🤝｜tech-support

[quick forge]

like I have a fk ton to learn

[maiden haven]

welcome to python which has one of worst (compared to more recent langauges) package/dependency management systems especially for newbies

[quick forge]

which actually i'm quite happy to do, but I have to go back to work on tuesday lol

[maiden haven]

i write rust, go, ruby, terraform, python and every time i touch pytho nafter 6 months everythings broken and versions mismtach and my venvs on fire

[quick forge]

also find out if we're going to get evicted on wed at our hearing :/

so there's that

[maiden haven]

thats not good should you be deaing with that?

[ornate elk]

download python 3.10.11 64bit installer from python

and install it

[drowsy creek]

how to I update python in the venv specifically

[quick forge]

I am. it's about stupid shit, they just want higher rent.

[ornate elk]

you need to check "add python to path" when installing python
then delete the venv folder

[quick forge]

and are claiming that I've made extensive changes to the buildings electrical because I have a washer and dryer and the dryer plugs into our stove outlet with an extension cord.

[noble belfry]

they think you're a hydroponics lab, the cops have been coptering over your mining days 😄

[quick forge]

i've made no changes, but the extension cord has a 30A fuse/shutoff so it looks like some industrial shit

[noble belfry]

are you, killavolt, who is, named, killavolt, abusing residential electricity?

[quick forge]

it's honestly up in the air. they're also claiming we've subletted the place to someone

which I can't even figure out lol

[quick forge]

no actually

[maiden haven]

What modles go in the /Models/StableDiffusion folder?

[quick forge]

my mining rig was only 3 cards and under 400W

[noble belfry]

i just have one, two, 20 parties here a week! tops! they're calling it a hotel! i had to fire the vale- i mean my friend isnt allowed to stand outside and help people locate cars anymore 😄

[noble belfry]

the main models you download to generate things with.

[maiden haven]

3 hours to finally realize my sd is looking for checkpoints a folder extra for some reason

[drowsy creek]

there was no option for that

[noble belfry]

there literally is

[drowsy creek]

I just did the standard install

[noble belfry]

the standard install does have that. ive done it so many times, try it again

[maiden haven]

i read the note on why you have to do that and its stupid

its not even a good reason

im high right now and dont remember why though it was some false equivalency to safety i think

[noble belfry]

the model folder thing?

[maiden haven]

on why the python installer doe sthat weird path thing

[drowsy creek]

where the heck is it then

[maiden haven]

rofl of course theyre doing it through vs

[noble belfry]

i thought it was so other users could use other versions and you could use other distros and not be confused

[maiden haven]

i have no idea if thats a good or terrible idea

[drowsy creek]

yall literally told me to use the install exe

[maiden haven]

im laughing because its well known that the python has that installer checkbox so he had a 100% checkmate UNLESS you installed python some weird othe rway

[ornate elk]

looks good

[maiden haven]

isnt that the visual studio feature add thing?

[drowsy creek]

welp I'm not gonna get tech support here

[fair oxide]

hey guys
why can I not use finetuning?

the bottom rows of options are all greyed out

[maiden haven]

the right one is python i think, go with the python install

[noble belfry]

isnt adding python to variables the same as path? sorry this install screen looks so different to the one i see over and over

[drowsy creek]

no it's the EXACT version CS1o told me to use, with the 64 bit installer exe

[maiden haven]

im not sure how msvc handles python versions or anything like that it might be weird

ah

[noble belfry]

the old installer used to literally ask you the question, add to path. anyway, we can help you just do it yourself

[drowsy creek]

also why did SD auto install the wrong python version to the venv to begin with

[noble belfry]

its not hard

[ornate elk]

it didnt come with python

[noble belfry]

https://tenor.com/view/pointing-at-you-spider-man-spider-man-unlimited-spider-man-across-the-spider-verse-it-was-you-gif-14641812833362592646

[maiden haven]

youre gonna be fixing version miss matches like whackamole so dont worry too much about it

[noble belfry]

all the pythons

[maiden haven]

python is dependency hel

[quick forge]

also, how can I set this computer up so that I can access the WebUI's from any computer? I want to just have this running and then access SD from my main windows machine and possibly a discord bot.

[maiden haven]

im simultaneously fighting pip/python and trying to rememer what directrions slashes go in windows

[quick forge]

windows is C:\

[maiden haven]

assign your computer a static ip, start the server with systemd and go to the hostname in your browser

it already is aw ebserver etc you dont need to do anything

[quick forge]

i feel like i shold have known that

[maiden haven]

most distros dont come with ufw or iptables on

[quick forge]

it says something about changing something to true in Launch()

[noble belfry]

im finding it hard to follow this channel. its like a cross between being a midwife helping childbirth and a batting pen

[quick forge]

anyway, im just going to back to the beginning and take it one step at a time and actually try to understand what it is im doing so im not just doing things blindly

[noble belfry]

runs out field GOTTEM

[maiden haven]

in the end you're just saying python run webserver so you just need something to trigger when to tell it that, reboot, etc

[drowsy creek]

yeah imma probably not be able to use SD it seems

[quick forge]

you haven't been in many chat rooms have you?

[ornate elk]

why?

[maiden haven]

smoking and troubleshooting was a bad idea i have no idea what i was fixing

[ornate elk]

you need to delete the venv folder

[drowsy creek]

wait will it just use the normal python on my PC without it?

[noble belfry]

please dont swozey, you'll kill us all.

[ornate elk]

yes

[drowsy creek]

ok

[ornate elk]

it now will use 3.10.11

[quick forge]

dude's gonna accidentally make skynet and not remember how

[maiden haven]

where am i adding an extra Stable-diffusion at? cmd line ? i cant find it

[quick forge]

anyway, im off to see the wonderful wizard of storage

[drowsy creek]

so it's making a new venv using the correct version now?

[maiden haven]

did it load packages, you might need to pip install --file requirements.txt or something in there

[noble belfry]

please swozey ilu but dont help while high

[maiden haven]

?

i engineer high 24/7 lmao

i just keep forgetting where im focusing

[noble belfry]

did you not just ask where to put your models for SD?

[maiden haven]

i asked if anyone knew where i'm possibly injecting an extra line into the command, im not sure if its on the startup or python end

hows that relevant?

my global search isnt finding anything

[noble belfry]

🙂 we just need to slow the chat a little.

[maiden haven]

cool, talk as little as you like bud, i'm not the Wit police

i'll tech support as much as i like 👍

[noble belfry]

i mean youre asking questions while other people are asking and youre giving solutions at the same time (good ones) its just impossible to follow 😄

[maiden haven]

It's not my problem if you can't keep up with basically an IRC chatroom

You don't need to read everything lol

[noble belfry]

well i was trying to help you specifically but it got busy, it doesnt matter now. its fine.

[maiden haven]

bro, i am not asking if it's ok

i am not controlling my helping other people or not because you can't follow the channel, lmao, sorry man

weird request

[torn lotus]

hello, I'm looking to use a method of replicating images to create a virtual model. What is the best method, faceswap+pyracanny or using kohya_ss?

[ornate elk]

yes

[drowsy creek]

nice it works now and only takes like 45 seconds again instead of 4 minutes

[maiden haven]

@noble belfry SOrry I misunderstood what you were doing, I meant if anyone knew off the top of their head real quick, I'm looking in the code not sitting here really waiting for help or anything but i appreciate it

[noble belfry]

yeah no probs mate 🙂 was just like a cat trying to catch a laser pointer 🙂

[maiden haven]

lmao great way to explain it thanks

[noble belfry]

https://tenor.com/view/larry-david-seinfeld-pretty-good-enthusiasm-curb-gif-3938316

[maiden haven]

do these have any affect on it where its scanning for checkpoints? the options for these seem weird \Base Model \Author Name \ etc they look more like outputs maybe?

this must be something else i cant put in my own entries

[noble belfry]

C:\AI\stable-diffusion-webui\models\Stable-diffusion this is my dir. throw yamls into it.

dont tweak the others, you put loras into C:\AI\stable-diffusion-webui\models\Lora etc or C:\AI\stable-diffusion-webui\models\lycoris

unless you want your dirs central. in which case put symlinks in them

[candid walrus]

Anyone able to help me learn how to do inpainting or generate images (I can't get any of the loras to generate something I'd be able to say, oh that's the anime character I'm trying to generate let alone match the example image)

[ornate elk]

can you show an example when you use a lora?

[maiden haven]

So is the checkpoints dir not supposed to be the root Models dir? It's Models/Stable-diffusion everywhere?

ive kind of equated model = checkpoint

[candid walrus]

I haven't used it in a while but sure let me fire it up. I'll show you example prompt and image vs what actually got spit out.

[candid walrus]

I'll ping you in a little bit.

[ornate elk]

the better channel for this is #📝｜prompting-help

[candid walrus]

Alright sure I'll ping you there in a bit.

[mighty topaz]

Someone know how update SD via Stable Matrix from 1.7.0 to 1.8.0 ? i can only add packages new instal and pick 1.8, no update on a 1.7

[quick forge]

ok so i now have a 1TB secondary SSD. how do i install SD (orSDXL?) onto that non-boot drive? im not sure how to install stuff fro github, and am on Arch

[quick forge]

formatting the new partition, should it be btrfs?

[maiden haven]

i always wind up on ext4 because i forgot i even had a choice

then im sad i dont have snapshots

[dense sandal]

@ornate elk I think I finally fixed my issue by importing the Cv2 package from another computer

after reinstalling windows like 3 times and cleaning 2 of my drives

[quick forge]

how do i actually install this? Do i just need the safetensors files in the stable diffusion models folder?
https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/tree/main

or do i need the directory structure there too?

[mighty topaz]

you can just download file and put this to folder models and use it like normal model

[quick forge]

so whats all the scheduler, vae, unet and other folders then?

[mighty topaz]

you work on comfyui or sd a1111?

