🤝｜tech-support

[ornate elk]

Never launch the webui.bat

Always and only launch the webui-user.bat

[solemn girder]

Ï edited the second to bottom one then launched the top

So deleted the Venv again?

[ornate elk]

Yep

Also open up a cmd and type
Pip cache purge
To remove the old torch stuff

[solemn girder]

Ive ran it like that before

[solemn girder]

In the command line?

[upbeat gyro]

Traceback (most recent call last):
File "E:\Forge\webui\modules\call_queue.py", line 57, in f
res = list(func(*args, **kwargs))
TypeError: 'NoneType' object is not iterable

Im now getting this error

[ornate elk]

no just in a cmd

[solemn girder]

Gave me an error

[ornate elk]

whats the error

[solemn girder]

Well now it worked, eternal curse!

[ornate elk]

thats how your webui-user.bat should look like

[solemn girder]

Mines
set COMMANDLINE_ARGS=--medvram --autolaunch --no-half --xformers --medvram-sdxl --no-half-vae --pip-cache-purge
git pull

No spaces

Set python and such too

[ornate elk]

remove --medvram --no-half --pip-cache-purge

[ornate elk]

nope

[solemn girder]

No I meant I had them

[ornate elk]

?

the args?

[solemn girder]

I didnt send my full file in the copy paste

[ornate elk]

oh you had the pythond and venv paths set?

thats not needed at all

[solemn girder]

@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set SAFETENSORS_FAST_GPU=1
set COMMANDLINE_ARGS=--autolaunch --xformers --medvram-sdxl --no-half-vae
git pull
call webui.bat
--gradio-img2img-tool color-sketch

[ornate elk]

looks better, but --gradio-img2img-tool color-sketch wont work if its not in the line with the other args

[solemn girder]

Some tutorial at some point probably told me to add it

[ornate elk]

bad tutorial then xD

i can take a look if you have the lnk

[solemn girder]

I probably didnt understand half of it so some of the code isnt there

[ornate elk]

but i doubt someone adds commandline args at the bottom

[solemn girder]

Idk what I wanted fixed

[ornate elk]

yea maybe

[solemn girder]

I installed this thing around when it first came out

[ornate elk]

you dont need --gradio-img2img-tool color-sketch so you can also remove it

[solemn girder]

Its been stopped here for a bit

[ornate elk]

yea it will install a lot of stuff in the backround

when its done it should be on torch version 2.3 and xformers 0.0.23

webui version will be 1.9.3

[solemn girder]

Should I update that?

[ruby oracle]

Whenever I try in startups stable diffusion with the run.bat I keep on getting this

[ornate elk]

you can later do that with this command: python.exe -m pip install --upgrade pip
or just install python 3.10.11 64bit and then delete the venv folder again. (it wont download everything again as the files are cached)

[ornate elk]

hey, thats not the recommended way to install the Stable Diffusion Webui
checkout my Install guide in the Pinned Messages of this channel, or linked here:
#🤝｜tech-support message

[ruby oracle]

I will go ahead and do that thank you for directing me to the right spot

[solemn girder]

Those 3 supposed to go in the same one?

[ornate elk]

thats a single command, that dont goes in any file but has to be run inside a cmd

[solemn girder]

in args

CMD worked

AH NEW OPTIONS!

[upbeat gyro]

?

[ornate elk]

a lot of new stuff xD

you can now also use SDXL models

[solemn girder]

Yeah, I just have zero clue how to SDXL, also I get blue sponge

[ornate elk]

oh whats the model?

[upbeat gyro]

Hey, when you're done, can you help me through dm's?

[solemn girder]

It was auto picked but its an old one

[ornate elk]

cant help much with forge :/ but you can show me the whole cmd log

[upbeat gyro]

Alright, wait a moment

[solemn girder]

Model is called final

[ornate elk]

for SDXL you need to download a SDXL model from civitai.com
and you also need an sdxl vae file.

[ornate elk]

can you try an other one?

[solemn girder]

Yeah

Wait, Ive always had my models in the stable diff folder

Do I need to change that

Okay new ones are working

But do I need to place the loras in loras now that Ive updated? (They didnt work when I did that before.)

[ornate elk]

loras and lycoris and everything lora related goes into models/lora

[solemn girder]

I feared so...

[ornate elk]

they should work normaly

[solemn girder]

I just have to figure out which are loras

[upbeat gyro]

This is the errors i am getting

[solemn girder]

And which are checkpoints

[ornate elk]

every file thats larger than 2gb (or 1.98gb) is a checkpoint (model)

[solemn girder]

So put all under 2gb into loras?

[ornate elk]

yep

if you had them in the models/stable-diffusion folder then move them

thats only for the large models

[solemn girder]

Then I select the loras from the lora menu right?

[ornate elk]

exactly

[solemn girder]

Only 13 of them show up in the menu

[ornate elk]

in the prompt you then get a lora:name:1

[ornate elk]

they only show up if a compatible model is selctet. if you downloaded them in the last days, you may downloaded some SDXL loras by accidant

[solemn girder]

A lot of them are old

[ornate elk]

ah okay

try the refresh button

also check the CMD for any errors

[solemn girder]

No cmd errors

But refreshing gives one on screen

[dense tapir]

No problem and thank you for your reply

[ornate elk]

what did you tried, it seems the error comes from jsut loading the model

[upbeat gyro]

The only thing i tried is booting up webui

[solemn girder]

Maybe anything in this is an error, gives me this when I try and refresh

[ornate elk]

would need to see the full cmd log

did you used any embeddings (textual inversions) before? because they dont go into models/lora

[solemn girder]

nope

[solemn girder]

It goes off screen

Imma try restarting

I pulled the files when the program was started

[ornate elk]

ohh okay

[solemn girder]

Hmm, I cant generate anything either

It just keeps loading stuff

[ornate elk]

what does the cmd shows?

[solemn girder]

A load of idk

[ornate elk]

its generating

[solemn girder]

It began when I got pinged

[ornate elk]

make sure the webui is whitelisted in any browser adblocker

like ublock origin etc
Gradio middleware error is mostly caused by browser blocked stuff

[solemn girder]

Ah, course

[solemn girder]

This didnt work, Im still missing loras

[ornate elk]

and your sure they are loras?

[solemn girder]

idk

they are old

so possibly not

[ornate elk]

they could be also hypernetworks

or embeddings

[solemn girder]

mmm

They worked when they were in the stable diff folder years ago

[ornate elk]

can you show me the files?

[solemn girder]

Theyre all safe tenors

[vernal hare]

Does anyone know what causes this error?
_MultiThreadedRendezvous: <_MultiThreadedRendezvous of RPC that terminated with:
status = StatusCode.UNAVAILABLE
details = "Received http2 header with status: 504"
debug_error_string = "UNKNOWN:Error received from peer {created_time:"2024-05-27T14:23:29.595436297+00:00", grpc_status:14, grpc_message:"Received http2 header with status: 504"}"

Randomly started getting it

[pale echo]

Can anyone explain me how i can fix this message?
inference_feedback_manager.cc:114] Feedback manager requires a model with a single signature inference. Disabling support for feedback tensors.

[ornate elk]

what webui do you use?, do you have a full cmd log?

[ornate elk]

would also need to see the full cmd log

[pale echo]

INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
W0000 00:00:1716819778.436678 278532 inference_feedback_manager.cc:114] Feedback manager requires a model with a single signature inference. Disabling support for feedback tensors.
W0000 00:00:1716819778.442801 278532 inference_feedback_manager.cc:114] Feedback manager requires a model with a single signature inference. Disabling support for feedback tensors.

u mean that? there is nothing else that happens.

[ornate elk]

what tool are you triying to use?

[vernal hare]

I'm just interacting with the API through collab

[pale echo]

ADetailer. It just started to break all of the Sudden.
Should i try updating my Stable Diffusion first?

[ornate elk]

okay, api and collab, cant help with both xD sry

[ornate elk]

yes and the extensions

in the extensions tab

[pale echo]

extensions i updated, maybe that broke it.

[ornate elk]

whats your webui-version?

