work, I get a broken image everytime.

full lark
#

Today stability will release a new model right? 👀

finite cloak
fervent thunder
#

how do you use stable video diffusion to generate 360 views

#

is there a demo

deft locust
#

I have an important question since i cannot really find an answer anywhere else.

#

I have somehow subrscribed to "Black Magic" via the Stable Diffusion Webpage but what is it exactly?

#

I feel like i am being scammed when it comes to that.

#

It looks similar to Stable Diffusion but has none of the other functions except Text to Image

full lark
deft locust
#

yes blackmagic.cc but Stable Diffusion send me there 3x before i did the subscription so i thought it was renamed or something

finite cloak
#

There are websites that label themselves "Stable Diffusion" that have no relation to Stability AI

deft locust
#

crap

#

So i got dubbed

#

It was just odd, they looked exactly like in the videos for Stable Diffusion

full lark
#

I just searched "stable diffusion" on google and the first webpage is a scam, it leads to blackmagic.cc

celest dune
#

Do I really need to write an email to stabilityAI to enquire about the price for selfhosting SDXL-turbo for commerical use?

#

I'm just interested to know a rough estimate so that I know if it's worth pursuing

iron pendant
celest dune
#

$100 per month isn't much, stuff is worth pursuing

iron pendant
#

the issue with buying a commercial license will be around what expectations there will be of future competition and further changes in terms

#

they could decide to change terms in another year and stomp your business into dust, as is with anything that is a ongoing cost, the cost or terms can change

#

maybe they decide to step into business in a way that overlaps you, and they want to capture that market

frigid wolf
spring notch
#

odd that they specifically mentioned indie devs. the average indie game makes $15000

radiant grail
#

anyone making banks $$$ with stable diffusion?

iron pendant
frigid wolf
#

I love the original vision of stability, being a Benefit Corp, funding research, giving researchers a way to be able to publish their models openly to the community. It's what makes stability different, it's what makes it special. I hope they reconsider this direction.

frigid wolf
next leaf
#

я предоставляю мощные сервера для любых задач.
если хочешь испытать Stable - на мощных настройках
то обращайся - предоставляю выделенные сервера свои личные по часовой оплате. обращаться в лс если надо!

frigid wolf
#

I’m at the beginning of my career as a researcher. I’m just entering my PhD, my first journal article just passed peer review and is about to be published. Hearing Emad speak early on about stabilities vision really resonated with me. It’s something easy to get excited and be passionate about. I don’t want them to lose that.

finite cloak
#

to make sure it's clear re the above:

#

Under that plan, if you're not using SD commercially to make money, SD remains free for you as always

hasty hornet
finite cloak
#

we were going to but between when i said that and now we accidentally dropped another model woopsie

frigid wolf
iron pendant
#

"we see you are successful using stable diffusion xyz model, thank you for discovering that market, we will take it now, as our terms have changed to disallow that and/or we are now charging x% of your revenue to do that" etc that's really my concern if investing in a particular set of software or models

#

there are a lot of folks making small change off patreon subs and such and that probably kills them all off

#

I definitely get it from SAI's perspective, in many hussle bros out there that were farming them for free tech

frigid wolf
iron pendant
#

is it now against TOS for huggingface to host demos? they charge $1-3/hour for T4/A10G instances for demos, that's a for profit venture

#

I guess I would presume huggingface needs to pay for a commercial license else anyone who might put up a research project on HF as a demo and pays for that, not that the researcher is for profit, but huggingface is

languid minnow
#

Anywhere I can go in here for help with a --skip-torch-cuda-test error

bright nest
#

how do i remove this " WARNING:accelerate.big_modeling:You shouldn't move a model when it is dispatched on multiple devices."
i have a brand new comfui install.. :/

#

i never moved a model

iron pendant
#

perhaps SAI can consider similar terms to Unity engine etc, i.e. revenue or funding limit so the little guys don't get smashed

slim raptor
#

LCM lora is nice, i can generate higher res pic with the same time or less

thick vale
#

if u guys create a picture in txt2img, u start with 512x512 then after few time u like a picture but its in 512x512 but u like to have ik hd, what do u guys do upscale it in extras or go to img2img and upscale it there?

vapid tangle
gleaming mica
#

Probably a better suited question for ComfyUI discord but is there a way to get an interface with Comfy That is simply my prompt and the output? I remote in to my machine from a laptop and it's hard to nav around Comfy

#

(actually not sure there is a comfy discord)

thorny frost
visual pier
violet latch
#

does anyone know about Segmentation ADE20K color dataset?i found werid thing,all ADE20K color dataset PDF or image all displayed SKYSCRAPER and ROAD the same color (140,140,140)or #8C8C8C, all 150 objects in ADE20K list has only those two objects use the same color, is this correct?or the first person who sorted the table made the mistake?