[quick forge]

right now a1111

[mighty topaz]

when i watch tutorial on YT how use refiner in comfyui they tell ppl to downlaod file this and use it , and i do the same

i recommended to search on yt refiner stability dyffusion and u know everything

[quick forge]

yeah but im trying to understand how it's all working together and what im doing, not just blindly follow random youtubers. it's always conflicting advice. that's why im here trying to learn all about it

and also completely new to running linux so that's a whole learning curve by itself/.

[mighty topaz]

i dont know what youtubers u watching but i watch only those who explain everything here you go

https://www.youtube.com/watch?v=zNZFelT_-Tk

[quick forge]

i want to have this setup a particular way if possible, where I have just SD installed, and different webui's all pointing to the same install/extensions/models etc.

[mighty topaz]

i know what all this do but i dont know english well so i cant help just watch this movie

[maiden haven]

you can download them by clicking link from hugging face every time it updates or git clone it into your models dir and manually go in and git pull updates

[drowsy creek]

any idea why adetailer is deepfrying my generations now?

it didn't used to do that back when I used to use SD

[quick forge]

how do you git clone something into a specific directory like the models directory?

[ornate elk]

You need to add --no-half to the webui-user.bat

For inpainting and adetailer to work on AMD

[drowsy creek]

2nd image: the image with the same parameters now after re-installing SD 
3rd image: 2nd image's parameters but with adetailer enabled

[drowsy creek]

ok

thanks it works now

[crimson rover]

Is there any good extension to fix face in Anime style? Iam struggling to fix it in Inpaint

[nocturne topaz]

adetailer

[crimson rover]

okay, thanks! and one more thing, if i want to Inpaint something very specific, how to do that? i tried many attempts but it's not inpainting one specific weapon

[sleek breach]

So.. my stable diffusion insists on searching in the Alberto user folder, when there is no such folder, what am I meant to do now?

[nocturne topaz]

reinstall python on your computer

there will be something in the PATH settings that is pointing at that

[quick forge]

how do i install SD on Linux (Arch or Ubuntu) without a webUI.
And then second to that, how do I install Just a webui?
I'm going to be installing it on a separate SSD to the OS SSD.

is there a guide for that sort of setup because i haven't really been able to find one that has SD installed separately.... I don't think.

[nocturne topaz]

you still struggling chief

install without a webui is just downloading a model and then either writing a python script or opening python and calling diffusers to make a pipeline to use the model (yeah, that is quite technical and not really an easy way to explain that)

normally your best bet is to go for comfyUI or a1111, with the caveat that you are on amd

so for comfy, it looks like you create a venv, activate it, install pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7 and then git clone the comfy repo into the right location and then navigate to that and run python main.py

I don't have an amd setup so can't help you much more than that sorry

[sleek breach]

How do I check this? Already reinstalled it

C:\Users\Usuario\Documents\Stable Diffusion\stable-diffusion-webui>git pull
Already up to date.
venv "C:\Users\Usuario\Documents\Stable Diffusion\stable-diffusion-webui\venv\Scripts\Python.exe"
No Python at '"C:\Users\Alberto\AppData\Local\Programs\Python\Python310\python.exe'
Presione una tecla para continuar . . .

Still getting this

[nocturne topaz]

in your start menu type path, then select Edit the system environmeent variables

in there click "Environment Variables" then look through the PATH variable for something odd relating to python

[sleek breach]

Something off like what?

This is all I got

[nocturne topaz]

and in the system path?

(as opposed to the user one)

I'm 50/50 expecting Alberto to pop up there

[quick forge]

not struggling so much as not understanding a lot. I want to run multiple webui's but i dont want to have to have duplicates of all the models and stuff so im not really sure how to go abut doing what i want to do. I just needed to sort out a few things first. like adding more drive space.

[nocturne topaz]

both comfy and a1111 have config files that you can use to point them at a shared models directory

a1111 is the webui-user.sh file, and in the commandline_args you add --ckpt-dir /path/to/folder/models --lora-dir /path/to/folder/loras

[sleek breach]

I'm sorry, where is system path?

[quick forge]

oh ok cool. what about stability matrix?

[sleek breach]

Also, before anyone says it, yeah, I'm the guy who comes look at Marcelo

[nocturne topaz]

in the system area just below the user area in the same window

[sleek breach]

Erm

[nocturne topaz]

looks cool, never tried it

[nocturne topaz]

ah ok.. so it's not that causing it..

so what happens if you run webui-user.bat?

[sleek breach]

I just get this message

[nocturne topaz]

or is that the results..

[sleek breach]

Already up to date.
venv "C:\Users\Usuario\Documents\Stable Diffusion\stable-diffusion-webui\venv\Scripts\Python.exe"
No Python at '"C:\Users\Alberto\AppData\Local\Programs\Python\Python310\python.exe'
Presione una tecla para continuar . . .```

[nocturne topaz]

whats in your webui-user.bat file?

[sleek breach]

git pull
@echo off

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

call webui.bat

[nocturne topaz]

add --xformers to the commandline_args line, delete the venv folder and run webui-user.bat again

[sleek breach]

creating it now in the cmd

seems to be working

[muted flame]

So my diffusions were painfully slow and it would only let me do one everytime before i had to restart my computer due to lack of vram for a 2nd diffusion, so I upgraded my graphics card. It now has much, much faster generation, but crashes at 95%
what do I need to do to fix this?

Also, it crashes my computer at the 95% mark about every other time I attempt it

[finite gull]

Hello. I had trouble running a trainer using Automatic1111. The I am using anaconda, so after all the workarounds i attempted, i decided to delete all my environments and start anew.

On the Automatic1111 github, I followed the instructions and downloaded the packages required for Nvidia cards. I was not expecting it to include the stable diffusion with GUI.

So what I am trying to figure out is since this was done outside anaconda, I have to start over and run those nvidia prereqs with anaconda, or am I overthinking it?

Can just make a new environment and run the webui via anaconda as is?

[nocturne topaz]

I would delete the venv folder and restart the system

[nocturne topaz]

running a trainer - use OneTrainer or kohya_ss (bmaltais' fork has the gui)

a1111 is for image generation

[muted flame]

Just delete the whole venv folder and it won't need to be replaced or anything like that?

[flint gyro]

Hello, I'm new here, how can I generate images 😅😅

[nocturne topaz]

when you re-run the webui it will rebuild what it needs to work

[muted flame]

Okay sounds good. I will give that a try. Thank you!

[finite gull]

you mean if i run it through anaconda?

[nocturne topaz]

head over to clipdrop or civitai or aiscribbles.com to get started

[nocturne topaz]

nope.. that wasn't your reply

you need a seperate tool to train a model

[flint gyro]

Is it not possible anymore to do it from one of the channels here?

[nocturne topaz]

nope - #1047610792226340935

[sleek breach]

Yup, it works now

[flint gyro]

Thanks for your help 🫶

[sleek breach]

Now for anime stuff, which models do you recommend?

[nocturne topaz]

how much vram do you have (i.e. can you run sdxl models?)

[muted flame]

How much VRAM is needed for SDXL?

I have 12 vram now, does that mean I'm able to run those models?

Can all models just run in local stable diffusion 11111

[fair oxide]

Hey guys can anyone help me out, why is the fine tuneing option not available in the settings?

[sleek breach]

15,6 GB usable

Wait, I think I checked for normal RAM rather than Vram, how do I check for it?

[nocturne topaz]

Will be the dedicated GPU memory (not the shared)

[sleek breach]

Unsure of what you're looking for here, if there's something in spanish you want to know the translation of, I'll tell you

[nocturne topaz]

the gpu 1 tab

[sleek breach]

Oh, holy

I just noticed it said G, not C

There ya go

[fair oxide]

How do I generate an image?

[nocturne topaz]

#🤝｜tech-support message

[fair oxide]

Is it not possible to generate images in a bot server now?

[nocturne topaz]

7.8gb vram - anime models in sd1.5 - https://civitai.com/models/24779/dark-sushi-mix-mix?modelVersionId=93208 is probably a good starting point - heaps to choose from on civitai

[nocturne topaz]

nope - #1047610792226340935

[sleek breach]

Feels a bit too detailed for my tastes, I'd prefer something with less realism

[muted flame]

It crashed my computer again.

[nocturne topaz]

error message? or hard crash?

[muted flame]

Hard crash

I attached the screenshots earlier where I did have error messages when it didn't crash it but that was before deleting the folder and having it make a new one

[sleek breach]

Like this one I'd say

[sleek breach]

https://civitai.com/models/119012/bluepencil-xl

[nocturne topaz]

https://civitai.com/models - have a look there, filter by sd1.5 models and go to your hearts content!

[nocturne topaz]

that's a sdxl one, you may be able to run it, but might not

[cosmic furnace]

some sdxl need 48gb ram 😦

well at least if u try do smthing higher than 1024

[ornate elk]

On AMD it can work with 12gb when using 768x768 or 768x1024

But I wouldn't recommend it

[muted flame]

Okay

And I'm having an issue with my new graphics card where when I get to 95% it's crashing My whole PC. I posted the pictures a little bit earlier. Any advice on how I can fix that?

[ornate elk]

Your better with using 1.5 and upscale

[ornate elk]

What's your GPU?

[torn lotus]

what is better to produce a model in different setups? faceswap with pyracanny or lora training

[finite gull]

got bmaltais gui working (at least loading in browser). havent played with it yet, but its a milestone in my book. thank you!

[nocturne topaz]

instantID or ipadapter

[muted flame]

How do I find that? I can't seem to find it anywhere

[nocturne topaz]

what graphics card you got?

[muted flame]

Power color fighter Radeon RX6750 xt

I couldn't find it on power color or AMD site

Both of them just talk about GPU in vague terms

[ornate elk]

Okay and when does the PC crashes?

Only when using Stable-diffusion ?

Only when using sdxl models?