[pale echo]

Zluda, version 1.9.3-amd-13-g517aaaff

and for extensions update this error happens:
*** Error checking updates for kohya_ss
Traceback (most recent call last):
File "C:\SD-Zluda\stable-diffusion-webui-directml\modules\ui_extensions.py", line 113, in check_updates
ext.check_updates()
File "C:\SD-Zluda\stable-diffusion-webui-directml\modules\extensions.py", line 194, in check_updates
for fetch in repo.remote().fetch(dry_run=True):
File "C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\git\remote.py", line 1015, in fetch
res = self._get_fetch_info_from_stderr(proc, progress, kill_after_timeout=kill_after_timeout)
File "C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\git\remote.py", line 889, in _get_fetch_info_from_stderr
output.append(FetchInfo._from_line(self.repo, err_line, fetch_line))
File "C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\git\remote.py", line 460, in _from_line
old_commit = repo.rev_parse(operation.split(split_token)[0])
File "C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\git\repo\fun.py", line 378, in rev_parse
obj = cast(Commit_ish, name_to_object(repo, rev))
File "C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\git\repo\fun.py", line 184, in name_to_object
raise BadName(name)
gitdb.exc.BadName: Ref 'febc5c5' did not resolve to an object

[ornate elk]

you installed kohya_ss as extension but its not an extension xD

its an standanlone tool

that should not be in SD

[pale echo]

ah ok i can delete it from there then right?

[ornate elk]

yep

delete it from the extensions folder

then you might need to delete the venv folder too, if the error is still there

@pale echo if you want to train loras with AMD, that only works with OneTrainer (its also a standalone tool)

https://github.com/Nerogar/OneTrainer

[manic grove]

Can anyone confirm that REST API is currently not working?

[heavy perch]

@ornate elk hi again 😂

[ornate elk]

hey,
install python 3.10.11 64bit

[heavy perch]

don't work with 3.10.6 anymore?

[ornate elk]

it should but the update could fix the error

[heavy perch]

kk

[worthy nymph]

ok people, sorry to change chanel, but ive been told to come here to fix my problem, its driving me crazy

as said before, there is a problem in there apparently I cant find. Ive tried unistalling/installing torch, Pytorch, check PATH, now try with a different version of Python, and nothing changed I did check the shm.dll file which is apparently weird out, but it gives nothing to me
I dont know where to look anymore nor what to do

[vernal hare]

It started working again, looks like the api endpoint was down

[vocal burrow]

what is generate_art.py ?

[worthy nymph]

my file for generating a stable diffusion picture

im partially using the help of chat GPT. Its rough around the edges, but I just wanna learn for now

It told me to create a notepad file, create a bunch of lines with some prompts and sh*ts , name it "generate_art.py" and put it somewhere on my computer, then run command prompts with python

[pale echo]

is it normal after deleting and reinstalling it, that it takes ages to launching the web ui?

[vocal burrow]

yeaaaaaah as expected from chatgpt

don t trust chatgpt

that program is probably hot garbage, I m not surprised it does not work

use already existing and proven solutions instead

[worthy nymph]

well I cant really always trust people either, they arent allocated to answers all the times either, so eitherway im fucked if I wanna learn

[ornate elk]

yes, its like the first zluda install. it can take 10-30 mins, just let it load

[worthy nymph]

thousands of different youtube tutorials with poor quality and 30min-an hour for bad results is just a waste of time, I dont call that learning

[ornate elk]

check out the pinned messages of this channel.
there are guides to install Stable Diffusion localy (for nvidia and AMD)

these are the best and easiest

[vocal burrow]

No offense but copy pasting from chatgpt is not learning.
If you want to learn how to use SD, use already existing programs.
If you want to learn how to write a SD program, trace back the error and read tons of documentations. Error is on line 1 so you re probably trying to import something that is not installed.

[worthy nymph]

wait I think ive found the problem

its the location of torch

• its in python 312

which is weird, because ive still reinstalled it at least 20 times eversince I changed to Python 310.6

should have taken the right torch version

[lost crescent]

"Control Filter List" in ComfyUI - how does it work?

[worthy nymph]

I was actually able to correct all my problems with ChatGPT

[ornate elk]

You still should install it the correct way with my guide if you don't want to have problems in the future.

[worthy nymph]

problems like what, if I may ?

[ornate elk]

What can occur:
Models won't work or loras are incompatible.
Extensions won't work.
High vram usage and slow generation speed

[ornate elk]

That looks not that good. It installs software that wasn't updated since 2 Years.
Runway 1.4 model is very old, it was the second public model available in 2022

[left juniper]

I tried installing missing nodes but they're conflicting in plugin manager

I installed comfyui controlnet, now the only ones left are: Text box and SDXLPromptStyler, but i'm not sure if it's gonna break if I try installing the conflicting nodes

[wind thunder]

even when i put --skip-torch-cuda-test it didn't work i think i wrote it bad
can someone tell me the right thing to put i have this now --xformers --medvram --no-half-vae

[ornate elk]

please use my install guide from the pinned messages

and dont add --skip-torch..

[pale echo]

The Problem with ADetailer is still there. Strangly only on the 2nd Setting (mediapipe_face_mesh_eyes_only)

[wind thunder]

i just did it but is encounter this error ?

[upbeat gyro]

I got ForgeUI To work 😄

[ornate elk]

make sure your nvidia driver is updated

then open up a cmd and type
pip cache purge
then delete the venv folder and relaunch the webui-user.bat

[wind thunder]

i just added skiptorch idk ...
and it worked

i mean for now in the cmd its progressing

[ornate elk]

--skip-torch-cuda-test will not fix anything

it makes your webui run on slow CPU mode

[wind thunder]

oh ok

[ornate elk]

if its run trough you have to do the steps i mentioned above to fix it

but before that remove --skip-torch...

[pale echo]

@ornate elk i believe i found my problem. i guess my eye model i used on adetailer wasnt compatible anymore with the latest update of it

[ornate elk]

ahh you mean the face mesh thing?

[pale echo]

exactly. i had used mediapipe_face_mesh_eyes_only all the time, i guess they get rid of it?

[ornate elk]

maybe, if yo uwant one only for eyes, i can recommend that one:
https://civitai.com/models/330727/full-eyes-detection-adetailer

oh i meant that one:
https://civitai.com/models/150925/eyes-detection-adetailer

[pale echo]

tested both, both do well. can u recommend me any hand ones? i find it hard to get a good hand result, they allways look like from a silent hill game xD

[ornate elk]

nope, for hand there isnt a good one. but if you want good hands try Pony models, they are incredibly good at it

[wind thunder]

how can i add a VAE model ? on stable vanilla i dont see option

i mean i know how but can we ?

[ornate elk]

put the vae into models/vae
then in the webui go to
Settings/ UserInterface/ Quicksettings
There add sd_vae
then hit apply and reload ui

then you get a vae dropdown next to the model dropdown

[hollow ocean]

Well last night I went and did a clean install after that issue and it was fixed all night installed new extensions reinstalled models and loras. So I obviously reloaded and closed the session multiple times and it was fine search and refresh worked and there were no issues. Now its broken again. Disabled all extensions so nothing should be blocking it. Any other suggestions?

I should also add that I am not seeing any errors when launching so I really dont know what to look at as far as what may be causing it.

[ornate elk]

you can reset your ui by deleting the ui-config.json

also still make sure the webui is whitelisted in any browser adblocker

disabling extensions wont fix problems they cause

if they install stuff in the venv it will always be broken (no matter if disabled or enabled)

[fair oxide]

does anyone wanna tutor me for NSFW images? i wouldnt ever post it, but if you are good at it please let me know. DMS are open

i have pretty good general knowledge in stable diffusion

just wanna see how some pros do it

[worthy nymph]

ok so whats Automatic1111 exactly guys ? As I said earlier im new, im not familiar with all the terms yet sorry

[ornate elk]

hey, Automatic1111 is a webui (a cmd programm used with a Gui for easier usage)
it uses the Stable Diffusion process to let user generate images based on text prompts.
As there is no official "Stable Diffusion" Programm for End Users, Automatic1111 made the Webui so anyone can easily setup and use Stable Diffusion

[cosmic furnace]

Yet again i get this error

on 1st try to do generation

img2img

NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.
Time taken: 59.2 sec.

i was trying to make pixel art of 64x100 😄

mb its not trained to do so or whut? o.0

hm i guess it cant indeed, cant make it do anything like arrow o.0

ah seem to work now when i altered promt a lil bit and generated in txt2img first and then back there... still weird , no ?