thick vale
visual pier
thick vale
#

ah thank you, i figger out if i keep it low picture woulds stay the same more,

thick vale
visual pier
#

Oftentimes I move between 0.46 and 0.44 even. Thats how much of a difference it makes :D

thick vale
#

yh

thick vale
visual pier
#

No just upscaling by 2

finite ravine
#

is it possible to combine depth controlnet with the sdxl inpainting model?

deft locust
#

Can someone please share a Link to the right website? I think i got the wrong website again for the second time and i can't find a reliable Link here on the Discord channel.

full lark
#

This is the original from stability

deft locust
#

Thank you very much!

full lark
#

Np!, I still dont know why the first website that appears is a scam and nobody from stability reported that

deft locust
#

What's worse is at the time i tried generating pictures with the scammed AI program i was far not the only one using it.

full lark
#

I think it still works because it connects to the api, but its still sus

bitter laurel
#

hey all! I can't seem to find this for the life of me!! I just need to convert a string to conditioning in comfy.

iron pendant
bitter laurel
#

oh they had a proper emoji for it

#

lol that took me much longer than i intended. THANK YOU!!!

hollow dust
#

Is this the chat to ask about LORA creating quesions?

gleaming mica
#

Are there any readings on how Stability.ai envisions SD "activating humanity's potential"?

desert badge
#

anyone have any tips on why my LoRA isn't showing up in SD webui? I have the correct base model selected to apply the LoRA to but nothing 🤨

#

(lora trained on sdxl using dreambooth)

thorny frost
marble knot
#

My graphic card can't handle the kind of things i want to achieve with AI
Is there a way for me to rent calculation power for IA without having to sell one of my kidneys?

desert badge
#

you tried colab?

#

there's PAYG and different plans available

marble knot
#

looks expensive, i dunno what they mean by comput (but when i spent my free credits there were gone really super fast) unit but yea thats probably the best bet

desert badge
#

yeah i'm not sure what the compute units equate to either tbh

warm junco
marble knot
#

RunDiffusion seems to offer great plans, im looking into that

warm junco
marble knot
#

set COMMANDLINE_ARGS=--xformers --medvram --no-half-vae

warm junco
marble knot
#

I don't do normal generations, just SDXL but that could help some day

#

thanks anyway, and yea there is lot of alternative on the web

#

like RunDiffusion or others so its all good

leaden sluice
#

Hi all
no one can help with a small animation in deforum or just explain a little how to make it better and more beautiful
I will be very grateful if someone can help me in text form

or at least tell me where I could write so that they can help me

vagrant depot
#

Heey!, I'm kinda new using stable difussion web ui, and I'm trying to use my model that I trained with 1.5 sd version with some other checkpoints that I found online, but It doesn't seem to be working as I expected, Is it posible calling promps from other checkpoints just texting it? or I must merge checkpoints and then type the promp to get the style that I want?

fossil phoenix
#

I know it may sound silly, but is it in the right model directory, and is the extension properly named, and did you hit refresh for the loras? Just asking , cos for the last one, my loras tab, there's two refresh buttons, the little blue one, that does something... but not refresh it, and the one literally says "refresh" which does refresh properly lol

#

and I always get stuck wondering, why is it not here... and oh, i hit the wrong refresh 😐

desert badge
#

maybe it's something to do with the metadata?

fossil phoenix
#

does the wrong model see it? lol

#

but yeah probably metadata, not sure how 1111 discriminates

#

hey, what if you just call it, not from the click and add, but just type it in?

#

does it work?

#

again not sure if a1111 respects <> if not showing in the list.

desert badge
#

worth a try!

fossil phoenix
#

if it loads on the backend, but not displaying for some goofy reason it may

desert badge
#

hmm, when i try that it doesn't take effect

#

but in the logs i do get
Calculating sha256 for (my_lora_path)

#

however it doesn't seem to be doing that when i trigger other loras

vagrant depot
#

Why some Lora models doesn't work on sd web ui ?

formal stratus
#

Anyone of you guys able to make the color specific to hex code?

rare grove
#

need smart person in text support chat

#

My additional networks it not functioning properly

minor harbor
#

Hey im trying to install the reactor extention but keep getting this error. import insightface
ModuleNotFoundError: No module named 'insightface'. anyone know how i remedy this?

orchid parcel
#

Hi,

#

anybody having a simirar error on ComfyUI?
TypeError: forward_timestep_embed() got an unexpected keyword argument 'time_context'

teal hazel
grand dust
desert badge
#

what's the workflow for upscaling images with webui, is it img2img? Do I need a different model?

woeful anchor
desert badge
#

🔫 👽 now answer

woeful anchor
#

its probably img2img

#

i hope

#

i tihnk

#

i dont relaly know tho

desert badge
#

wait lol there's a SD upscale script in the img2img section

regal jungle
#

Hello, just started using Stable Diffusion and trying to create consistent characters. I have a close up face picture and am trying to put that same face on a full body picture using img2img, but it keeps giving me more portrait photos. Among my other prompts used to create the original face, I have "full body image, cute black outfit, streets of brooklyn" Any recommendations?

woeful anchor
regal jungle
#

i have no idea what that means tbh

woeful anchor
#

just put it in

#

the ai will understand

#

its like a secret terminolgolically termial factor

#

yk some high level shit

#

big brain shit

regal jungle
#

gotcha gotcha

woeful anchor
steep canyon
#

Why is Stable Audio not open source? Is Stability starting to pull an "OpenAI" and just release minor models as open source sometimes?

prime loom
#

Hey all, hope you are well.