[muted flame]

When using only say able to fusion. It crashes at 95%

[ornate elk]

Have you tried a 512x768 resolution with a not sdxl model? Any 1.5 model for example Dreamshaper v8

[muted flame]

I did not adjust any of the size settings so I was on whatever the original scale is. I was using majicmix

[small turret]

why I can't find deliberate anymore on civitai? I remember It was pretty good and popular a year ago

It's paid rn or they renamed the project(model)?

[ornate elk]

So it crashes every time?

[maiden haven]

when i go to mov2mov and put a movie in, set a prompt and click generate.. . what is it supposed to do? mine just sits there with interrupt/skip

[muted flame]

I attempted five times, three of which crashed my computer completely, two of which just left me an error message that I posted in those screenshots

[maiden haven]

i thought i'd be able to inpaint the video or something

[ornate elk]

How much RAM do you have?

The error says its to less

[nocturne topaz]

deliberate went exclusive to try and monetize their work.. you can find the files on huggingface https://huggingface.co/XpucT/Deliberate/tree/main

[small turret]

yeah, already read about it, but ty

[muted flame]

8gb ram

[ornate elk]

Thats to less and the cause of the crash mostly. You need to increase your windows pagefile to not get it crashed.
Before doing that make sure you have 15-20gb free space on the C drive

[muted flame]

So it's just an issue of storage space?

I have stable diffusion on my D drive, does the space need to be on the c drive or the D drive with that fact?

[ornate elk]

No its an issue of to less RAM (Memory).
Increasing the windows pagefile gives you "virtual ram"
But that pagefile takes space from the SSD

The free space need to be on C

[fair oxide]

if anyone is using https://platform.stability.ai/ do you know why I cant use finetuning?

do I have to request access?

[muted flame]

Okay. Thank you.

[nocturne topaz]

if you are logged in and it's greyed out, I would suggest it's likely that the feature isn't available yet

[ornate elk]

NP, if you made some free space on C, you can checkout the guide how to increase the pagefile.
https://www.thewindowsclub.com/increase-page-file-size-virtual-memory-windows
Make sure its enabled for C and not for any other Drive.
After you adjusted it (Min 16000)
You need to restart the PC.

[fair oxide]

nothing is working lol. I cant even create images on dreamstudio

[muted flame]

I'm actually waiting on a larger hard drive that is supposed to arrive tomorrow, so I bookmarked that page you just sent me and I will make another attempt tomorrow when I have the space. I do greatly appreciate the assistance

[fair oxide]

are there any alternatives to stability.ai that can allow me to use my own lora and have an api? @nocturne topaz

sorry new to sd

[nocturne topaz]

is your lora saved to Civitai?

and do you have your own graphics card?

[fair oxide]

civitai only gives me the option to upload it publicly, once i do it redirects me to other services like runpod

[fair oxide]

yeah but I only have a 3060 and want to generate thousands of images

[nocturne topaz]

3060 won't be that much slower than what you'll get from the public api

[fair oxide]

not public, Im looking for paid solutions

[nocturne topaz]

why the api requirement?

[fair oxide]

cause I want to generate thousands of images and its a hassle to do it manually. I have a lora and image prompts that I want the images to be based off of, after that Ill pick the best ones

Right now im using foocus on runpod and a lora I trained on google collab

[quick forge]

alright now im really confused. i installed kubuntu but when trying to follow this guide i run into errors about not being able to install certain things.
https://civitai.com/articles/1028/running-automatic1111-on-ubuntu-for-amd-gpus

[nocturne topaz]

whats the error?

[quick forge]

E: The repository 'https://repo.radeon.com/amdgpu/@AMDGPUVER@/ubuntu jammy Release' does not have a Release file.
N: Updating from such a repository can't be done securely, and is therefore disabled by default.
N: See apt-secure(8) manpage for repository creation and user configuration details.

so i can't seem to install amd drivers, and rocm doesn't install either

[nocturne topaz]

This might be what you need chief - https://github.com/konieshadow/Fooocus-API

[fair oxide]

Thanks. i found this as well but I was looking for a hosted solution that allows me to use my own lora's, that api can be hosted on replicate but how would I use my own loras with it?.

[nocturne topaz]

If I was to use a private lora, I wouldn't be using replicate - and all of the other hosted APIs I know of allow use of public loras from civitai

[fair oxide]

I dont mind uploading my lora to civit ai and then using it that way, and which host apis are you talking about?

[quick forge]

ffs, everything i try i seem to run into some kind of errors.
Is there a step by step guide for installing SD oin unbuntu with AMD cards?

the one i linked doesn't seem to be working for me

[nocturne topaz]

aihorde.net is the main one I'm thinking of

even then, that's not really what you are after

lemme get this straight

you have a lora, and some sort of file with a pile of prompts that you want to run against the lora that would generate several thousand images, you want a way to be able to call a script (be it python or powershell or bash or similar) to call the api to generate and save the image to your local machine. You have a 3060 on your computer but would like to use something faster like an a100 or h100

am i correct thus far?

[quick forge]

this is fkn frustrating, everything i try to do is met with some kind of error. how tf do i install SD on ubuntu?

is 23.10 too new for rocm yet?!

[nocturne topaz]

Fuck it, lets figure this out together

[quick forge]

im sure it's user error lol

[nocturne topaz]

Instructions on amd site are for 20.04 and 22.04

[quick forge]

ya i just found out i have 23.10

[nocturne topaz]

we can try the 22.04 instructions in 23.10, or did you want to downgrade to 22.04 first?

also, is ubuntu installed to hard drive, or are you using windows subsystem for linux?

[quick forge]

it was working ok in Arch, but everytime i needed to install something i couldnb't use pip and dont understand how to use pacman in the same way

Kubuntu is installed on an SSD. no windows on this sytem.

[nocturne topaz]

yeah.. everything is written for pip

[maiden haven]

its pretty much just pacman -Syu

[nocturne topaz]

ok.. lets go with 23.10 then

i'm assuming this is a fresh install?

[quick forge]

i forget exactly what I was trying to install, but following instructions and i tried using pacman -Syu instead and just got errors

I can downgrade if necessary.... it's a totally fresh instal

[nocturne topaz]

I'd start by running these commands

sudo usermod -a -G render,video $LOGNAME
wget https://repo.radeon.com/amdgpu-install/6.0.2/ubuntu/jammy/amdgpu-install_6.0.60002-1_all.deb
sudo apt install ./amdgpu-install_6.0.60002-1_all.deb
sudo apt update
sudo apt install amdgpu-dkms
sudo apt install rocm```

[quick forge]

ok so do I do that one by one or just paste the entire thing into Konsole?

[nocturne topaz]

one by one into console

apart from the last echo

and the see prereq...

[quick forge]

see prerequs where?

oh thats a place im getting stuck too.... getting errors on those.

[nocturne topaz]

which line gave what error?

[muted elm]

hello! i tried to download ReActor extension but it doesnt appear in my stable diffusion web and this is what shows on CMD

[quick forge]

hang on im not there yet

[nocturne topaz]

in windows go to the file explorer, open up C:\IA\stable-diffusion-webui\venv\Scripts

[quick forge]

well so far im at "7 packages can be upgraded. Run 'apt list --upgradable' to see them."

after apt update

[nocturne topaz]

then in the address bar type cmd and press enter

[nocturne topaz]

we can ignore that for now

[muted elm]

ready

[nocturne topaz]

type activate and press enter

[nocturne topaz]

should add (venv) to the front

[quick forge]

can't install rocm.

The following packages have unmet dependencies:
rocm-gdb : Depends: libtinfo5 but it is not installable
Depends: libncurses5 but it is not installable
Depends: libpython3.10 but it is not installable or
libpython3.8 but it is not installable
E: Unable to correct problems, you have held broken packages.
killavolt@SDBox:~$

[muted elm]

now what do i do

[nocturne topaz]

one moment, will sort doc out

[nocturne topaz]

pip install insightface

[muted elm]

i write that?

[quick forge]

when i tried to install the dependancies i get this

killavolt@SDBox:~$ sudo apt-get install -y libtinfo5
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Package libtinfo5 is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package 'libtinfo5' has no installation candidate

[muted elm]

ready

[quick forge]

i ran that sudo command.... didn't edit the file though

i dont know how to disable the cdrom source though

line 1 is : # deb cdrom:[Kubuntu 23.10 Mantic Minotaur - Release amd64 (20231010)]/ mantic main multiverse restricted universe

[nocturne topaz]

hmm.. maybe not then

[muted elm]

what do i do now

[nocturne topaz]

https://github.com/Gourieff/sd-webui-reactor?tab=readme-ov-file#insightfacebuild

follow the insightfacebuild instructions from there

[quick forge]

and i just get the same error trying to install libtinfo5

[nocturne topaz]

ok.. lets go with the rebuild to fresh ubuntu 22.04 then (and not kubuntu)

[quick forge]

how do i do that?

[nocturne topaz]

reinstalling the operating system..

yeah maybe not then

[quick forge]

oh ok lol.

lemme grab the ubuntu 22.04 iso then

[nocturne topaz]

are you able to install from iso?

I'm assuming so given you've gone from arch to kubuntu

[quick forge]

ya, using Ventoy

mind blowingly easy

this is all a brand new computer im setting up for the specific purpose of running stable diffusion and nothing else

[nocturne topaz]

oh nice, never heard of it, I always used to just build the bastards manually

which gpu is it?

I know it's amd, I don't remember you saying which one

[quick forge]

ventoy is great, you just drop the iso on to the usb stick and it will auutomatically find it and let you run it.

RX6800

I can probably have ubuntu reinstalled again and be back here in like 20 mins

[nocturne topaz]

I'm still here for a couple of hours.. it's a work from home day

[quick forge]

is rthis what im looking for ?
https://releases.ubuntu.com/jammy/

[nocturne topaz]

looks good to me

[quick forge]

be back shortly on a new install lol

[quick forge]

well this is.....interesting

[edgy plinth]

is dream studio down again??

[quick forge]

well im just waiting for Building initial module for 6.5.0-18-generic

after running amdgpu-install

ok... i think i have rocm installed, and went through all those commands above

[thick crow]

is this normal ?