[solemn moon]

hi, i have two paths for ControlNet

1 - \stable-diffusion-webui-master\models\ControlNet
2 - \stable-diffusion-webui-master\extensions\sd-webui-controlnet\models

Both of mine have models in them, trying to free up space, do i need models in both

[ornate elk]

no you only need them in models/ControlNet

so they wont disappear if you delete the extension

[solemn moon]

thaannkkkss you

[round edge]

where do I put an upscaler pth file I put it in the gfpgan folder but it does not show up

[calm cairn]

i feel bad asking questions because they always have simple answers

I want to machine-generate captions for an image
I'm sure I used to know how because I tried to make my own LoRA and succeeded at multiple textual inversions/embeddings, but I'm looking at this "BLIP/Deepbooru caption" option and I really don't remember what it does/how to work it.

Can someone tell me how to caption a single image?

[round edge]

https://ibb.co/QdDHm87 what am I supposed to do? (my image keeps getting compressed by discord)

[ornate elk]

models/ESRGAN

you can create it

[round edge]

how do I add pth models to this screen

[ornate elk]

are they in the models/esrgan folder?

and have you relaunched the webui-user.bat after putting them in there?

[calm cairn]

I have a sprite from an old video game and I want to make automatic/machine captions for it

I think i have korya installed but I don't remember anything about it

How do I get auto captions? Thanks in advance

[proper pike]

Does anyone have a patreon I can sub to right now that can show me how to get a smoother pixel animation? Because my current workflow isn't working.
Trying to achieve like inner reflections recent pixel animation. Mine aren't working properly

[round edge]

Tried that nothing

[vivid token]

hi, this might be a dumb question, but where do my .part files go? I have a .safetensor and a .part. I'm running on webui, and if you need more specs I'll get them to you. I'm just dusting off this program after not using it for over 10 months

nvm I'm tripping, turns out it wasn't done downloading

[cursive cobalt]

I assume when it says "Checkpoint trained" and then "Base model: Pony", its just needs to be placed in the Lora folder, ye? (Pack in question https://civitai.com/models/331116?modelVersionId=464939)

[calm cairn]

hi, I posted a little earlier today, I don't think anyone was around who could help

I want to know how to use Automatic1111 to caption pictures (img 2 text), I know it's possible when making LoRA/embeddings, I just don't know how

Anybody know how? Thanks in advance

[vocal burrow]

Img2txt tab -> interrogate (clip)

[cursive cobalt]

(If you need a screenshot of it) The Clip and the box on the right. Use the clip.

[calm cairn]

hmm, do I need to update? I don't have img2txt tab

ah never mind, I see a paperclip in img2img--though I probably should update too

Thanks to both of you! It's parsing now

[vocal burrow]

img2img, my bad was on phone

[plush apex]

how do you update the webUI I cant seem to find that update.bat script?

[solemn moon]

ever had a generation get stuck, like everytime , pls help

[plush apex]

nvm solved it.
for the record you litterally just git pull

[cursive cobalt]

What i do, is i open a new tab, open the webui console window, spam "Enter" a few times", go back to the webUI browser tab, open the webui console window, spam enter a few times, and go back to the empty tab, and open window and spam enter a few times... It sounds weird, but oddly.... It works for me

It is a very, very weird method yes, but if it works, it works.

"It Just Works" - Todd Howard,

[solemn moon]

I'll try that

[solemn moon]

now it freezes here
Already up to date.
venv "C:\Users\o0\stable-diffusion-webui-master\venv\Scripts\Python.exe"
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Version: v1.9.3-1-gddb28b33
Commit hash: ddb28b33a3561a360b429c76f28f7ff1ffe282a0
CUDA 12.1
Launching Web UI with arguments: --xformers --no-half --medvram --api
[-] ADetailer initialized. version: 24.5.1, num models: 10
*** Error loading script: main.py
Traceback (most recent call last):
File "C:\Users\o0\stable-diffusion-webui-master\modules\scripts.py", line 508, in load_scripts
script_module = script_loading.load_module(scriptfile.path)
File "C:\Users\o0\stable-diffusion-webui-master\modules\script_loading.py", line 13, in load_module
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 883, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "C:\Users\o0\stable-diffusion-webui-master\extensions\openpose-editor\scripts\main.py", line 14, in <module>
from basicsr.utils.download_util import load_file_from_url
ModuleNotFoundError: No module named 'basicsr'

---

ControlNet preprocessor location: C:\Users\o0\stable-diffusion-webui-master\extensions\sd-webui-controlnet\annotator\downloads
2024-05-28 17:16:59,863 - ControlNet - INFO - ControlNet v1.1.449
*** Error loading script: main.py
Traceback (most recent call last):
File "C:\Users\o0\stable-diffusion-webui-master\modules\scripts.py", line 508, in load_scripts
script_module = script_loading.load_module(scriptfile.path)
File "C:\Users\o0\stable-diffusion-webui-master\modules\script_loading.py", line 13, in load_module
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 883, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "C:\Users\o0\stable-diffusion-webui-master\extensions\sd-webui-depth-lib\scripts\main.py", line 10, in <module>
from basicsr.utils.download_util import load_file_from_url
ModuleNotFoundError: No module named 'basicsr'

---

*** Error loading script: m2m_ui.py
Traceback (most recent call last):
File "C:\Users\o0\stable-diffusion-webui-master\modules\scripts.py", line 508, in load_scripts
script_module = script_loading.load_module(scriptfile.path)
File "C:\Users\o0\stable-diffusion-webui-master\modules\script_loading.py", line 13, in load_module
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 883, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "C:\Users\o0\stable-diffusion-webui-master\extensions\sd-webui-mov2mov\scripts\m2m_ui.py", line 23, in <module>
from modules.ui import (
ImportError: cannot import name 'create_sampler_and_steps_selection' from 'modules.ui' (C:\Users\o0\stable-diffusion-webui-master\modules\ui.py)

---

17:17:01 - ReActor - STATUS - Running v0.7.0-b7 on Device: CUDA
Loading weights [54ef3e3610] from C:\Users\o0\stable-diffusion-webui-master\models\Stable-diffusion\meinamix_meinaV11.safetensors
Creating model from config: C:\Users\o0\stable-diffusion-webui-master\configs\v1-inference.yaml
Applying attention optimization: xformers... done.

[ornate elk]

What's your GPU?

[solemn moon]

1660ti

gpt told me to do this so i did 😅

[ornate elk]

That could fix the depth lib issue yes

But you have to install it in the venv

[solemn moon]

oh lord

[unique bough]

Need help with it, I want to generate a video using deforum and when I click "generate" it says an error and that's it

[mortal compass]

What's in your terminal when this occurs

Yo @ornate elk, I wanna chance my directories from pic 1 to pic 2. I reckon I'm doing something wrong cause it ain't working lol

chatgpt tells me to chance versions of fastapi and pydantic cause they are causing a mismatch in regards to this

But I'm not sure if it's correct

[mortal compass]

requirements doesn't even have pydantic in the txt file

[mortal compass]

you have installed your a1111 in c:\users\druma

Your root installation should be directly in C:\

Otherwise it will or may cause problems

[unique bough]

ok, ty, I`ll try

[mortal compass]

If you are not experience with this stuff I highly recommend you watch videos on how to install whatever you want to run. There's a lot of ways to fuck up your installation

[unique bough]

I installed according to the tutorial on git

[mortal compass]

It's easy to miss one small step. I did many times and it made the program unbootable. I succeeded when I started to look up different videos and guides to actually learn how to install it rather than just reading steps

Anyway, that's just how I learned

[unique bough]

ty mate, have a nice day

[ornate elk]

the second screenshot doesn look right

@mortal compass it should look like this in forge:
`@echo off

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

@REM Uncomment following code to reference an existing A1111 checkout.
set A1111_HOME=H:/AI-Generator/stable-diffusion-webui <---- Only Change this Line to your Auto1111 Path
@REM
@REM set VENV_DIR=%A1111_HOME%/venv
set COMMANDLINE_ARGS=%COMMANDLINE_ARGS% ^
--ckpt-dir %A1111_HOME%/models/Stable-diffusion ^
--hypernetwork-dir %A1111_HOME%/models/hypernetworks ^
--embeddings-dir %A1111_HOME%/embeddings ^
--lora-dir %A1111_HOME%/models/Lora ^
--vae-dir %A1111_HOME%/models/vae ^
--controlnet-dir %A1111_HOME%/models/ControlNet
git pull
call webui.bat`