I would appreciate some help here.

How do I use stable diffusion for image-to-image translation? Also how can I fine-tune this for my very specific use case?

grand dust
#

there are many yt tutorials for that

#

(sry for the not rly useful response)

rare cipher
#

Should I be able to train LORAs with Kohya against non-base models? Example, using CapabilityXL as the model instead of base XL. Every time I try this the results are overwhelmingly cartoonish. With base XL the results are quite decent, but then when I gen the LORA using other models the results are meh.

#

My LORA I trained against CapabilityXL looks great in gens at the 50% mark and then by the end is a weird grinch oversaturated mess. Ideas?

prime loom
rare cipher
#

Or asking simply how to use img2img?

prime loom
#

Image to image translation in my case is changing the background of my image

rare cipher
#

If you just want to change the background, what you want is inpainting.

#

If you can be specific about where you're stuck I will offer what advice I can.

prime loom
#

What's in painting?

pastel lynx
#

How to lock/unlock nodes in ComfyUI ?

rare cipher
karmic granite
#

For instance when you train using anime its best to use the Nai model because all the anime models stem from it basically

rare cipher
#

Right on, thanks. Doing some extensive testing with one of each and sure seems that way. Kinda sucks that different models will handle my LORAs soooo differently. I get why, was just hoping I could sorta "cut through" so to speak. Thank you!

karmic granite
#

Better question tho. Is the stability foundation ever going to release its own official training tool? Because im absolutely over Kohya breaking every other month lol

rare cipher
#

100% with you on the Kohya bit. It's very fragile. And weird bugs. IE, all my directories are on a network share. That's perfectly fine for all but logging. If I choose a logging dir that's on a network drive, Kohya crashes. All other network directories are fine. So random lol.

vagrant depot
#

Why can't I load some Lora models ?, I put them in the right fodler but sd web ui only recognize one :c

prime loom
solar minnow
#

How can I boost using more VRAM and GPU in SD ?

bleak drift
#

What are yall’s opinions on using stable diffusion to remove certain things from images of its for private use?

karmic granite
karmic granite
#

thats even optimized for SDXL if you use it.

karmic granite
wraith iron
#

hey guys, just entered the dc, I watched a recent video on Stable Diffusion and how to use it with fooocus, I have low Vram and I was wandering if there is a way to use my normal ram instead of the graphics card one, Tell me if i should be asking this somewhere else

#

when Im generating an image it gives off a error about ram and I think its because of that

sand holly
#

Shouldn't it in theory be possible to generate a bunch of models with random weights and eventually one will be good enough while not resembling any artstyle that exists in reality?

#

in theory it could also be much faster than training as you can get a perfect model on first try

sand holly
sand holly
wraith iron
#

I just ran it again to get the error message and suprisingly it just worked for the first time

#

but just so i know , how do I make it use the cpu instead?

sand holly
#

i wouldnt recommend it but you would have to watch youtube videos on that one, i really dont know how to do that

#

by the way what gpu you got?

wraith iron
#

3050

#

but its laptop

#

now the generation works wonderfully but the image prompt dosent, when i give him a image the error is always the same "connection errored out "

#

and on the cmd it just says "paused"

fervent thunder
#

is there a way to continue videos

#

like get 5 to 10 second videos

quick marsh
#

when the

#

this is my hand --->diffusionhand

pseudo coyote
#

Hey guys, does anyone know how to update fooocus, I think there is an update but it’s been a while since I used it

surreal perchBOT
#
FAQ: Why are my images blurry?

In order to ensure a safe experience, the DreamStudio website has a NSFW classifier that will detect and blur any potential NSFW images. While in most cases the classifier will appropriately identify NSFW images, there may be occasional false positives due to the nature of how these systems work. We will continue to work on and improve the classifier to make false positives less and less likely! You are not charged for any images that are blurred

#
FAQ: What is SDXL 0.9?

SDXL 0.9 represents our research version of the model, soon to be followed by SDXL 1.0, which will be released as open-source. SDXL 0.9 showcases a remarkable advancement in image and composition detail compared to its predecessor models. With an impressive parameter count, SDXL 0.9 stands among the models with the highest number of parameters available of any open-source models, boasting a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline. To achieve exceptional image coherence at an higher native resolution of 1024x1024, SDXL 0.9 leverages the power of two CLIP models, including OpenCLIP ViT-G/14, one of the largest OpenCLIP models trained to date.

coarse hare
#

anyone here looking for Website Dev for your launches? Price is very negotiable!

exotic violet
#

如何生成动画

#

How to generate animations

sand holly
#

if it still exists

hallow plaza
#

What image to video solutions are there?

cerulean bane
#

Is 1.5 getting a turbo version?

latent flower
#

dalle-3-trained-model when

knotty fog
stray trout
oak violet
#

Back in my day you had to colour the pixels by hand, manually addressing them to a specified colour.

stray trout
fervent thunder
#

May someone point me to a good source to learn stable diffusion as a beginner? Should I use a UI or install it locally? Thank you, very appreciate

primal shuttle
thorny frost
fervent thunder
#

Thank you guys!

verbal osprey
#

Hey guys I need some major help here. I'm trying to set up comfyui. I've been using automatic1111.