I would like it to use more resources

[silk estuary]

It worked! Thanks

[nocturne topaz]

press where it says 3d and select CUDA

[nocturne topaz]

It's going hard 🙂

[thick crow]

oh okay, thanks !

[compact pier]

Where's the bots?

[nocturne topaz]

#1047610792226340935

[thick crow]

I also have a Intel® Core™ i7-13700K Desktop Processor 16 cores (8 P-cores + 8 E-cores) 30M Cache, up to 5.4 GHz, should I use that instead of my gpu?

[nocturne topaz]

Nope.. tis 20x slower than the gpu

[thick crow]

okay!

thanks for the help

appreciate it

[nocturne topaz]

No worries

[quick forge]

how the hell do i edit ~/ .bashrc ?!

[nocturne topaz]

what have you found now that is asking for that?

[quick forge]

following this
https://civitai.com/articles/1028/running-automatic1111-on-ubuntu-for-amd-gpus

things seem to be working now without the errors as before

[thick crow]

i'm new to this, wondering if there's a way to open the ai art in Photoshop and see the layers ? (if it has any)

[quick forge]

but now i have to edit that file

Edit the ~/.bashrc file:

Add alias python=python3 at the end of your .bashrc file, save and close.

source ~/.bashrc

[nocturne topaz]

nano

[quick forge]

oh neat

[nocturne topaz]

so nano ~/.bashrc

[nocturne topaz]

no layers, all flat

[thick crow]

ahh thats ashame

[quick forge]

i wonder if i should just make the bash_aliases file it suggests

just added it to the end

so this is where I clone the git for installing the webui..... instead of cd ~, would I change that to my SSD where i want stable diffusion and webuis to be installed?

ummmmm ...... how..... no drive letters? where do I CD to?

[nocturne topaz]

just go cd and press enter (should take you to home)

and then mkdir sd-webui

then cd sd-webui

then do the git clone there

[quick forge]

this is where i want t put it.... i think im in the right spot
killavolt@SDBox:/media/killavolt/SECONDARY-SSD$

yup

i guess the python alias didnt work

ok here's where i have questions. we're at rocm 5.7 are we not?

pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm5.5

At the time beeing, pytorch for ROCm5.6 is not available, but the 5.5 version should be ok.

i guess just changing the number to 5.7 worked lol...at least so far

[spark yarrow]

really dumb question. I have a membership as of today but can't find where my images are stored after I generate them?

[nocturne topaz]

membership usually has nothing to do with image storage?? Are you using dreamstudio?

[nocturne topaz]

so far so good then..

[quick forge]

fuck

Traceback (most recent call last):
File "/media/killavolt/SECONDARY-SSD/stable-diffusion-webui/launch.py", line 48, in <module>
main()
File "/media/killavolt/SECONDARY-SSD/stable-diffusion-webui/launch.py", line 39, in main
prepare_environment()
File "/media/killavolt/SECONDARY-SSD/stable-diffusion-webui/modules/launch_utils.py", line 384, in prepare_environment
raise RuntimeError(
RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check

[nocturne topaz]

what do you get if you try and run rocminfo

[quick forge]

ROCk module is loaded
Unable to open /dev/kfd read-write: Permission denied
killavolt is member of render group

unless i do sudo rocminfo

[nocturne topaz]

ok, so that tells me something isn't quite right with rocm if you don't have permissions

[quick forge]

so how do i correct that?

[nocturne topaz]

https://github.com/ROCm/ROCm/issues/1211

sudo usermod -a -G video $LOGNAME looks like the command needed

and then reboot the system

[quick forge]

sudo usermod -a -G video $LOGNAME fixed it

stil get the same error when trying to run launch.py

[nocturne topaz]

and you have rebooted?

[quick forge]

whoops no hang on

collecting torch 2.1 now apparently

2.0.1

[nocturne topaz]

yay 🙂

[quick forge]

ffs

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

i get that i have an AMD gpu but i dont want it usiong my cpu

[nocturne topaz]

which torch did it grab?

[quick forge]

2.0.1 i think

[nocturne topaz]

you may have to go and activate the venv manually and reinstall torch using pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7

[quick forge]

does it not like that it isn't installed in my user directory but instead on a secondary SSD?

[nocturne topaz]

nope..

should be fine

[quick forge]

i thought i did that but im running that command now

i think its getting 2.2

oh FFS
ERROR: Exception:
Traceback (most recent call last):
File "/usr/lib/python3/dist-packages/pip/_internal/cli/base_command.py", line 165, in exc_logging_wrapper
status = run_func(*args)
File "/usr/lib/python3/dist-packages/pip/_internal/cli/req_command.py", line 205, in wrapper
return func(self, options, args)
File "/usr/lib/python3/dist-packages/pip/_internal/commands/install.py", line 389, in run
to_install = resolver.get_installation_order(requirement_set)
File "/usr/lib/python3/dist-packages/pip/_internal/resolution/resolvelib/resolver.py", line 188, in get_installation_order
weights = get_topological_weights(
File "/usr/lib/python3/dist-packages/pip/_internal/resolution/resolvelib/resolver.py", line 276, in get_topological_weights
assert len(weights) == expected_node_count
AssertionError

why is this so god damn stupid

[nocturne topaz]

you did the source activate first?

[quick forge]

...... did the what now?

[nocturne topaz]

you need to cd to your stable-diffusion-webui folder

[quick forge]

i am there

[nocturne topaz]

then do source venv/bin/activate

[quick forge]

ya i just did that and ran the pip3 command again.

[nocturne topaz]

should put (venv) in front of the location

[quick forge]

now it's a bunch of "requiorement already met" messages

something is working now...so far lol

[nocturne topaz]

ok

[quick forge]

installing requiorements for codeformer

jfc i hate this keyboard

[nocturne topaz]

that's a good sign

[quick forge]

finally, launching webui with arguments:

but downloading stuff

might load this time

[nocturne topaz]

probably stablediffusion 1.5

[quick forge]

yup

is that the latest version?

[nocturne topaz]

nope.. but it's a starting point

[quick forge]

good god lol

so when that's installed how do I upgrade it?

[nocturne topaz]

you cd over to the models/StableDiffusion folder and then use wget to download models from huggingface or civitai

or in a1111 go to extensions, search for the Civitai Browser + extension and install that, which allows you to search from within a1111

[quick forge]

fkn loaded the webui finally....

[nocturne topaz]

ok.. just hit generate and see if you get an image

[quick forge]

so now when I want to launch this again, how do I do it? Do I have to activate the source, then launch the python script again?

yup. got an image

[nocturne topaz]

you can do it that way, or just use bash webui.sh

from the stable diffusion webui folder

that activates the source and fires it up

[quick forge]

is there anyway to make a shortcut like in windows?

[nocturne topaz]

I guess you could make a shortcut to webui.sh on the desktop and it should work

[quick forge]

ok, so if i want to fire it up and access it remotely how would I do that?
Like I want to run this linux machine like a server and access the webui from my main windows machine

or any other computer...

[nocturne topaz]

to run it remotely you would want to go nano webui-user.sh and there will be a line with COMMANDLINE_ARGS on it, remove the # from the front and add --listen to that line

then run it again and it should say "available on URL: 0.0.0.0:7860" which means you can access it from any machine that can access that computer

[quick forge]

eexport COMMANDLINE_ARGS="--listen"

export COMMANDLINE_ARGS="--listen"

[nocturne topaz]

you'll also want to add --enable-insecure-extension-access as well if you want to install extensions from the network without having to jump back onto the main system

[quick forge]

export COMMANDLINE_ARGS="--listen --enable-insecure-extension-access"

i need to assign a static ip too don't i?

[nocturne topaz]

no.. on the local network it should be findable, but you could fix the ip of your machine in the network settings so it's easier to find later

[quick forge]

well.... im finally back to where i was this morning lmao

with the webui running, just on ubuntu now

[nocturne topaz]

I guess that's a yay?

[quick forge]

yes it is

[nocturne topaz]

https://tenor.com/view/tuesday-happy-elmo-gif-13143615

[quick forge]

i wonder if i can get kubuntu with 22.04
or is it possible to somehow change ubuntu around .... i keep looking in the bottom corner for the clock and it isn't there

i thought linux was configurible like that

[nocturne topaz]

No idea.. but if you are only remoting into it, shrug?

[quick forge]

true

not really on my priority list, just a curiousity

so how do I go about installing other webui's like comfyui? just follow their install instructions? how do I move the models and shit to a central location for every webui to use?

[nocturne topaz]

dont even need to remote into it.. just set bash webui.sh as a startup task and then use it headless (ie you just go to the browser on the windows device and navigate to http://linuxbox:7860 and that'll do the job

[nocturne topaz]

comfy has a file that you configure where you tell it where the a1111 install is and it uses the models and loras from there

as for installing comfy on amd.. no idea, and i'm not up for that challenge today sorry

[quick forge]

no worries for installing it. just curious

isstabilitymatrix the same way?

there are a few different things IW ant to run so I can try them out and see which I like best

[nocturne topaz]

I think that has a whole pile of 1-click installers, but im 99% sure that isn't setup for AMD

[quick forge]

god damnit i hate this keyboard

AMD GPU support is experimental and may not work as expected in some configurations.

apparently

[nocturne topaz]

yeah, and with the luck you were having....

[quick forge]

I would chalk most if not all of that down to my lack of knowledge of linux

like this all legit feels like I'm 15 again fucking around in dos with fdisk and all the stuff normal people didn't do on computers haha

[nocturne topaz]

hahaha yeah I can see that.. definitely dos 6.22 flashbacks

and setting up optimisations to get the soundblaster fx card to work

[quick forge]

omg soundcards. i always had trouble with those

well now im going to try and get comfyui installed lol

[nocturne topaz]

https://tenor.com/view/good-luck-good-luck-gif-20034388

[quick forge]

hmmmm..... i wonder if i could/should drop my media drives in this bitch and run plex on box too

im on a roll now man 🙂

[nocturne topaz]

I wouldn't run plex and stable diffusion concurrently

as a file host yes, but as a transcoder to push across the network no

[quick forge]

yeah you're probably right. plex is running well enough on my main compouter anyway

what i am actually very confused about though, is how do i actually install something from github?