[mortal compass]

Where did you find out that this is what it should look like? I made a fresh install just a couple of days ago, and the first picture is basically unedited, except for the args

[ornate elk]

i tested it for myself, and thats how it works by removing the "@REM" stuff

and adjusting the lines like in the example

[mortal compass]

I'll see if I can make it work

[mortal compass]

Does this look correct to you?
`@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS= --pin-shared-memory --cuda-malloc --cuda-stream --xformers

@REM Uncomment following code to reference an existing A1111 checkout.
@REM set A1111_HOME=Your A1111 checkout dir
@REM
@REM set VENV_DIR=%A1111_HOME%/venv
set COMMANDLINE_ARGS=%COMMANDLINE_ARGS% ^
--ckpt-dir C:/Ai_Root/Stable-Diffusion_Forge/webui/models/Checkpoints ^
--hypernetwork-dir C:/Ai_Root/Stable-Diffusion_Forge/webui/models/hypernetworks ^
--embeddings-dir C:/Ai_Root/Stable-Diffusion_Forge/webui/models/embeddings ^
--lora-dir D:/Program_Files/Ai_stuff/lora ^
--vae-dir C:/Ai_Root/Stable-Diffusion_Forge/webui/models/VAE ^
--controlnet-dir C:/Ai_Root/Stable-Diffusion_Forge/webui/models/ControlNet
git pull
call webui.bat`

[ornate elk]

looks okay

[mortal compass]

something about pydantic

I feel like this is one of those issues where I'm missing something obvious cause I'm not into programming xd This feels like it should be easy to do but no matter what people say on different threads or what chatgpt says, I'm running into errors

[ornate elk]

put albumentations==1.4.3 and pydantic==1.10.15 in the requirements_versions.txt file

then relaunch

[mortal compass]

Ran pip install requirements, everything seemed to install as it should, but I noticed an error at the end of the process:
Installing collected packages: pydantic, protobuf, albumentations Attempting uninstall: pydantic Found existing installation: pydantic 2.7.1 Uninstalling pydantic-2.7.1: Successfully uninstalled pydantic-2.7.1 Attempting uninstall: protobuf Found existing installation: protobuf 4.25.3 Uninstalling protobuf-4.25.3: Successfully uninstalled protobuf-4.25.3 Attempting uninstall: albumentations Found existing installation: albumentations 1.4.7 Uninstalling albumentations-1.4.7: Successfully uninstalled albumentations-1.4.7 ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. mediapipe 0.10.14 requires protobuf<5,>=4.25.3, but you have protobuf 3.20.3 which is incompatible. Successfully installed albumentations-1.4.3 protobuf-3.20.3 pydantic-1.10.15
I ran the bat file, and it looked like it was installing protobuf, and now it has launched correctly, everything is at it's place! I take it as the error has been taken care of?

[ornate elk]

looks good

[mortal compass]

Thanks for sharing your green circuit fingers! So, why wasn't those requirements already in the text file? Is it bad programming from forge's devs side or something else?

[ornate elk]

np, its caused by a controlnet dependency

[cosmic furnace]

Csio?)

Did that guy shared how he did that very speedy Webui with zluda amd?)

[ornate elk]

nope, but the outputs where not the same quality like with the zluda version we use now

[solemn moon]

worked, im back in, now to see if it generates

would be cool if you could drop the text file into automatic, i know theres other ways, just dropping is kickass

[cosmic furnace]

Oh... Not that good then i guess.)

[west mirage]

Does anyone know how to fix? I tried reinstalling git but that did not work

i also tried git config --global --add safe.directory "*" but that does not work either

[solemn moon]

turns out to hold true, lmao thanks bruh

[cursive cobalt]

Its a super weird "fix" xD But it works

[ornate elk]

did you followed the guide from the pinned messages of this channel?

[west mirage]

no

i did it like i always do it by following automatic1111 guide

on the repo

[cursive cobalt]

Apparently, that guide does so much wrong

[ornate elk]

okay, i would still recommend following my guide.
but first let us know what De syntaxis van de bestandsnaam, mapnaam of volumenaam is onjuist means

[west mirage]

The syntax of the file name, folder name, or volume name is incorrect

[ornate elk]

okay, can you try using the guide from the pinned tabs and make a new folder for it?

[frail bobcat]

Does anyone know why underscores don't work anymore? Dynamic prompt doesn't recognize my wildcards anymore but changing the underscores with two tildes in settings works. So something is overwriting the use of underscores??? Anyone have an idea?

[west mirage]

i should clarify that this error is in webui.bat and not webui-user.bat

i want to use webui.bat since it always has been faster

[ornate elk]

Its not recommend to use the webui.bat
Also webui-user.bat should be faster depending what's inside it.
So what's in your webui-user.bat? And what's your GPU?

[west mirage]

RTX 3070ti - 8GB

[ornate elk]

Then add --xformers --medvram-sdxl --no-half-vae
To the webui-user.bat at the line commandline_args=
And you have the best performance you can get for that card in Auto1111

But it only works when launching the webui-user.bat

Because the webui.bat doesn't have launch args

[west mirage]

doing that now

well, it works

ty

[ornate elk]

No problem 🙂

[left juniper]

Hi. is it incorrect to download PNGs from civitai and load them into confyui to see the workflow OP used? when i do this, the output is completely distorted and only looks like a silhouette of the rough sketch of what's it supposed to be.

I make sure to include all the checkpoints and loras it's using but still get a messed up output

[chrome rampart]

hey all. question. Is it smart to activate restore face option in SD-forge and were do i find it ?

[brazen sun]

Hello there, small question, its been a while since I have used this, has there been any updates on using it on amd or is still the automatic1111 thing that barely uses it?

[ornate elk]

Hey, there was a huge improvement for AMD cards as Zluda released in February

You can find an install guide for it in the pinned messages of this channel

[brazen sun]

oh great! I'll look at it. Thanks!

[ornate elk]

Np, if you have any question feel free to ask

It more than doubles the speed of directml

And uses less vram so sdxl works easily too

[brazen sun]

wow, and here I thought I was going to have buy another graphics cards for more speed

[ornate elk]

What's your GPU?

[brazen sun]

radeon rx6600

[ornate elk]

That will work with Zluda 🙂 as 3 others already got it to work here

[brazen sun]

I am seeing a video tutorial and it seems I will have to do a diferent thing related to a "hip sdk" thing

[ornate elk]

Don't follow the video guides. There is currently no correct one

Thats why I made the guides here

For zluda

[brazen sun]

oh then ill look them up, they are pinned right?

[ornate elk]

But yes hip SDK is needed

[brazen sun]

Okayp I was just installing it

[ornate elk]

Here:
#🤝｜tech-support message

[brazen sun]

it gives a 215 error, and I have the latest amd driver installed

any idea?

[ornate elk]

Would need to see the full cmd log

[brazen sun]

I am trying again and seems to be working, if there is an error again ill try and get it

how do I get the log from this?'

[ornate elk]

ohhh

[ornate elk]

try installing it again but disable the visual studio plugin installation option

[wind thunder]

i've putted the VAE in the good files but i dont see it in my SD

i checked the setting for so it always choose the model i selected

but still i dont see it

?

[ornate elk]

in the webui go to
Settings/ UserInterface/ Quicksettings
There add sd_vae
then hit apply and reload ui

[brazen sun]

seems to work, now its blue, ill continue reading the guide

[ornate elk]

blue?

[brazen sun]

the blue here used to be red

[ornate elk]

ohh you installed the AMD pro driver :/

[brazen sun]

oh

[ornate elk]

in the guide it says to dont do that

[brazen sun]

do i have to unistall it or install it over it?

[ornate elk]

i think you need to uninstall it before installing adrenalin again

[brazen sun]

wait I have to install adrenaline all over?

[ornate elk]

i think so

[brazen sun]

darn, well ill get to it, brb and thanks again

[ornate elk]

np

also restart after the driver install

[brazen sun]

this one I have to unisntall right?

[ornate elk]

yep

[brazen sun]

and do I have to restart after unstalling too?

[ornate elk]

only if it says so

[brazen sun]

ookayp

[brazen sun]

I keep this like this right?

and the path like this right?