I'm setting up the xtra_model_paths.yaml file. Do I need to change ONLY the base path for automatic1111?

Also, do I use C:\Stable-Diffusion\stable-diffusion-webui\ or C:\Stable-Diffusion\stable-diffusion-webui\webui\

And what about the backslashes vs forward slashes. The rest of the document uses forward slashes...but typically windows uses backslashes.

#

Would this be correct?

#config for a1111 ui
#all you have to do is change the base_path to where yours is installed
a111:
base_path: C:\Stable-Diffusion\stable-diffusion-webui\

checkpoints: models/Stable-diffusion
configs: models/Stable-diffusion
vae: models/VAE
loras: |
     models/Lora
     models/LyCORIS
upscale_models: |
              models/ESRGAN
              models/RealESRGAN
              models/SwinIR
embeddings: embeddings
hypernetworks: models/hypernetworks
controlnet: models/ControlNet
verbal osprey
muted peak
#

Why does the discord SDXL is better than the base SDXL available to public?

shell rose
#

Greetings, i am new to SD und i created a image with 904*512 - but i want it to be in a 21:9 format (at the End 3440x1440)... but when i change the format of the image, a hole new image appears... is there a way to get the same image but in 21:9 ?

warm junco
#

Your path is correct but you may need to switch the slashes

fiery wasp
#

I think we need a separate section for LCM since that one can be a very tricky beast. I have no idea where to suggest that and who can make it happen.

vast skiff
#

hi

latent flower
#

👋

#

legit dalle 3 model when

#

jk

#

ime dalle 3 is soo gud

#

back to sdxl im like

#

😐

#

maybe its the prompt completion using gpt? dunno

bleak matrix
#

Good morning, everyone! How are we all today?

latent flower
#

ops

#

sorry

#

I forgot

bleak matrix
#

It's all good. What's up?

latent flower
#

good morning! Im great and rn im playing aroud with dalle 3!! Im testing prompts and checking the differences from local SD... Wbu?

#

btw, its been a while since I join you guys here... work is being extra tuff... but I quess im back now with a new gpu 🙂

bleak matrix
#

I'm here most mornings! 🙂

latent flower
#

I see! Its cool to see ppl I know here again, like CS1o... I remember when they just joined lol

latent flower
solar minnow
#

Everyone , which is the best models and VAE for anime girl now ?

bleak matrix
latent flower
fiery wasp
sturdy cape
#

Any good generators for text to stl for 3d printing some novel Christmas present shapes?

opaque wharf
#

Is it possible to add this bot to the server with my friends ?

sudden ruin
sudden ruin
#

But no official release

raven tendon
#

hi guys how i make the img2img feature ?

thorny frost
floral umbra
#

Hoi, do you guys kow of a good open source text/image to 3d on git?

I really like the results of luma, but it's not public/FOSS sadly

bleak matrix
# sturdy cape Any good generators for text to stl for 3d printing some novel Christmas present...

I don't have anything on stl to 3d printing, but if you're interested in learning about 3D modeling, there's a lot of super easy ways to user Blender+SD to make simple shapes. If you check out my tutorials here: https://www.youtube.com/@atypicalconsortium There's quite a few on Blender+SD. There's a tool in Blender (the cut tool) where if you bring a plane into Blender (images as planes) and simply cut the image, and extrude it, you can make some cool little 3D things without a significant amount of effort.

floral umbra
bleak matrix
floral umbra
sturdy cape
#

I'm looking to make the shape, export as stl. From that I would go with the usual slicing software like Cura or Flashpoint. (Though some better support structure generation and outcome simulation modeling would be interesting.)

bleak matrix
floral umbra
bleak matrix
floral umbra
#

Aye, but i ain't got 40 a month for it :P

bleak matrix
#

(Hence Sculptris having it.)

floral umbra
#

Ohhhh. So sculptris is the free version of zbrush Thunk

sturdy cape
bleak matrix
floral umbra
bleak matrix
floral umbra
#

Oh by the way, as we have a few blender users in here, do you know of a model for SD that is great for generating textures? As i wanna make VDM's for a model i'm designing, but wanna use textures as a starting point :P

floral umbra
bleak matrix
bleak matrix
#

When I have some time, I'll get to that!

bleak matrix
sturdy cape
sturdy cape
# bleak matrix What specific shape?