[nocturne topaz]

a1111 extension or comfy?

[quick forge]

like do i havbe to clone the git? run a file?

[nocturne topaz]

git clone usually

[quick forge]

generally anything really.....

ivenever understood github...ever

except when i could get a basic windows executable file

is this the right comfyui?
https://github.com/comfyanonymous/ComfyUI

and i would just clone that git to my drive?

friggin instructions right there

[quick forge]

anyone know why i can't view my /etc folder?

[quick forge]

do i have to do that source activation every time before I load something?

[jolly gorge]

anybody know how to optimise forge for mac m1?

i have it running but the performance seems about the same

[nocturne topaz]

Generally yes. Think of the venv as a python install inside your python install

[maiden haven]

i just made my first lora and put it in the models/Lora dir the json and .safetensor, when ic click lora it just says loading

is there anywhere else i needfor it

i hop over to kohya to train a model hop back to sd and my sd is broken again lmao

[quick forge]

how the hell do I quit the server from the terminal without just closing the terminal?

anyone know why comfyui wouldn't be loading my checkpoints?

[nocturne topaz]

Did you make a seperate venv?

[nocturne topaz]

Ctrl-C

[clear basalt]

wich LR scheduler is the best if i have a big dataset images to train a LORA for a face?

[fierce hollow]

Hello everyone, I want to upload additional files to Lora. I'm uploading them to the folder at the path stable-diffusion-portable-main\models\Lora. However, they are not displaying within the neural network itself. Can you please help?

[hazy agate]

Having a problem with UI, turned every extension off.
This is a state if i click on result.

This is after i click on cross.

This didn’t happen before; the picture was minimized as much as possible and was located in the corner of this field.

And the same problem if generating batch.

[nocturne topaz]

if they aren't appearing there, they are incompatible (i.e sd1.5 lora with sd-xl model)

[nocturne topaz]

does it do the same if you use incognito mode

[hazy agate]

Yes.

[wanton stirrup]

howdy, having an issue i couldn't find a solution for:
(currently using python 3.10.6, cuda 12.3)

it also shows this when pip is already 24.0

[vocal burrow]

what s the full log ?

[uncut jewel]

Hey! Anyone could help me regarding this issue when i'm trying to update ComfyUI ?

[vocal burrow]

it s telling you everything you have to do

you re probabably running comfyUI on a fat32 partition => no permissions suppport => git gets angry => need to declare safe.directory

[uncut jewel]

Alright i see, i run comfyUI on my Nas, maybe that's the reason..
However, i did make ComfyUI updates on the past, that's why i don't understand why now, it doesnt want anymore 😅

[vocal burrow]

git works in mysterious ways.... there are plausible explanations such as it depends if file xxx or yyy has been modified or not.

[uncut jewel]

Yeah i see, recently i add my root disk letters change for no reasons, it can be the cause.. Do you know any way how i can declare the safe directory so it can update properly ?

[vocal burrow]

to declare comfyUI directory as """safe""" aka ignore permissions (aka not a proper fix but as long as you don t dev for comfyUI or modify its files you should be safe).... you just have to run the command it tells you to

git config --global etc etc

[uncut jewel]

Ok i see, should i do "cmd" on my comfyui or venv folder to run this command ?

[vocal burrow]

it s system wide (not venv related)

open a cmd, copy paste the whole line, press enter, voila

[uncut jewel]

thanks my man, gonna try it out 🙂

[mighty topaz]

Guys i cant enable controlnet because when i click check and apply, then restart and again is off all the time is off and cant enabled

[hollow wraith]

i'm trying to re-run image generations from a few months ago but the results are always slightly different. i checked to make sure i use the exact same model and lora versions, generation settings (taken from png info sent to txt2img), clip skip settings, VAE, extensions and command line arguments, but the images still don't come quite the same. what else can i check still?

[nocturne topaz]

You are running forge, controlnet is built in already 🙂

The extension that needs the sd-webui-controlnet needs to be updated to use forge

[nocturne topaz]

gpu vs cpu seeds, xformers on vs off, discard penultimate sigma, k-diffusion quantization

[mighty topaz]

oh ok thank you

im new in forge i know they remove --xformes etc they no need it, so what i can on from this thats help me with 12gbvram 4070

[nocturne topaz]

none of them, should be fast out of the box

actually --xformers is still required

it's --medvram and the ilk that have been removed

[vocal burrow]

sdwebui forge should come with its build-in controlnet, you don t want to use those extensions, cf https://github.com/lllyasviel/stable-diffusion-webui-forge?tab=readme-ov-file#about-extensions

welp someone else already answered, I m too slow :p

btw @nocturne topaz I ve tried forge this week end. I get nowhere close to the advertised 30 to 45% inference speedup. At best I get 10% and it breaks my tensorRT engines. I ll give it another go this week to see if I messed up something.

[nocturne topaz]

10% ahead of tensorRT is good..

[mighty topaz]

ok so i need to write manualy in window --xformes

[nocturne topaz]

some people are getting 0 :S

[nocturne topaz]

in the webui-user.bat file on the commandline_args= line

[vocal burrow]

No I get 10% at best under normal circumstances. Going from 9 with xformers to 9,6 it/s with forge. I still get about 16it/s when using TensorRT with auto1111.

[mighty topaz]

i have via stability matrix so i dont touch files only in stability

[nocturne topaz]

ah right.. nah in that case go tensorRT.. I hate the fact that tensorRT has to be rebuilt each time you wanna try a new lora 😦

[mighty topaz]

thanks

[nocturne topaz]

but I can see why at that speed one would do it

[vocal burrow]

yeah it s a real chore. I probably won t switch to it.

[mighty topaz]

oh @nocturne topaz i add --xformers and

[nocturne topaz]

does it stop it from working?

[mighty topaz]

no i can work and generate images but i dont like errors always disturb me 😄

[vocal burrow]

that s expected

don t mind the triton is not available thing for sdwebui forge

[uncut jewel]

Hey again 🙂
I'm trying to use the IPAdapterFace checkpoint on comfyUI but have this issue.. Anyone knows what to do about it ?
It works well with the classic IPAdapter that's why i don't really understand what's happenning..

[mighty topaz]

ok but in a1111 is? because there this working,

[vocal burrow]

what do you mean ? your screenshot is from a1111 ?

[mighty topaz]

no no, i mean when im install first time a1111 i turn on xformers and works but today im install forge and cant use? this is a1111

so i think problem is with forge triton... to bad

[vocal burrow]

there s no real point using xformers with sdwebui forge

cf https://github.com/lllyasviel/stable-diffusion-webui-forge/issues/152

[mighty topaz]

yeh ok, i turn off it , i turn on this before because someone tell me here i can on xformers in forge 😄

[vocal burrow]

you can even force disable it with --disable-xformers

[mighty topaz]

yeh i udnerstand that:D.. but dont need i just remove text --xformers from it and works

[indigo ivy]

what commandline arguments should i use for an rx 6700xt?

have this rn --use-directml --medvram --opt-sub-quad-attention --autolaunch

[nocturne topaz]

Yeah that was me, as I have it on my setup, but its linux

[dense sandal]

I have my models on forgeUI and I want Auto1111 to load the same. (usually it's the other way around but as I have them in different drives I need to do it this way.) how can I do it?

[quick forge]

howdo i install a tar.zst file?

[solemn thorn]

how do i get rid of this error ?

i tried reinstalling aswell

[tiny abyss]

I'm getting an error when running a batch process in the Depth Extention

[tiny abyss]

Tried with CPU also, error still there

[vocal burrow]

use command line arguments --ckpt-dir xxx --vae-dir yyy --gfpgan-dir zzz --lora-dir www --embeddings-dir aaa --hypernetwork-dir bbb to set its directories

[vocal burrow]

what s your gpu ?

[fierce hollow]

Please tell me which models allow for artifact-free image creation at resolutions higher than 1024x1024 (for realistic human images). I've tried DreamshaperXL, and it works perfectly, but if the resolution goes above 1024, people end up looking like giraffes

[dense sandal]

ok I can't just use the models folder? do I have to do it for each one

[vocal burrow]

not to my knowledge, or you can set a symbolinc link for your model folder instead

[quick forge]

anyone know what this means?
Can't load /var/lib/shim-signed/mok/.rnd into RNG
4037A97D227F0000:error:12000079:random number generator:RAND_load_file:Cannot open file:../crypto/rand/randfile.c:106:Filename=/var/lib/shim-signed/mok/.rnd

[vocal burrow]

How to get High resolution results :

SD 1.5 and derivatives models are trained on 512x512 pictures and should give better results when trying to generate (before upscaling) pictures close to that resolution. As for SD 2 and derivatives, they are trained on 768x768, so same spiel. Aim for something close to that. and then upscale to some crazy resolution if you want. And SDXL works at 1024x1024 but theorically should be able to handle a much larger range of resolution (cf PS)

The usual workflow is :
1/ Do first step at low res (for fast and artifact free iterations)
2/ Upscale it in latent space (with hires.fix or img2img) to add some details while going higher res.
3/ Then upscale it one last time with ESRGAN based upscaling (cf Extras Tab) to make it 2k/4k without modifying its composition.

If you generate at high resolution in one pass (without hires.fix/upscaling) you're gonna get "duplicate"/"doubling" artefacts

PS : here's a (incomplete) list of some possible resolutions for SDXL:

• aspect ratio: 4:3 - 1152x896
• aspect ratio: 2:3 - 832x1216
• aspect ratio: 9:16 - 768x1344
• aspect ratio: 21:9 - 1536x640
• aspect ratio: 1:1 - 1024x1024
• aspect ratio: 3:4 - 896x1152
• aspect ratio: 16:9 - 1344x768
• aspect ratio: 5:4 - 1152x896
• aspect ratio: 3:2 - 1216x832
• aspect ratio: 4:5 - 896x1152
• aspect ratio: 9:21 - 640x1536

[fierce hollow]

Okay, thanks

[vocal burrow]

not without a full log

[fierce hollow]

Is there a way to fix the face? Sometimes (often) artifacts appear around the eyes, distorting and skewing them.