[ornate elk]

yes

looks correct

[brazen sun]

and the 4.5 step is if I have below a 6800 right?

great, and now it aint blue

and sorry for asking so much but what does this mean

and i have had this error

nvm just read the important step

[viscid dagger]

How do i use SD3 via the discord?

[brazen sun]

nvm nvm still same issue

[ornate elk]

ohh do you have an amd cpu?

sry for the late response

[brazen sun]

nope, intel

[ornate elk]

this can be ignored

[brazen sun]

I have an amd gpu and intel cpu

[ornate elk]

okay, you need to move the rocm libs files into the right folder

you have to download the right files for your gpu and then:
4.5 Go into C:\Program Files\AMD\ROCm\5.7\bin\rocblas folder. There copy and rename the library folder to old_library Open the .zip file and drag and drop all files of the library folder into the library folder. Not into the old_library folder.

[brazen sun]

I did that part yes

Last time I didnt restart but this time I did

(if it works now ill say dont worry)

or is there extra steps for intel cpu?

[ornate elk]

nope

[ornate elk]

restart is needed after the rocm library steps

[brazen sun]

and mostly out of curiosity, is there extra for amd cpus?

[brazen sun]

did it this time, seems to at least be taking longer

[ornate elk]

there is an fix if you have an amd cpu with igpu

thats why i asked

[brazen sun]

oh okayp

also, can this be ignored too?

[ornate elk]

yep

all normal

xD

[brazen sun]

seems to be downloading a thing now (I opened before as said in the tutorial=

)*

and now its open, now I have wait a while for it to load right?

Works or not, thank you so much for your patience

this taking a while doing this little animation is normal and the console doing nothing is normal as well right?

[ornate elk]

Yep it should load

[ornate elk]

Yep normal

[brazen sun]

so I just, wait, right?

[ornate elk]

Yep

15-40 minutes

After pressing Generate

If the button is grayed out and the cmd doesn't shows any error its good

[brazen sun]

I mean the interrupt and the skip are grayed out

and it put this

[ornate elk]

looks very good

just wait 🙂

[brazen sun]

oh nvm it just worked

quite a creepy pic but that makes sense with no lora lmao

[brittle lantern]

Quick question does the stable assistant (the site) use stable diffusion 3 to generate those images? (Idk where else to ask this)

[brazen sun]

one last question if anyone knows, any way I can access my local webui from my phone?

[regal ridge]

hi! i'm new to stable diffusion. I've been getting an error - ModelNotFoundError: No module named "modules". any help would be appreciated

[ornate elk]

hey, can you show the complete cmd log?

[regal ridge]

Microsoft Windows [Version 10.0.19045.4412]
(c) Microsoft Corporation. All rights reserved.

C:\Users\AJ20L\Downloads>python app.py
python: can't open file 'C:\Users\AJ20L\Downloads\app.py': [Errno 2] No such file or directory

C:\Users\AJ20L\Downloads>

I apologize. That was another error I got

I tried using python app.py on the CMD & that's what I got. i'm just trying to open the program & start it for the first time

[ornate elk]

if you checkout the pinned messages of this channel,
there is my complete install guide for Stable Diffusion Webui localy
try that and test if it works

its not against the TOS, but i cant help you with API stuff sry

Adetailer is not an upscaling tool, its used for fixing faces, eyes, hands etc, it uses Inpainting and automaticly detects the parts

what settings did you used?
like resolution, steps etc

[regal ridge]

sure thing. i'll check it out. i learned about stable diffusion from youtube so i probably should have looked for the discord first. tyvm for the reply!

[ornate elk]

np 🙂 there are a lot of old and outdated or just bad Install tutorials out there, so your at the best place by following my install guide from here

okay, yea for that you can just use Hires fix as an upscaler with the Esrgan4x+ or the 4x Ultrasharp upscaler model

hires fix is in auto1111 by default, also the Esrgan4x upscaler

4x ultrasharp has to be downloaed manually

yep and if you have a hr_step set it to 10
but it depends on the model

[regal ridge]

a quick question: where can i find the install guide? i don't want to scroll through the wrong category for minutes 😅

[ornate elk]

click on the Pin needel on the top right of this channel, then scroll down to the Nvidia Guide if you have an Nvidia GPU

damn xD

mayb you can find Face restoration as an option?

[regal ridge]

I see it. thanks again! 👍

[ornate elk]

Can you try Real-Esrgan4x+
Or Latent Nearest Exact
Or Bicubic

Oh

Yes, there is an option in the Auto1111 settings to get a Json file for each generated image

Hmm I'm not at the PC rn, can check tomorrow

Maybe it was only save as text file

Or search for generation data

I'm off, I'll take a look at it tomorrow

[regal ridge]

Just wanted to say tyvm! As it turned out I did everything wrong last time. It all worked out & I was able to use the program. 👍

[delicate citrus]

hi guys

What PC configuration I need to make videos?

I wanna start work with SD profissionaly, I will buy a PC

can someone sugest me?

I talk about video but is a default configuration to do anything, I imagine that video is more heavy

[solemn moon]

as long as you setup has a gpu i think

the bigger the gpu the faster it works

[delicate citrus]

ok

can you give me a floor? a minimal GPU, for example?

[solemn moon]

well, i have a 1660ti, and although it works, i want a 4060

[delicate citrus]

nice, thanks ^^

I will show you a example

[solemn moon]

might settle for the 30's range, im a laptop type of guy

[delicate citrus]

@solemn moon I need do something like this too
https://www.reddit.com/r/StableDiffusion/comments/18m3aa4/webcam_drawing_to_architectural_concept/

it's for achitecture propose

[delicate citrus]

sorry my english is bad, 30 cm of dimension to put in a laptop? laptop interest me

maybe the company give me one : D

[delicate citrus]

https://github.com/s-du/FocusPocusAI

[solemn moon]

A 30 series gpu

[delicate citrus]

ah ok

nice ^^

[solemn moon]

I’m sure there are lots of tools even if they are not for architecture

[delicate citrus]

yes, if you know anyone and wants share be free, I apreciate ^^

I see now a model so cool

it's for signal language

I was learning is very interesting

[solemn moon]

Found this with a simple search, easier than creating your own assets

https://github.com/ashawkey/stable-dreamfusion

[delicate citrus]

I will check it, thanks!

yes is very usefull

@solemn moon what do you think about 3050?

[solemn moon]

Honestly I don’t know much, but, I’m sure it’s fine, I mean, it must be the bigger the better 😅🤷‍♂️

[delicate citrus]

ok haha I understand the concept

[solemn moon]

I will probably get this next

[delicate citrus]

nice

hey

I discovery now SD.Next

I was out of the updates

next can run controlnets on SDX models?

I'm in doubt

@solemn moon

[solemn moon]

Bro control nets are so frustrating, I end up doing all basic controls with SDXL, then use 1.5 models and 1.5 control nets to do anything more

[delicate citrus]

I cant run on SDX only 1.5

if next can run controlnets with SDX will be very cool

[solemn moon]

There are SDXL control nets, not many I don’t think 🤔 I think for architecture you would only need canny and line art, for most things

I tried to find it, can def remember seeing it but I uninstalled a lot of extensions lately and now can’t find it for you 🫠

Maybe something like SD Next or something that enhances the user interface, def saw it b4, and now it’s gone 😅

Have a feeling you might be able to add something in there too 🤷‍♂️ just a guess

Probably:
, .json
or
; .json
🤔

[subtle quarry]

does controlnet work with ponyxl?

i'm struggling hardcore with it

[delicate citrus]

preprocessor? there nothing there

[subtle quarry]

does there need to be a preprocessor...?

with 1.5 i've never had to use one.

[delicate citrus]

I think yes

[subtle quarry]

i'll give it a shot.

[delicate citrus]

I have almost certain that is necessary

ok ; )

[solemn moon]

Noted, Must be the extension idea 😅

[subtle quarry]

nope, but i haven't refreshed the seed yet

[delicate citrus]

try it

mark open pose the SD will pick the preprocessor and model for you

man

you are using the wrong preprocessor

for track body is other model I think

can you show me the list?

@subtle quarry '

oh

XL

dont run controlnets I think too

or not all the controlnets...

[subtle quarry]

It runs controlnets, i've gotten it to work before

just ponyxl is being a headache.

[delicate citrus]

ok

what is ponyxl? a checkpoint:?