Specifically I want to make a Christmas gift of two dragons forming a heart shape. I've tried some shape searching though thinking more I might be able to multi load some models to compose & assemble without 3d modeling in detail.

thorny frost
sturdy cape
sand holly
#

Does anybody know if it's worth more to use a 13B LLM model with a higher wbit number rather than going for a 34B model with a lower wbit number (8 wbit vs 3 wbit)?

bleak maple
#

Hi

#

I have a doubt

#

What exactly happens when we vote an image on this server, like does it use it to train something?

restive stratus
#

Is there an appropriate place to place requests/commissions? I have an image I have made that I just want to have a simple cohesive vector/cell shaded style applied to that won't obliterate details. I do not currently have SD set up on my computer, and don't have time for the next couple days, and need to send the image out to get a custom puzzle made before christmas.

empty fox
#

Seems like the Tile Texture style does excellent aligning horizontally but not quite there vertically.

gaunt bridge
#

What's the best resource for learning comfyui fully? e.g. how the nodes are supposed to fit together, etc

lavish current
#

I still haven't figured out why anti-AI people come to AI servers

#

Truly baffling

azure nebula
#

hey suddenly my stable diffusions stopped running

#

ok i t started running that was weird

reef hull
#

Does someone here work with IPAdapters in ComfyUI?
I have a strange Issue the nodes don't finde the models

wary hazel
#

Does anyone know how to covert ckpt to safetensors please

#

I see the link to download safe-and-stable but how do I pull it up for it to work

rotund forge
#

In the GUI it's under Utilities (tab top), then "Convert Model".

fervent thunder
#

is there a way to prevent stable video diffusion from getting so grainy

white kraken
#

hi guys!

#

I'm trying to get the new Turbo model to work, but I'm not getting the option for the TurboTimer in ComfyUI. Does anyone know what I could do? ❤️

bronze harness
#

Hey folks! new the server, looking for some help

#

got automatic1111 to do auto update today. after the update, it starts looking weird

#

not sure if I have permission to post picture or not, but basically the button where it used to be "Inpaint masked" has became "Inpaint masked, Inpaint masked", and when selected, return errors ValueError: ['Inpaint not masked', 'Inpaint not masked'] is not in list

fervent thunder
#

Is it possible to run stable diffusion locally with old GPUs like 2x R9 280X or should I use a cloud solution? I also could use a Mac Book M1 Air. Thanks

warm junco
warm junco
bronze harness
#

thank you, will try that

warm junco
plucky skiff
#

Is there a ComfyUI node that allows me to render an Output video of the Denoising stage, you know the Preview you see on the KSamplers at the Denoising stage if you have Preview method set to Auto or TAESD is there a way to output that as a video?

hollow dawn
#

I am using comfyui. But I am trying to use a checkpoint with a built in VAE. What node should I connect to ksampler in that case. The default has the VAE node. Any help?

gusty cradle
#

hello there! just want to make it sure after spending a lot of time trying to figuring out... Animatediff SDXL motion module doesn't work on A1111 right?

bleak matrix
#

Good morning, everyone! How are we all today?

rose hollow
#

yo

#

can u tell me , how do i use stability video generation

forest apex
#

Man this is astounding

rose hollow
#

huh

forest apex
#

Watching my front webcam through img2img almost realtime on my potato

#

@rose hollow on your computer?

rose hollow
forest apex
#

Or online, not sure it's offered

rose hollow
#

is clipdrop paid now ?

#

could you tell me how to use that hugging face webiste

forest apex
#

Perhaps, use it for what? I only download models from it

rose hollow
#

i am new , but i know basics

forest apex
#

You'd want the cloud spaces, have you found any? You pay monthly but not sure about credits/limits

rose hollow
#

i want everyhting free

#

if i had money , i wouldnt be here

forest apex
#

Ha. Sure, collect and repair old hardware, but you'd still have to conjure a graphics card

rose hollow
#

i am talkinga about that

#

i am talking about creations

forest apex
#

To generate video you either have to build or acquire a moderately powerful computer with adequate graphics

#

Or, pay for services online. Some are cheap, others are easier to use.

rose hollow
#

lenovo lappy , specifically for coding and video edtiing

#

8 gb ram , what else, to create an image

#

?

forest apex
#

8G is good, what model of GPU?

rose hollow
#

amd ryzen 3 and amd radeon

#

could u tell my anything that would be helpful for me

#

is stability AI paid now

#

?

#

?

forest apex
#

Not sure.. playgroundai is free for images. You can run most stuff depending on how technical you wanna get

#

There are easy installers for things like automatic 1111 or comfyUI, but just gotta listen here, reddit, hackernews, it's changing pretty fast

rose hollow
warm junco
rose hollow
#

gen 3 ig

warm junco
#

you can check it in the taskmanager under the performance tab

rose hollow
#

yeah , so how do i use these AI,s

#

its very difficult

warm junco
# rose hollow its very difficult

you need to follow this install guide:
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs
but before make a new folder on a Drive and not on desktop, downloads, documents. Call it Ai.
Then go into that folder and click into the file explorer bar (not searchbar). Type cmd and hit enter.
then paste the git clone command from the website in it and hit enter.
Then launch the webui-user.bat
if you have any questions about it feel free to ask in #🤝｜tech-support

rose hollow
#

?