[fair oxide]

if you are using a1111, After detailer is a very good option https://stable-diffusion-art.com/adetailer/#Windows_or_Mac

you could install it from extension tb as well

[shut moat]

Hello, my images usually come out with somewhat pink spots. What could be causing this?

[quick forge]

ok, so I reinstalled Kubuntu, and setup Stable Diffusion, but for some reason now when I run it, Automatic1111 is just white, and not a nice stylized blue background or anything

what did I fuck up this time?>

[vocal burrow]

the 1st looks really grey, you might be using an incorrect VAE
2nd looks ok ish, red pink/blue artifacts are usually due to a combination of too low/high cfg value and sampler/sampler steps. So in order if you want minimal impact on the output, try to increase. decrease the number of steps, then the cfg, then the sampler itself. (Using a different VAE could also fix those)

[vocal burrow]

theme depends of the system, use the --theme dark to force the dark theme

depends of your budget and usage of it.

[quick forge]

ok that explains a bit.
How do i change my system to datk theme...because it's not theo nly thing thats wrong lol

[quick forge]

I just spent about $700 setting up a tower for SD, and I already had an RX6800

[vocal burrow]

dont know depends of your sytem, google should be able to answer that faster than me

[vocal burrow]

(that answer wasn t for you)

[quick forge]

looking it up now. oddly enough this is the first install to not be dark,

[quick forge]

i also replied to the wrong person lol

[fierce hollow]

How do I find out what I'm using? Sorry, I'm new to this neural network.

[muted flame]

I'm still having crashing problems at 95% after making the space

Could you confirm for me what 8 need to make at this step because maybe I made an error when updating it to the new card

Edit the webui-user.bat (right click), At the line COMMANDLINE_ARGS= You add: --use-directml --medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling
If your GPU has only 8GB Vram and you run into out of memory, you may need to change --medvram to --lowvram.
GPUs with less than 8gb Vram need: --lowvram
RX 5xxx cards additional need: --no-half

[fierce hollow]

I have stable diffusion portable main I understand it’s better to uninstall and install a1111?

[fair oxide]

i wouldn't recommend using portable webui, are you on windows and what graphics card do you have?

[fierce hollow]

Yes, I have Windows 11, Graphics Card: 3070Ti

[fair oxide]

ok, click the pinned icon on top of this chat window , (top right) and follow the guide under
[NVIDIA] Here is a quick Guide to install Automatic1111 Webui for Nvidia GPUs (Stable Diffusion) on Windows

[fierce hollow]

I'll try to find it now, I understand you're talking about the pinned messages in this chat, right? Sorry, I don't speak Russian, I'm communicating with you through a translator, I hope it won't cause you too much trouble.
And one more question, if you don't mind - I should uninstall my current version, right? It's bad, and my graphics card allows me to install a better one, right?

[vocal burrow]

looks fine to me, maybe for the last line it should say "RX5xxx cards and lower" instead of "RX5xxx cards" but that s about it

[vocal burrow]

yes the pinned comment in this channel

[fierce hollow]

#🤝｜tech-support message
I understand, you meant those messages, right) Okey

@fair oxide "Is this version significantly better than what I currently have? Right?

[vocal burrow]

because you have a "portable" version we can t know for sure what you have, and it s harder to update

[fair oxide]

that's definitely more versatile and feature full than portable

[fierce hollow]

Alright, then I'll do as you say and check YouTube after that. Thank you very much, you're a good person. I'll uninstall my version now and install what you suggested. After that, I'll follow your advice on adjusting the faces

[fair oxide]

np and good luck, if you have other questions feel free to ask

[vocal burrow]

you can save your models, embeddings, loras, etc aside and put them back in the newly installed version

[fair oxide]

like Aryetis pointed out, if you have previously downloaded files like checkpoints, LoRAs or anything else you should back them up before deleting the portable webui

[fierce hollow]

Yes, I've already transferred one model to my desktop, with which I can create realistic images. I tried using a few more, but the result was blurry, even though the prompts were the same everywhere.

[quick forge]

do i place this ip_adapter.bin file in the stable-diffusion-webui/extensions/sd-webui-controlnet/models/
https://huggingface.co/InvokeAI/ip-adapter-plus-face_sdxl_vit-h/tree/main

[fair oxide]

..\stable-diffusion-webui\models\ControlNet

[quick forge]

ok, so pth files go in the folder I mentioned?

[fair oxide]

both go into the folder i mentioned, and face id LoRA would go into Lora folder

[quick forge]

so why is this tutorial im reading telling me to put them in the other folder?

https://stable-diffusion-art.com/controlnet/

[fair oxide]

you can follow that guide too, but the folder i mentioned works fine too and quicker to access and organize all the controlnet models in

[quick forge]

will it still read models in subfolders within the models folder?

[fair oxide]

should work fine, if you have webui running then hit the refresh icon next to model box

[quick forge]

was just curious, it's not running yet

currently

[mossy zodiac]

Hello, I just generated an image with my face using Control Net / ip-adapter_face_id_plus, but there are still some differences. However, I like the pose, the clothes, etc. How can I modify this image to make it more closely resemble my face? I would like to do an inpainting of the face but with a reference image

[fair oxide]

what are the settings are you using for controlnet weight, starting step and ending step?

[fair oxide]

should be ok and you have face id lora mentioned in postive box?

[mossy zodiac]

,lora:ip-adapter-faceid-plusv2_sdxl_lora:1

[fair oxide]

you can try lora weight between 0.6-0.8 and in case you are using adetailer, turn that off

[mossy zodiac]

how I can change the lora weight ?

what is adetailer ?

[fair oxide]

type it in <lora:ip-adapter-faceid-plusv2_sdxl_lora:0.6>

[mossy zodiac]

oh ok nice to know

[fair oxide]

so you might not be using adetailer, but just to be sure, after detailer tries to correct facial blemishes, and if thats turned on by any chance that will impact faceid

[mossy zodiac]

oh a guy tell me to test this extension

but it's disabled because it make my webui craches

[fair oxide]

ok, one other thing about preprocessor resolution, just to experiment with, try a value of 512

[mossy zodiac]

Hmm, I still notice some differences, which is odd. But regarding my initial request, is it possible to use inpaint using a reference image?

[fair oxide]

can you throw an example of your source and output ?

[mossy zodiac]

I'll try to do it with another face

eyes problem but it's ok

I think on my own picture I have too much things behind me, maybe I have to take a picture of me with a wall behind me

[fair oxide]

for detection that's necessary if you have too much noise in the background

[proven prawn]

Im using a google collab notebook how can I add hires fix to it?

[quick forge]

right now i have A1111 and ComfyUI installed on a secondary SSD.

All of the models and everything are in A1111's folder and ComfyUI points to it.

Is there a way to move the models, in some organized manner, into a folder outside of these webui's, and have every other webui I install point to the separate models folder, so everythingcan use all the same models?

[worldly pagoda]

--ckpt-dir "Path/to/models/stable-diffusion" --lora-dir "Path/to/models/lora" --controlnet-dir "..." --embeddings-dir "..."

[quick forge]

I don't specifcally know what to do with that information

[worldly pagoda]

add it to parameters in webui-user.bat

with the path to your models folder

[quick forge]

im on linux

linux/amd

[worldly pagoda]

ok just do the same thing with ur file

webui-user.sh

[fierce hollow]

I saw a section with videos on the website https://civitai.com/videos. Can you please tell me how to make similar videos?

[mighty topaz]

YT animated diffusion, deforum etc

[dense sandal]

it didn't work...

[fierce hollow]

@fair oxide @vocal burrow Please advise, after installing the new version of the neural network, it takes significantly longer to load images. Is this normal or am I doing something wrong? The settings in webui-user are the same as they were in the previous version: 'set COMMANDLINE_ARGS=--autolaunch --xformers --medvram-sdxl --no-half-vae'.

[vocal burrow]

what does the console log says ?

[dense sandal]

its because I used the wrong / right? I changed it

its loading again.

[vocal burrow]

it might be

[dense sandal]

yes its good now

[vocal burrow]

same, what does the full console log says ?

[fierce hollow]

Are you referring to this? Or where can it be found?

[vocal burrow]

yes this. I already see lots of error. so restart it, start a generation. copy paste the whole text in a .txt and drop said file in here

[fierce hollow]

To start the generation and immediately copy all the contents and send it to you?

[vocal burrow]

yes everything from the start

[fierce hollow]

The second upload went relatively quickly, but the computer lagged heavily around 95%. I observed this when using ControlNet

[quick forge]

Automatic1111 has the venv folder, and when i run either A1111 or ComfyUI I have to activate the source?
I didn't have to do that when I was on arch and I could jus run webui.sh.
Do I always have to activate that source when I run a UI?

and why is A1111 the only one with a venv folder?

[vocal burrow]

what s your gpu ?

[quick forge]

Linux/AMD RX6800

[vocal burrow]

(not you)

[quick forge]

oh, poo

[fierce hollow]

3070ti,

[vocal burrow]

no you don t have to activate the venv when using the UI it does it by itself.

[vocal burrow]

(I dont think it s the only one having a venv, it might be the only one where it s obvious that it s using one tho)

[quick forge]

it doesn't though, if i just run the launch.py or main.py without activating the source i just get an error about not being able to use the gpu

[vocal burrow]

because that s not how you re supposed to launch it.

[quick forge]

how am i supposed to launch it then?