[subtle quarry]

mhm. it works a good bit different than sdxl though

[delicate citrus]

you want show me the list of preprocessors?

mabe I recognize the right for you

@subtle quarry

[delicate citrus]

open pose full

is what you need

full*

[subtle quarry]

still not workin

[delicate citrus]

or "open pose", but i think its the ful

my god LOL

dont recognize

[subtle quarry]

i'll just check google real quick.

[oak wing]

#🤝 Can Anybody tell me how can I generate videos using stable diffusion

[delicate citrus]

you will need a yt tutorial

will help more

[oak wing]

Can you share the link please

[solemn moon]

https://www.youtube.com/watch?v=HbfDjAMFi6w

[oak wing]

Thanks

[subtle quarry]

hey so i'm trying to get my computer to run a1111 from another profile on the computer and it keeps making me add new exceptions and it's driving me up the wall

nvm, got it figured out. i had to do it anyway but it's working now.

[formal trout]

Hi, in Comfy, extra_model_paths.yaml, how to add a extra model path other than the Auto1111 one?

so it loads checkpoints from both

[subtle quarry]

why is controlnet ignoring me?

no matter what model i use, it just absolutely fucking refuses to listen.

gonna try 1.5 and see if it's just an sdxl issue.

[solemn moon]

SAME VIBES

I envisage, a SD3 model, with all extensions included and all extensions universally compatible

[visual orbit]

Hey guys, on a 3090 in forge I get about 3.4 to 3.7 it/s in ponysdxl 25 steps sde ++ karras 2m. I run xformers, cuda stream e.t.c. I am running on an undervolt of 893mv( around 330 watt) at 1860mhz and a +250 boost on memory.(9705) Any way I can speed this up more? My GPU is zotac amp extreme holo

[delicate citrus]

try with 1.5 and you will know if is the model or other thing

[formal trout]

I tried to do something like and got an error

[subtle quarry]

i think you have to flip the slashes

use \ instead of /

that's what i remember

you also need to include what drive it's from
Like C: or D: etc etc

so the path should look like C:\path\to\the\folder

[subtle quarry]

1.5 ignores it as well.

gonna try updating extensions

[delicate citrus]

It's a good idea

[subtle quarry]

god dammit! WHy is it not working???

I'm going to try another model that isn't open pose.

still isn't working

i'm getting kinda frustrated.

can anyone help?

Please?

[subtle quarry]

i tried deleting the venv and it did nothing

[solemn moon]

Driver update?

Is it “enabled” 😅😂

[subtle quarry]

..i guess updating the drivers wouldn't hurt

and yeah , it's enabled

[solemn moon]

was my issue at one point

[subtle quarry]

and it's also registering shit so like i dont know

i dont know what the problem is, but hopefully updating drivers will fix it.

i really don't feel like making a bug report...

nope. that didn't do shit.

gonna try comfyui i guess

it works in comfy ui. This is so stupid.

[formal trout]

It's Linux, directories are like that

[subtle quarry]

so are windows directories, but i still have to flip them

they should look like this

[formal trout]

Oh I think I might have the answer, when you use the image with the line dummy, you don't have to use the preprocessor, set it to none.

[subtle quarry]

i started with that.

[formal trout]

Or use an image for the preprocessor to work and generated the dummy

it works here

Why do you get two more images and they look black?

you should get only one with the dummy

[formal trout]

you should always do like that, if you make it preprocess the dummy it will certainly not work

[subtle quarry]

hmh, alright lemme mess around real quick.

[formal trout]

like that

[subtle quarry]

oh. NOW it decides it's gonna work.

alright that's 1.5 figured out

now time to try sdxl

[formal trout]

I think it was that all along, it happens lol, you have to check the setup

[subtle quarry]

this is what i did for 1.5

well what i just now did

and that's what got it to work

so i'm gonna try the same method but for sdxl

oookay it's ignoring it

gonna try with just the dummy and not the preprocessor

okay it's still ignoring it

hmh.

maybe it's the openpose model i'm using... im not sure though.

[formal trout]

try the one I'm using, thibaud_xl_openpose

[subtle quarry]

that's what i used

i'm trying to get the ijmage to finish but it seems to be stuck.

...ahhhh.

it's confusing layers.

it things a leg is an arm.

so it was making it kinda right, but it got confused.

[formal trout]

well good, it is working

[subtle quarry]

i'm gonna have to figure out how to get the open pose model right, but

[formal trout]

yeah better use an anime one for anime I guess

[subtle quarry]

did a little bit of editing on it, hopefully that somewhat fixes it

and back to ignoring it outright. Hm.

i love how the interrupt button doesn't work half the time.

adds an extra minute to the process of ending an image early.

[formal trout]

yeah it takes some time to do something, sometimes skip is faster I think

If you do Generate Forever

[subtle quarry]

i do not, i generate one at a time

[formal trout]

or various

[subtle quarry]

it's not working again.....might have to stick to comfyui for sdxl cnet.

[formal trout]

If you do complicated poses don't expect to get the exact pose, also try with non anime for that openpose model

[neat bay]

hey good morning ! i m so so excited for the HUG and stability course more than you think but when i play for it it askes me for 180 dlr, ik it will give it back to me once i finish but the problem is i live in Egypt and paying with dlr is suspended in my country how can I apply now

[wide pecan]

Hi, I am trying to reach out to Stable Audio for API integration for enterprise plan, I have sent serveral emails to your official email address and there is no reply, can anyone reply me back (Mandy), we really wish to start the cooperation asap

[torn girder]

hey guys im getting blurry images in response using stable core, why?

[rain arch]

Hello! Long time no chat. I've been tinkering with ComfyUI again and I have a question about schedule prompts. Specifically, is there any way possible to grab the positive and negative prompts from a given frame in FizzleDorf nodesand save the data when running multiple prompts? Please see the example images above. I have an input going in to where the "normal" set of scheduled prompts (like below) normally go, and have not been able to find a node that reads those values

"0":"A Toucan with butterfly wings, hopping gracefully through the meadows with a kaleidoscope of colors",
"1":"A castle built in the shape of a flower pot",
"2":"A Hiroshima Peace Memorial, a somber yet impactful example of modern Japanese architectural resilience",
"3":"Interstellar Wanderer, long-range interstellar exploration vessel, designed for extended missions into uncharted space",
"4":"A sailor ascends the mast to adjust the crow's nest, a solitary figure against the backdrop of a cloud-streaked sky"

I'd like each of those prompts to be read as the name of the prompt so I can save individual frames with the prompt as the filename. Is that at all possible? There is a reason for this madness, I promise.

[brazen sun]

one question if anyone knows, any way I can access my local webui from my phone?

[obsidian carbon]

Any comfui people here any good with working with Face Detailer? I need an issue resolved xD

[obsidian carbon]

TL;DR, a lora that has a clear face in it can't be detected with the face detailer in comfyui (impact pack)

[neon swallow]

Hello @vocal burrow. Dream Studio is having issues generating images. Could you help us? CC @north cedar

[vocal burrow]

Looks like there s some network outage
https://dreamstudio.com/api/status/

best is probably to wait it out and contact support directly from their website if it lasts for more than a couple of hours.

[lean ocean]

@vocal burrow can you help me

Traceback (most recent call last):
File "C:\Program Files\Python310\lib\runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "C:\Program Files\Python310\lib\runpy.py", line 86, in run_code
exec(code, run_globals)
File "C:\Users\harla\Desktop\Kohya-Demo\kohya_ss\venv\Scripts\accelerate.EXE_main.py", line 7, in <module>
File "C:\Users\harla\Desktop\Kohya-Demo\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 47, in main
args.func(args)
File "C:\Users\harla\Desktop\Kohya-Demo\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 995, in launch_command
args, defaults, mp_from_config_flag = _validate_launch_command(args)
File "C:\Users\harla\Desktop\Kohya-Demo\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 928, in _validate_launch_command
raise ValueError(err.format(mode="bf16", requirement="PyTorch >= 1.10 and a supported device."))
ValueError: bf16 mixed precision requires PyTorch >= 1.10 and a supported device.

What is the fix to this Errorcode?

[regal ridge]

hi! does anyone know what the best model to use for Radeon RX 5500 XT (32 RAM)? I typed Cat to practice loading an image but I got an error. I'm also wondering if maybe there was an extra step I had to take to get images to generate.

[ornate elk]

You need to select fp16 on kohya

[brazen sun]

A bit of a stupid question but does regional prompter still work?