#

not now ofc

#

u give me your time

warm junco
rose hollow
#

how do i style these images

warm junco
rose hollow
#

listen , tell me how do i precisely communicate with the AI

#

whats the advice

warm junco
#

where?

#

you first need access to an ai

rose hollow
#

can you give some samples of your generations

#

here now

warm junco
rose hollow
warm junco
#

in this channel here you cant post images

rose hollow
#

could you send me some images that u created

#

why arent you guys straight forward in talking

#

its annoying

warm junco
#

your pretty direcly unfriendly with your messages like "here now" and "listen"

#

so please change your writing style or i stop helping you

grand pasture
rose hollow
#

i want to see, what do you type and what comes out

#

just answer , so i can go or stay

warm junco
#

i can make some quick image if you want, but i dont get what your after

rose hollow
#

forget it man , i am stunned

#

i will actually die out of annoyinngness if i find people like you in real life

#

u have too much patience

#

just for your information , u havent helped me a little in this long conversation

#

i am gonna watch big brother now

warm junco
#

take it or leave it. The bots like #1100170312106127410 are free to use. lots of examples in the channels here too.

fervent thunder
#

Which stable diffusion model should I download and why? 🙏

outer crescent
vapid bone
#

Hey peeps, just started experimenting with ComfyUI and want to check out img2vid. In my Image only checkpoint loader... the ckpt_name shows "undefined" and I can select any of the models I have installed. What is the directory I should have these in?

lusty beacon
#

why are people messed up in turbo

tender cove
#

@frigid swan here's a good place too

frigid swan
#

I have way too many kudos, so here's a shared API key with 5 million of them free, to generate using stable diffusion on a voluntary distributed network of GPUs. It's a free service, but kudos speeds up generation when demand is high and supply is low (which is always)

fossil phoenix
#

But then I ask it this

olive arch
#

Anyone Have Youtube Channel. U can contact with me. Good work available

unborn pumice
#

whoa just got stable diffusion running and the first image i generated had some of the scariest looking humans ive ever seen! I mean nightmare fuel lol. i got some learning to do.

echo kite
#

any good non-sdxl models for non-portrait photorealism?

#

a lot of photorealistic models are good but they are mostly for generating photos of faces

#

looking for more stuff that can make stuff like cats browsing the internet etc

fossil phoenix
#

So I was playing with suno.ai and I guess they're using a multi-model generative workflow clearly NLP lyric generation, the little square image they give to each song like a CD cover or something is image generation, probably based on a prompt generated from keywords parsed from the lyrics and music style prompt, the music output of course is generated from text to music of some sort, and I'm not sure if they use some sort of non-AI distillation of the music into cadence and key for the text to speech of the lyrics. I know the lyrics are using keywords such as verse and porous in Square brackets, and the music itself is partitioned into a intro, verse, chorus, and sometimes ending segment, but they only last 1 minute and 20 seconds so often they end right after the last word of the chorus. Maybe if I included an outro position in brackets. But I'm not trying to recreate that. Rather perhaps train a model on audio using groups of open pose annotated with style of music that the dance is part of, and then annotate the individual shifts in motion within the open pose segments, to key in on the beat portion of the music or some sort of cadence tagging. I wonder if you could extend something like the forum or anime diff using a model like that to create music videos from an audio, pre-existing or generated

daring dew
#

Hi Every One I am new to this discord, and this is my first time using it, one of my friends told me about it that I need to join a discord about Stable diffusion if want to know how to generate AI Images. Can someone guide me how to do it here. Thank you if you can help me 🙂

fervent thunder
#

Hi, I am a huge AI fan (you can probably tell from my name 😄). I discovered this shop where there are very cool calendars that were designed with AI. Is the shop trustworthy? I mean, they accept PayPal, so nothing can go wrong, right? https://calendardreams.com/

unkempt hatch
#

anyone using an eGPU?

green radish
#

Hi, I want to try to create fashionweek ai, someone try it before, please help with promt how to create base on fashion drawing design

jade tangle
#

What is "no module 'xformers' proceeding without"?

warm junco
jade tangle
#

Ooh

warm junco
#

What's your GPU?

#

You can add --xformers to the webui-user.bat at the line COMMANDLINE_ARGS=

warm junco
primal dirge
#

So does anyone know how to enable loras and embeddings in comfy UI?