[vocal burrow]

you should be using webui-user.bat for windows and webui.sh for linux

[vocal burrow]

and what settings did you use for that generation ? (screenshot the webpage)

[hazy agate]

Question.
Are there any settings in UI section that can change and fix this problem?
#🤝｜tech-support message
Or maybe this is a wrong channel?

[quick forge]

for some reason my computer is suddenly slow and choppy

[vocal burrow]

ok I don t see anything wrong in your settings nor the console. speeds do look a little slow, if you have some overclock settings try removing them.

[quick forge]

odd... that didn't work for me earlier.
But now it does, but it doesn't open the webui automatically in the default browser

[vocal burrow]

never seen this issue, did you clean your cache, cookies, etc ? Does the console says anything weird ?

[quick forge]

so how am I supposed to launch comfy then? by also using something that isn't the python script?

[vocal burrow]

you can change that in the settings or use --autolaunch command line argument (set it in webui-user.bat or webui-user.sh depending of your os)

[vocal burrow]

hmmm haven t launch comfyui in months but for it I do remember launching main.py, launcher. py or something along those lines directly indeed.

I m not at home right now. I m about to commute back home

so see you in a bit

[quick forge]

yeah that's how as far as i know. but then I do have to activate the source, but also have to point it to the venv folder in the A1111 directory so it uses those models and such

[vocal burrow]

oh no don t do that for comfyui

sharing venv between programs is not a good idea.

there should be a extra_model_paths.yaml file in comfyui to indicate which folder to use for models and such

(and with that I m off)

[hazy agate]

Used different browser, still an issue.

Nothing unusual in console.

[neon ruin]

tried this guide, followed everything to the letter but still got this: OSError: [WinError 126] The specified module could not be found. Error loading "C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\torch\lib\cublas64_11.dll" or one of its dependencies.

[ornate elk]

did you moved the file in this location?

[neon ruin]

i created the copy and pasted it in like it said

[ornate elk]

and replaced the old one?

[neon ruin]

yes

[ornate elk]

can you check if the file is in there?

[ornate elk]

are you on the latest AMD driver?

can you show me the rocm install and the path entrys ?

[wanton stirrup]

sorry for the late reply the full log is:

[ornate elk]

outdated webui version, wrong installation method, and most important wrong python version

[wanton stirrup]

context is that i reinstalled windows but kept the files, it was working before but now... yeah

isn't 3.10.6 for sd?

[neon ruin]

Well this is weird, i definitely made the path entries but now everything looks different

[ornate elk]

python 3.10.6 would be okay, but you have installed the 32 bit version, but only 64 bit is supported

do you want to update this webui version or do you want a clean installation?

[wanton stirrup]

update you mean, reinstall python correctly xd and then update through the ui?

[ornate elk]

updating your webui, cause your on 1.5.1 and latest version is 1.7.0, but your webui doesnt have git so you cant update it easily until i give you the steps to fix it

but yes, please uninstall python

and install python 3.10.11 64bit
then tell me when your ready

[ornate elk]

looks okay

[neon ruin]

AMD making me want to kill myself

[ornate elk]

is it in the right path menu?

[neon ruin]

everything is such a struggle lol

[neon ruin]

is what in the right path menu?

[wanton stirrup]

got both python and git

[ornate elk]

perfect

[neon ruin]

to be clear, you want the ZLUDA folder there right?

[ornate elk]

@wanton stirrup
To fix your webui update go into the stable-diffusion-webui folder and click in the File Explorer bar (not searchbar) and type cmd and hit enter.

Then use the following commands:
git init git remote add origin https://github.com/AUTOMATIC1111/stable-diffusion-webui.git git fetch --all git reset --hard origin/master git branch --set-upstream-to=origin/master master git pull
Then delete the venv folder.
Then launch the webui-user.bat

[neon ruin]

not the sd-zluda?

[ornate elk]

yes, the zluda on c

not the webui

[neon ruin]

that's what I thought yeah, looks right

[dense sandal]

what should I put for my controlnet models? sorry

[ornate elk]

--controlnet-dir

[dense sandal]

can I put two of those? I have some in the extension and some in the models/controlnet folders..

[ornate elk]

whats your gpu?

[neon ruin]

well something changed

[ornate elk]

nope you need to move them in one folder

[neon ruin]

Wondering if I formatted incorrectly, directions said add space

maybe you meant new line?

[ornate elk]

at which step?

[neon ruin]

Now go to "C:\SD-Zluda\stable-diffusion-webui-directml\modules" and there search the shared_init.py
Right click and edit it with IDLE 3.10.
After the line "from modules.onnx_impl import initialize as initialize_onnx" make a space and add:
torch.backends.cudnn.enabled = False
Then save and close the file.

[ornate elk]

ohh yes

a new line

will edit it

[fierce hollow]

Which sampling method do you recommend for the most realistic image? (Currently using 'Euler a') Thank you in advance

[ornate elk]

can you show how it looks like?

thats how it looks for me

[ornate elk]

change it so it looks like mine

then it should work

[neon ruin]

just did

probably just me but I followed "space" literally

[ornate elk]

yea that was my fault sry 🥲

[chrome rampart]

hi all . any idea why i am getting msg about missing xformers ?

understood in forge you dont need argument statement?

[wanton stirrup]

it worked, tyvm!

[ornate elk]

nice, no problem 🙂

[ornate elk]

@neon ruin did it worked now?

[ornate elk]

dont know

its webui-forge

it should use it by default

[neon ruin]

it does open

[chrome rampart]

@ornate elk ill try to figure it out thks

[neon ruin]

having difficulty changing checkpoint though, just goes to "processsing"

[ornate elk]

ah make sure its not an sdxl model

thats bugged rn

[neon ruin]

should just be 1.5

[ornate elk]

yes any 1.5 model will work

[maiden haven]

i'm trying to get a smooth workflow between kohya and webui going and im having a hard time flipping between one to the other. i''ll get one side of it (sd or kohya) working, usualy i have to either delete the venv and pull everything or activate the env and manually compile torch with cuda etc. but then when i stop webui and move over to kohya, now kohyas venv complains, almost always that torch wasnt compied with cuda.. so i do the same thing there.. then back to sd and .. .same.. what am i doing wrong? am I potenetially not getting venved into the other apps venv? only thing i can think of

[vocal burrow]

hmmmm try to delete everything in %appdata%\..\Local\Temp\gradio

[maiden haven]

oh im also wondering if its because of SD extensions installing/changing package versions i didnt realize those go through and basically deal with their own deps, but i dont ge twhy i would ever wind up with a not-CUDA compiled torch..

[quick forge]

does venv have to be inside of the stable-diffusion-webui folder?

[hazy agate]

Nope. Going to reinstall and try again.

[vocal burrow]

:/ let me know if it fixed it

you can also try to simply delete venv before that

[hazy agate]

venv?

[quick forge]

the venv folder

[vocal burrow]

the venv folder in stable-diffusion-webui

[fierce hollow]

Which sampling method do you recommend for the most realistic image? (Currently using 'Euler a') Thank you in advance

[hazy agate]

It worked, LESSGO!!!
Thank you very much!

[nocturne topaz]

Dpm++ 2M Karras

[fierce hollow]

Initially, it was set up worse than what I'm currently using. I'll try it now, but it's worse than what I'm currently using

[vocal burrow]

Models and prompt are way more important than the sampler used. Euler A tends to do ""simple / less detailed"" stuff but there are many counter examples.

[fierce hollow]

I use the following prompts to increase quality: 4k, realism, detailed skin, ultra realistic, 8k, natural lighting

[nocturne topaz]

which model - some models that will make the results worse and not better

[neon ruin]

@ornate elk just an update, haven't had any issues on this version yet

[nocturne topaz]

Agreed, I just answered the question as it was posed 😛 - Euler A tends to give a smoother result, which is great for certain image types, but when looking for skin detail it's not what you are after

[vocal burrow]

most of the time yes

[nocturne topaz]

again.. there will be a model out there that is the exception

[ornate elk]

nice, and its fast right ? 😄

[neon ruin]

It is yeah

this might have salvaged my build

Was beginning to have severe buyer's remorse going with AMD

[nocturne topaz]

install headache vs more performance and extra vram for same $$$....

[ornate elk]

what ive tested with zluda so far:
Controlnet works, (only depth, mlsd and ip-adapter not), also Adetailer dont work

[neon ruin]

if I crash again and see the same error then I'll be concerned again

[neon ruin]

not to mention the 7900XT looks great 😎

[vocal burrow]

That's why I tend not to answer to "what's the best model/prompt/lora/etc" questions. Usually I go with "there's no best xxx just like there's no best brush in photoshop". A more honest answer would probably be "AI is kind of a black box so try different things out, smash stones together until sparkles happen" but that's not what people want to hear :p

[nocturne topaz]

A business analysts favourite answer "It depends...."

I've tried most of the realism models out there - you'd have to go a long way to find one that doesn't do it's best detailed skin using DPM++ 2M Karras or Restart

[quick forge]

does ubuntu do relative paths like ../

[tropic mango]

Hello, 1 year ago, I generated images with Dreambot, since that period the dreambot channels were deleted, how can I find my pictures ? And where can I prompt again ?

[maiden haven]

if youre confused on what versions need to match what this is super useful https://www.felixsanz.dev/articles/compatibility-between-pytorch-cuda-and-xformers-versions

It taught me things like.. "Unfortunately I have not found any compatibility table between PyTorch versions and xFormers versions. If we do pip install xformers, it will install the latest xFormers version and also update PyTorch to the latest version, something we don't want." i didnt realize so many of these packages would trigers rollup/backs

[quick forge]

I renamed the webui folder to "Automatic1111" and now it can't use my GPU unless I rename it back?
RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check

what do I need to change to correct that so it can use the new folder name?

[maiden haven]

im just trying to train a new LORA on 40 images.. this is a 4090, i think i went too high somewhere lol .. way too many repeat checks (40-100?) or did my gpu get unplugged again nothings complaining

[quick forge]

why is my webui running at 0.0.0.0 ?