[ornate elk]

Should work

[brazen sun]

some times it just generates one character

[regal ridge]

updateL I got "disable-nan-check" error & when I put it now the generated images are all colored black.

[brazen sun]

for example

did I prompt it wrong?

[lean ocean]

that didnt fix it i tried that first

[ornate elk]

What are the other settings?

What's your GPU?

[ornate elk]

Hey, which install guide did you followed?

@regal ridge I don't think a 5500XT is supported by Zluda AMD, so you should install the Directml Webui from my second AMD Guide

[regal ridge]

I download it from AMD: ([AMD] Here is a quick Guide to install Automatic1111 Directml Webui for AMD GPUs (Stable Diffusion) on Windows:). surprisingly the website opened up & I was able to use the default Model to load some images. however, everytime I download new Models to test them out, I get this error:
NansException: A tensor with all NaNs was produced in Unet. Use --disable-nan-check commandline argument to disable this check.

[ornate elk]

Ah okay so you uses the correct guide

Make sure the model you used is a model and not a lora.
Check the file size of it.
It needs to be above 1.98GB to be a model

[regal ridge]

I see. I made both of the mistakes. 😅 are there any model file recommendations? For instance, I go to Civitai to download the models, I click on models, & I see a bunch of Base Models: SD 1.5 and others. Then I check the models & I see Types (Wildcard, Lora, Workflows, etc.). I would like to know which ones to select/avoid or if there's a guide here I could check out.

[ornate elk]

Checkpoints are models

[left juniper]

hi, i'm trying to troubleshoot why I'm generating distorted images when trying to replicate something from civitai, I import the png to comfyui and link all the loras and checkpoint, seed, and sampler, but in the end I get this jumbled mess of colors
Sometimes it works great, sometimes it's just a distorted bunch of colors, is it because the versions I download are different than the ones used in the example pictures from civitai?
It seems to me it happens a lot more with dpm++ 2m karras than with eula normal

[regal ridge]

thanks for the advice! i'll be sure to check them out

[noble belfry]

claymation looks good!

[ornate elk]

for your gpu, you should use only 1.5 based models. These are 2gb

[reef patio]

Where do I put --xformers --medvram-sdxl --no-half-vae

I do this and it gives me no module 'xformers'

and it's super slow

[ornate elk]

Make sure you launch the webui-user.bat and not the webui.bat

[reef patio]

ah thanks. Also, I cant seem to find prompt settings like guidance

sorry, I am a little rusty haven't done this in a while

[ornate elk]

You can find images with prompts and settings on Civitai.com

[reef patio]

as in, I can't find the controls for prompts

I am missing settings from last time

[ornate elk]

which settings?

[reef patio]

Like whatever Guidance is

[ornate elk]

CFG scale

[reef patio]

like how close the prompts have to be followed

ah

thanks for the help

[ornate elk]

no problem 🙂

[storm hearth]

How do I speed up my txt2img generations? A singular image already takes over 5 minutes which seems unreasonably slow

[reef patio]

oh yeah, is there a guide somewhere to speeding up my prompts

They seem much slower than before to even generate one image, same pc

I have an Intel7 if that helps

[ornate elk]

whats your GPU?

[ornate elk]

whats your GPU? and whats insie your webui-user.bat?

[storm hearth]

Pretty sure I have a 3070 ti, and only added xformers to webui-user.bat

[plain stream]

I am downloading Kohya SS, and I go to the point where I had to dowload homebrew, which I did, but now im stuck. What do I do?

[ornate elk]

a 3070 ti needs --xformers --medvram-sdxl --no-half-vae for the best speed

[ornate elk]

make sure you always launch the webui-user.bat
then also use 30 steps and i fyour using hires fix, set hires steps to 10

[storm hearth]

I'll give that a try, hopefully that improves my generation time from over 5 minutes to something lower

[reef patio]

i'm running webui-user.bat, using 20 steps and i'm not using the hires fix. I have these settings. Is one screwing me over

[ornate elk]

if your using sdxl models rn, it should be faster then

[ornate elk]

no these are needed for GPUs with 8gb vram

what are your txt2img settings?

make sure to dont enable Refiner

[reef patio]

haven't touched it from default

[storm hearth]

Not sure if the checkpoint I'm using is based on sdxl

[ornate elk]

looks normal

[reef patio]

Yeah weird. Not sure why its so slow now

I can run games at super high specs perfectly fine and get like 144 fps pretty consistently

[ornate elk]

you can check the models file size. if its below 6gb its mostly an 1.5 model. If its around 6GB then its probably SDXL based

[ornate elk]

Games and Ai Stuff dont use the same techniques

Also make sure you dont run Heavy programms in the background like Games or Wallpaper Engine. That will slow down the generation speed

[reef patio]

Yeah, I was just saying that my PC runs stuff well and I haven't had issues in the past

Should I be looking out for CPU or memory

[ornate elk]

@reef patio Can you make a speed test?
Try an 512x512 image with 20 steps,
positive prompt: cute cat,
negative: blurry, deformed
Select an 1.5 based model that is 2gb in size like Dreamshaper v8

then check the it/s you get in the cmd

[viral knot]

wait what... a1111 has automatic scheduler? how does that even work logically? like how does it select a scheduler automatically? lol

[reef patio]

Can't find anything that tells how long it takes unless I am blind

[ornate elk]

these are iterations per second

i would say they are pretty normal for that gpu

[reef patio]

hmm. My CPU and memory dont cap out with this

they only get to 80-90%

is it worth doing a factory reset on my PC or something

[ornate elk]

nope (only if you installed windows like 5 years ago) but you can do a quick system check and repair by
opening up a cmd as admin and run
SFC /scannow

Also make sure you have the latest nvidia driver installed

[reef patio]

Yeah. I do often get prompts telling me I don't have the latest driver installed

but I can't really seem to update it

[viral knot]

did nvidia fix the driver issues? i remember a while ago, you had to have like a specific driver version if you wanted a1111 to generate something good

[ornate elk]

yea, thats already fixed

[viral knot]

ah kk

[ornate elk]

why?

[reef patio]

i'll give it another go

But it was more like it said

[ornate elk]

make the system check before and a PC restart

[reef patio]

It was still out of date even after updating

[ornate elk]

okay, it found stuff and repaired it, now make a PC restart

[reef patio]

Yeah honestly, my PC is doing some weird stuff. There's a few too many thing srunning in the background. I will do a wipe tommorrow

See if that fixes it

[storm hearth]

So the model I'm using is indeed SDXL based and with the changes to arguments my generations now don't seem to take longer than 2 minutes compared to well over 5 minutes. Thanks a lot!

[ornate elk]

ah perfect, no problem 🙂

[lean ocean]

uninstalling venv folder fixed it i think xd

[viral knot]

kek

[lean ocean]

its weird bcuz i have trained a ton and it gave out randomly lol

[formal trout]

yes it is like that

[low basin]

Finally got a proper pc and am curious about what commandline args I should be using for optimization. Gpu is 4070 Super (12gb)

[ornate elk]

--xformers --no-half-vae should be enough

[lean ocean]

@ornate elk 17:52:03-319978 INFO Saving training config to
C:\Users\harla\Desktop\Kohya-Demo\model\vileaks_20240529-175203.json...
17:52:03-323987 INFO Executing command: C:\Users\harla\Desktop\Kohya-Demo\kohya_ss\venv\Scripts\accelerate.EXE
launch --dynamo_backend no --dynamo_mode default --mixed_precision bf16 --num_processes 1
--num_machines 1 --num_cpu_threads_per_process 2
C:/Users/harla/Desktop/Kohya-Demo/kohya_ss/sd-scripts/sdxl_train.py --config_file
C:\Users\harla\Desktop\Kohya-Demo\model/config_dreambooth-20240529-175203.toml
17:52:03-371835 INFO Command executed.
No Python at 'C:\Program Files\Python310\python.exe'
17:52:03-614674 INFO Training has ended.

Any idea of the fix?

No Python at 'C:\Program Files\Python310\python.exe' I have it which is weird

@vocal burrow

[calm cairn]

so here's a question about colors
I'm trying to inpaint, but the colors that come out do not match the colors in the original image. Is that a VAE problem? CFG is at 7, denoising 0.5-0.7

[pearl forge]

Hi @ornate elk

Is it possible to skin this model image so that we can mkae it realistic person?