#

Also are negative prompts required for comfy UI?

jovial chasm
#

for some reason imgtoimg alternative test will not load for me does anyone know how to fix this

#

it has an error when i try to use it

#

and does anyone have a good iypnb file for sdxl automatic 1111

neat cypress
#

This new model called as PixArt (not stable diffusion) is literally better than SDXL. have you tried and compared? in the beginning of video i compared with 58 prompts and 290 images https://youtu.be/ZiUXf_idIR4

light bolt
#

hello could anyone tell me how I can remove objects from photos? like a necklace

lament idol
#

Hi All. I'm getting very low it/s with my new 4090 (5 or 6 it/s on a 512*768 on a SD 1.5 model). Anyone has a link for a guide to tweak the 4090? thanks

delicate mural
#

Guys how in the world do I know what version/model of stable difusion I should use.

#

Should one decide on sticking to a model when tuning a prompt, or is it reasonable to tune the prompt and see how the diffrences affect mutiple models for every change in the prompt?
please at me if you respond

lethal venture
#

Hey guys we made a Text2Game app. You type in a sentence (e.g. ChatGPT becomes terminator) and it turns your idea into a "Choose your own adventure" interactive story game. You can even choose to use your face as the main character.

For example, we turned Elon’s “Go F yourself” outburst into an interactive story game. You play Elon and in order  to save Twitter, you have to win back the hearts of Bob Iger and Tim Cook. 
You can play it in this link: https://supermovie.ai/story/v1.html?k=021e145de7a94f8fa2587e87dab3e7a8

Would love your feedback! We are building a waitlist right now.

hasty hornet
#

⛔ too much Elon for my liking, I would prefer none

raw dagger
#

Oh and it unlocks custom text generation HarCoolBlaze

lusty beacon
#

How to create AI Models with SD

#

like how to get the same look

#

und whats the best gui

spiral inlet
#

hey

#

ive loaded webui and i dont have the olive or onnx tabs. trying to generate an image i just gat garbled output

warm junco
# spiral inlet hey

Hey that guide is old and also olive isn't recommended for AMD Users as it lacks support and features.
Remove the --olive stuff from the webui-user.bat and instead add:
--medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling
Then delete the venv folder and rerun the webui-user.bat

spiral inlet
#

this is all thats in my webui-user.bat ```@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=

call webui.bat```

warm junco
pale arch
#

Hi guys, I want to finetune SD models with real people, any script for this ? Many thanks

warm junco
pale arch
warm junco
pale arch
warm junco
#

Okay, but then you may need to change some settings for training. Did you watched some tutorials for that?

#

Also the source material needs to be good quality

pale arch
warm junco
#

Sry, I dont have one rn.

pale arch
#

Thanks sir

warm junco
#

No problem, I would check YT for newer training tutorials in the last 3 months

glass mirage
#

Hey, I'm wondering if training LoRas or embeddings could be used in brand content, training for logo, brand image, etc. Anyone have a take on why it would or would not be possible?

soft sparrow
#

Why is my automatic1111 suddendly running a LOT slower? I checked the settings i used before with the metadata and i am using the exact same settings as before, the only difference is that the generation takes forever

#

500 seconds per iteration.....

#

it was around 30 seconds PER IMAGE before

#

why is my GPU only at 63c while at 100% usage

warm junco
soft sparrow
#

this worked fine before 🤷‍♂️

warm junco
#

Also if its still slow you can show me the txt2img settings or the cmd log in the #🤝｜tech-support channel

soft sparrow
#

i dont think i have xformers but i'll see what happens

warm junco
soft sparrow
warm junco
#

Perfect 🙂

sharp pumice
#

hi all

#

I have a project that I've been told I could do with SD + Dreambooth, and my laptop is not so powerful ... can someone guide me on where to sign up? I was messing around with a handful of AI generators when they blew up, but I am severely out of touch now

sharp pumice
#

and also, let me know if I am asking this in the correct channel 🙂

faint solstice
#

anyone has rtx 4070 benchmarks for sdxl, deforum optional as well i would really like to know please 🙏 🙏 🙏 🙏 🙏

pulsar venture
#

Hi, i'm need an help i use controlnet v 1.1.419 i use model like this control_sd15_canny. There are any upgrade to do?

pulsar venture
#

Is this a tech support even for controlnet?

fervent thunder
#

I already left my problem in tech support, sorry

glacial ridge
#

hi guys. how to type negative promt

dense flint
#

hello. new user here.. i notice huge difference from trying out different sampling methodes.. where can I read about those and know which one is best for what I,m trying to do?

pulsar venture
#

@warm junco sorry for tagging again but could you help with with controlnet extension?

warm junco
pulsar venture
torn hearth
#

Hallooo

#

Voice

#

Im bored

proper rover
#

Who can I talk to about getting a SVD commercial license for a small startup?

shell citrus
#

what version of stable diffusion do you use? 1.5 or 2.x?

full lark
shell citrus
#

I was wondering if I should go through the slight headache of trying to get 2.1 or sdxl to work with the webui

#

I guess I will stick with 1.5 then lol

sullen oar
#

How do you force a specific version?
like:

set PYTHON=3.10

when i try to run, it prints out:

E:\AI\Stable-Diffusion-Web\webui>git pull
Already up to date.
Couldn't launch python

exit code: 9009

stderr:
'3.10' não é reconhecido como um comando interno
ou externo, um programa operável ou um arquivo em lotes.