[nocturne topaz]

you set --listen

which allows access from outside the computer

[quick forge]

yup i did

just looked hat up

im having an issue with not being able to use my gpu again though

but it's after I renamed the foldere to Automatic1111

however if i delete venv it'll rebuild it and then will run A1111 just fine

[fair oxide]

hi

i keep getting a cloudflare error when i try to do image-to-video using this code

but i can generate images fine in cmd

import requests

response = requests.post(
f"https://api.stability.ai/v2alpha/generation/image-to-video",
headers={"authorization": f"Bearer sk-MYAPIKEY"},
files={"image": open("./kittens-in-space.png", "rb")},
data={
"seed": 0,
"cfg_scale": 1.8,
"motion_bucket_id": 127
},
)

print("Generation ID:", response.json().get('id'))

(obviously have my own api key in tere)

There are several actions that could trigger this block including submitting a certain word or phrase, a SQL command or malformed data.</p>
</div>

[nocturne topaz]

im guessing here, but is it something to do with the file being passed in? (i.e. is it expecting an open call in there?)

[fair oxide]

no idea, this is my first time using stable diffusion

i literally took that code directly from thier website 💀

https://platform.stability.ai/docs/api-reference#tag/v2alphageneration/paths/~1v2alpha~1generation~1image-to-video/post

[nocturne topaz]

ah ok.. no idea sorry, hopefully someone with API experience can come along soon and help you

[fair oxide]

my api peeps 🥲

[craggy scaffold]

so is dreamstudio down again? anyone else having issues?

[nocturne topaz]

I mean if dreamstudio is down, that could be due to an API issue that could be causing Rose's issue as well... outside chance

No idea sorry!

[quick forge]

ok for some reason now when i hit generate there isn't really a progress bar at the top of the (gallery?) it just kind of blinks then disappears, but the progress is still running in the terminal window

which can probably b seen with the public link
https://cc560588709f001910.gradio.live

[unique bear]

What's the command line to changing localhost port?

[quick forge]

hmmmm it doesn't like the public link for some reason, when i enable share=true it bugs out the progress bar on generations

is there a way to end the stable diffusion/a1111 process without closing the terminal window

[ornate elk]

--port

[quick forge]

i don't understand

oh nvm

[unique bear]

thanks ser

[maiden haven]

If i'm working with independent characters do I want 1 dataset 1 model = 1 char, or 1 char = 1 dataset of many chars = 1 model, i can't use "Instance Prompt" if I have multiple chars in one model i feel like

[nocturne topaz]

CTRL + C can end the process

[quick forge]

ive tried that and it doesn't do anything

[nocturne topaz]

weird.. thats usually enough to end it 😦

[quick forge]

it should take me back to the command prompt right?

[nocturne topaz]

should.. may take a while as it will try to finish the current image first

[craggy scaffold]

does anyone know if stability.ai is just not maintaining dreamstudio anymore? it seems like clipdrop is working well, but according to reddit dreamstudio has been down for the last 2 days with no word from them and nothing even noted as wrong on the API status page. pretty frustrating

[nocturne topaz]

#🔧｜finetune

[desert totem]

Should I move the question over there? Doesn't seem like fine tuning since it doesn't work vs needs tuning...

[nocturne topaz]

That is where the people that do checkpoint and lora training hang out (or at least thats what I hear)

[desert totem]

then that's where I needs to be, thanks!

[quick forge]

can a source be de-activated?

[jolly gorge]

anybody know how to optimise forge for mac m1?

[quick forge]

how do i install stable video diffusion?

[languid tinsel]

Did something change in the SDXL? Only 4 out of my 100 Lora's are showing in the UI. (they were working before, I even generated thumbnails and keywords for them)

[quick forge]

ok where tf is the ffmpeg path in ubuntu?

[vernal sundial]

i have a pc with an intel i7 12700k 48 gb of ram but a crappy gtx 1660 super.

i know the gpu is crap but somehow stable diffusion is way faster on my m2 2022 macbookair. on that it will be like 1.4 s/it whereas the pc will have 5s/it

[old parrot]

Switch to SD Forge, my 1660S runs great

[vernal sundial]

ok I'll try that

[steady thistle]

Any ideas on this? I'm using AMD. I have no idea if there's a workaround for this to run Stable Cascade.

[unique bear]

Do I just add --port 7861 if I want to run it on a separate port in the webui.bat file?

[nocturne topaz]

Cascade not supported in a1111 yet outside of a janky plugin that isnt great

[pseudo flame]

does anyone know how to setup a good face swap? willling to pay $

[nocturne topaz]

https://www.youtube.com/watch?v=_A7PraBHyg0

[woeful cloud]

Hey, not sure if this is the right place to ask or not but I am using automatic1111 webui and I have a unet .safetensors file, I have this put into my stable diffusions models and it's loaded in automatic1111 but when I try to generate an image it's really jumbled up and then the final output is a solid grey, I know it works because the hugging face inference api lets me generate on it

Edit: I am able to generate images now but they aren't correct/generating in the right format, please ping me if you reply!

[unique anchor]

Any working Google Colab Notebook for Kohya SS?

[nocturne topaz]

Define right format??

[nocturne topaz]

They may have one in #🔧｜finetune

[severe coyote]

hi, i deleted my textual inversion.py from the modules folder, i get this massage in the cmd , please help

[nocturne topaz]

I would either recommend doing a git pull, or just heading to github and downloadin the zip file, get the specific file and put it in the right place

[severe coyote]

how do i do git pull? what i need to write?

[tranquil vigil]

Can somebody tell me why do I recieve this error while trying to start DragGAN?

[nocturne topaz]

Not sure - check the issues on the github

[nocturne topaz]

Try downloading the file manually from github then

[severe coyote]

thanks. did it now

[lost crescent]

ComfyUI - I just cannot get WAS Nodes to work!!!

[tranquil vigil]

Hello, sorry but I am not good at coding, The first line of coding shows the issue that I am facing while trying to launch DragGAN.. So, if somebody could explain me in steps on what to do in the solution, it would be great

[vocal burrow]

everything seems to be explained in the very screenshot you just posted ?

[tranquil vigil]

yeah, but I am no good at any program line cmds.. so, from that perspective I need help

[vocal burrow]

you open a cmd

cd xxxx where xxxx is the dragcan directory path

then conda activate stylegan3

then .\scripts\gui.bat

[fair oxide]

https://youtu.be/W6YLIyA3Kco

[tranquil vigil]

typing in conda actiavte stylegan3 always gives me this error

[vocal burrow]

because you don t have conda (anaconda) installed

[tranquil vigil]

i have

oops.,., I dont have

this one?

[vocal burrow]

looks like it

[tranquil vigil]

i instealld anaconda, but still facing the same issue

[vocal burrow]

You didn't add conda to your path then

[tranquil vigil]

can you explain me the steps plz?

[vocal burrow]

https://www.datacamp.com/tutorial/installing-anaconda-windows
cf "How to Add Anaconda to Path (Optional)"

just know that by installing anaconda you now have (at least) two python on your system

so for stuff such as A1111 you might need to precise which python to use, using webui-user.bat set PYTHON= variable

@unique bear what s your gpu ?

and what settings are you using for your tests ?

[unique bear]

rtx 4090

[unique bear]

should I try cudnn file change again?

[vocal burrow]

no don t think that s needed anymore

what s the full console log for both your test case ?

[unique bear]

To create a public link, set share=True in launch().
Startup time: 31.9s (prepare environment: 18.5s, import torch: 4.1s, import gradio: 1.1s, setup paths: 1.4s, initialize shared: 0.3s, other imports: 0.4s, setup codeformer: 0.1s, list SD models: 0.3s, load scripts: 3.2s, create ui: 2.1s, gradio launch: 0.2s).
Loading VAE weights specified in settings: D:\AI Projects\stable-diffusion-webui-master\models\VAE\vae-ft-mse-840000-ema-pruned.ckpt
Applying attention optimization: xformers... done.
Model loaded in 6.8s (load weights from disk: 1.7s, create model: 0.5s, apply weights to model: 4.0s, load VAE: 0.2s, calculate empty prompt: 0.2s).
Advanced elements visible: False
56%|█████████████████████████████████████████████ | 69/124 [00:29<00:23, 2.32it/s]
Total progress: 53%|██████████████████████████████████▊ | 69/129 [00:31<00:27, 2.18it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 30/30 [00:12<00:00, 2.34it/s]
100%|████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:03<00:00, 1.40it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 35/35 [00:19<00:00, 1.20it/s]
0: 640x640 1 face, 7.7ms
Speed: 3.1ms preprocess, 7.7ms inference, 18.4ms postprocess per image at shape (1, 3, 640, 640)
100%|██████████████████████████████████████████████████████████████████████████████████| 13/13 [00:06<00:00, 2.14it/s]
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
mediapipe: 1 detected.
100%|██████████████████████████████████████████████████████████████████████████████████| 13/13 [00:01<00:00, 9.45it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 35/35 [00:35<00:00, 1.02s/it]
100%|██████████████████████████████████████████████████████████████████████████████████| 30/30 [00:03<00:00, 9.29it/s]
100%|████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:02<00:00, 2.34it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 35/35 [00:08<00:00, 1.93it/s]
0: 640x640 1 face, 7.7ms

[vocal burrow]

why the lines about mediapipe and XNNPACK ?

screenshot the whole webpage

surely you forgot to include something in there

[vocal burrow]

1/ I said screenshot the webpage
2/ Why do I see a lot of things on this screenshot that you didn t mention when I asked for the FULL log

there s controlnet stuff and probably reactor too there

lots of stuff that can slow down things

[unique bear]

My bad

[quick forge]

how do I tell which version of SD I'm running, and how can I update it if necessary?

[vocal burrow]

it tells you so at the very bottom of the page and at the very top of the log

"Version: 1.7.0"

I d say if you don t need, remove dreambooth, redelete venv and try again

and do test runs without controlnet, animatediff, adetailer stuff enabled

Woops I mixed up cases again