Launch unsuccessful. Exiting.
Pressione qualquer tecla para continuar. . .
#

i don't want to uninstall 3.12

warm junco
sullen oar
#

oh

#

that would've been much easier than editing the path and uninstalling 3.12......

#

fuuuuuuuuuuuuuuuck

crimson grove
#

hey i am a n00b where do i get styles to use with a1111?

#

i am just playing around with deliberate but the author has a styles.css that is being a paywall

lavish lake
crimson grove
#

yea i just mean the saved styles in a1111

#

but maybe there is not that much value that you can get my downloading them

#

you can just look up example prompts and base yours off those i guess

merry lark
#

silly question but what's the name given to models that fine tune images to a certain after they've been generated using SD

#

i recall there's one that makes anime images better, and one that's optimised for making humans look more realistic

warm junco
merry lark
#

yeah that's the one. thanks

magic harbor
#

Can anyone recommend some Civitai alternatives?

maiden breach
#

Would anybody like to help me from start to finish set up fooocus im a complete idiot😭

tepid zephyr
#

Does anyone have a really good img2img settings to share?

#

I tried a bunch of different combinations to achieve a anime style effect but no success so far :/

clear oyster
#

is intel NPU support being added soon?

jagged umbra
#

Hey, so Im trying to sort some images with stable diffusion metadata in them from those without, and Im noticing that some images store their prompts in a "comments" section while others store it in a "parameters" section. as far as I can tell windows file explorer only has a "comments" tag I can enable in file explorer to separate them out. what would I enable that would sort by parameters? it doesnt seem to be an option in the details tag list.

pale latch
# glass mirage Hey, I'm wondering if training LoRas or embeddings could be used in brand conten...

100% possible. Brands usually have some idea thats consistent. Thats the juice that these image networks love to learn. Consistency. While it probably won't get a perfect logo shape 100% of the time, you could train a branding style into a lora and then use controlnet to smash out images and retain consistency.

I think it would work better for focused marketing campaigns rather than monolithic brand management

unkempt hatch
#

Anyone using an eGPU? I'm about to blow $2000 on one just for stable diffusion

wet jay
#

Is there a way to change the tab icon in the browser when using ComfyUI?

elfin cove
unkempt hatch
# elfin cove egpu?

External GPU. I have a powerful laptop with no good GPU and want an eGPU for training

elfin cove
#

oh right

gray wolf
#

What’s next on the stable diffusion roadmap?

bleak matrix
#

Good morning, everyone!

#

How are we all today?

haughty marlin
#

Hi! anyone knows how i can prevent the fail of the install of cuda toolkit?

bleak matrix
meager kayak
#

Hi guys, I suspect my A1111 has some corrupted stuff. Could someone remind me what command to do to cleanly reinstall everything without losing my config and other photos and such please?

waxen geyser
#

anyone got an idea why my lcm lora is not working

#

tried several times with a lot of different steps / cfg /models

#

always ends in something like its burned completely messed up

simple comet
#

Could you help me to improve my prompt to have better result and more good looking ?

"layered dark blond hairs, sixteen old student, hufflepuff clothes, realistic, sharp details, hd, Renaissance clothes, short wick, neat, disillusioned, boy, dark blonde hair, determined, short hair, slim shoulders"

I like 2 images I generate (and more broadly their moody expression and haircut)

https://cdn.discordapp.com/attachments/244544321938784266/1177621797798170655/image.png?ex=65732cc3&is=6560b7c3&hm=33f2ae69221c34f5376d7389cd3355d8ad8f3251d0e05211c33bdc34624c0e90&

https://cdn.discordapp.com/attachments/244544321938784266/1177621854463217674/image.png?ex=65732cd1&is=6560b7d1&hm=091d0364ff14ff338e64df07ae697518ffa4318ed54e939bebfc0c475c94640a&

restive flicker
#

is Civit crashed?

errant yacht
fervent thunder
#

hi everyone ! is there a way i can run stable diffusion with control net online ?

brazen orbit
#

Hey guys! Is there channel for creating images etc.?

#

I mean help channel

#

For writing prompt, ideas, models - I want to create an image with LOGO*! and make the logo more 3d, cartoonish mascot

#

not sure if thats possible

#

by the way wow its been 1,5 year since stable came out ;o

pale latch
frozen rover
#

Anyone has any idea how I could use CLIP models with the embedding size (77, 512) instead of (77, 768) for guidance in stable diffusion? Is it possible to do it?

chrome dove
#

anyone know how I can get SD to get ride of a lora hash, I can't seem to figure it out in my settings and its automatically being used in all image generations when I don't need it to

#

i think i might have figured it out, something i missed in the settings tab with lobehub ui