[viral knot]

yes

[livid lark]

does anyone know where to put the yolo face hand and gashion detection models in comfyu?

[hasty orbit]

What is the best way to remove metadata from images in bulk?

[formal trout]

Exiftool

[hasty orbit]

ehhh didn't work, can still see the prompt in PNG info

[formal trout]

It should create two images, one original with different name and the new one with no metadata information,.same name as original

[hasty orbit]

Does dragging and dropping the image into the .exe still count?

[formal trout]

I'm using a Linux version that works with a command line, I don't know which is the Windows one

[hasty orbit]

What's the command? I can just run it in gitbash or cmd

[formal trout]

You do something like exiftool all = *./ I'm not at home right now

[hasty orbit]

That worked!

Thanks!

[formal trout]

🙌

[warm otter]

edit: it's not. somehow - sorry - I just checked against official python notebook - somehow my rest client has changed overnight, doing something different.
stability.ai api is broken, and not doing what it should.

See control/structure endpoint - it's ignoring the original image. This is on production. This is live. This worked the days prior, now the original image is ignored.

[warm otter]

Can anyone explain to me, why this request is not doing what it should? It's ignoring the original image

INFORMATION: --> POST https://api.stability.ai/v2beta/stable-image/control/structure h2

INFORMATION: Authorization: <omitted>

INFORMATION: accept: application/json

INFORMATION: User-Agent: OpenAPI-Generator/v1/java

INFORMATION: Content-Type: multipart/form-data; boundary=36eed318-391d-41d3-a465-169e36ae6407

INFORMATION: Content-Length: 556151

INFORMATION: Host: api.stability.ai

INFORMATION: Connection: Keep-Alive

INFORMATION: Accept-Encoding: gzip

INFORMATION: --36eed318-391d-41d3-a465-169e36ae6407
Content-Disposition: form-data; name="image"; filename="image17408095375412641010.png"
Content-Type: image/png

�PNG
<some decoded bytes>
--36eed318-391d-41d3-a465-169e36ae6407
Content-Disposition: form-data; name="control_strength"
Content-Type: application/json; charset=utf-8

0.7
--36eed318-391d-41d3-a465-169e36ae6407
Content-Disposition: form-data; name="seed"
Content-Type: application/json; charset=utf-8

0
--36eed318-391d-41d3-a465-169e36ae6407
Content-Disposition: form-data; name="output_format"
Content-Type: text/plain; charset=utf-8

png
--36eed318-391d-41d3-a465-169e36ae6407
Content-Disposition: form-data; name="negative_prompt"
Content-Type: text/plain; charset=utf-8

--36eed318-391d-41d3-a465-169e36ae6407
Content-Disposition: form-data; name="prompt"
Content-Type: text/plain; charset=utf-8

watercolor squirrel
--36eed318-391d-41d3-a465-169e36ae6407--

INFORMATION: --> END POST (556151-byte body)

source image and target image. Changing the control_strength parameter does not do anything.
When doing the same request from python, the result is as expected

Okay, found the issue. The official python notebook from openai uses as example for structure2image => sketch instead of structure - and structure is indeed broken right now!

here is the notebook:
https://colab.research.google.com/github/stability-ai/stability-sdk/blob/main/nbs/Stable_Image_API_Public.ipynb#scrollTo=59RaZazXz0AU
here is the relevant code line:

#@title Structure

#@markdown - Drag and drop image to file folder on left
#@markdown - Right click it and choose Copy path
#@markdown - Paste that path into image field below
#@markdown <br><br>

#some ommited stuff


host = f"https://api.stability.ai/v2beta/stable-image/control/sketch"
#some more

When changing the url to /control/structure, its not working at all anymore.

[vestal anchor]

My GPU: GeForce RTX 3060 Twin Edge 12GB GDDR6
I just installed ComfyUI and tried to generate something, but I got this error:

torch.cuda.OutOfMemoryError: Allocation on device

[wise scroll]

anyone else having issues with StableAudio timing out constantly when using input audio on generations?

😢

[ornate elk]

Yes, can be a VAE problem

[ornate elk]

What model and resolution? Did you used a custom workflow?

[wheat flare]

Hey all,

I'm trying to run SDXL through AUTOMATIC1112 on an Nvidia GTX 1660 Ti (6GB VRAM). Whenever I hit Generate, my computer fan whines for several minutes while nothing happens, and then the generation fails due an all-NaN tensor.

I've tried booting up AUTOMATIC1111 use --lowvram but it still doesn't work. SD1.5 does work fine without --lowvram, but of course the image quality is much lower.

Any thoughts?

[ornate elk]

hey, you also need to add: --xformers

also make sure you have 16gb RAM or more

[wheat flare]

I have 32GB RAM Will try --xformers in a sec

[ornate elk]

also your gpu needs mostly --no-half because GTX16xx cards cant use FP16 made models without it

that should prevent NaN errors

[wheat flare]

OK, so I tried running AUTOMATIC1111 with --lowvram --xformers --no-half. No dice

[urban oyster]

Hey guys, I'm testing the sd3 api https://api.stability.ai/v2beta/stable-image/generate/sd3 image-to-image request and i'm getting very slow request times (~40s for sd3, ~2m for sd3-turbo) is that normal?

[ornate elk]

what does the cmd shows?

[ornate elk]

can you share the full cmd log?

[wheat flare]

Copied and pasted straight out of Anaconda. Couldn't fit it in a Discord message, so here's a text file:

[ornate elk]

you need to double click the webui-user.bat to start it with the launch args

and edit it before

it wont work like that: webui-user.bat --lowvram --xformers --no-half

[wheat flare]

Once I double-click, where do I put the args?

[ornate elk]

right click to edit, then at the line Commandline_ARGS=

[wheat flare]

Is this correct?

[ornate elk]

yes

[pallid crag]

gm guys, may I ask why I can't find lora training tab in Automatic1111 even I've installed dreambooth extension?

[cursive cobalt]

@pallid crag Should be there.

@ornate elk 👋 How does one find, and use the pose function of... ADetailer was it?

[ornate elk]

pose?

[cursive cobalt]

Wasn't there some dynamic poses thing that made u able to fine tune poses?

[wheat flare]

It's working! Thanks for the help

[pallid crag]

oh I thought it is for Checkpoint? isn't it?

[cursive cobalt]

[ornate elk]

you mean controlnets openpose editor?

[cursive cobalt]

Yes. Thats the one

[ornate elk]

yea thats available in controlnet

[cursive cobalt]

Got a link to a guide that tells in an ELI5 format, how it works?

[ornate elk]

do you have the controlnet extension already insatlled + the controlnet models?

[cursive cobalt]

I think so?

[ornate elk]

then do this, and use the second edit button, not the one with P

[ornate elk]

seems like you dont have the controlnet models

Here are they: https://civitai.com/models/38784/controlnet-11-models

[cursive cobalt]

That seems correct

D:\A1111\stable-diffusion-webui\extensions\sd-webui-controlnet\models, ye?

After an update, they showed up

Refresh***

[ornate elk]

Animerge isnt a controlnet model

you need the ones i linked you

[cursive cobalt]

Remnant of incorrectly installed.

I did install those. ControlNet 1.1 Models, is the one

[ornate elk]

which one ?

[cursive cobalt]

The Controlnet(space)1.1(space)Models.

ControlNet 1.1 Models, not the Controlnet11models_tileE

But presing the edit button opens this window

[ornate elk]

what webui are you using?

looks like SD Next

[cursive cobalt]

Auto1111

The colors are nothing but a theme

[ornate elk]

thats how it should look like

you need the controlnet openpose model

[cursive cobalt]

Yarsh?

The one you linked was the Tile (e) version

(https://civitai.com/models/38784?modelVersionId=44811)

[ornate elk]

i just linked the site

where you have to select the right model

[cursive cobalt]

Figured, ye. I picked the OpenPose one

I moved the irrelevant stuff, THis is how it looks now

[ornate elk]

okay good

[cursive cobalt]

OOoohhh... I c. I clicked the wrong Edit button. This is how it looks now

[ornate elk]

yep perfect

[cursive cobalt]

Once edited, i press this, ye?

[ornate elk]

Yep

[cursive cobalt]

"RuntimeError: mat1 and mat2 shapes cannot be multiplied (154x2048 and 768x320)"

[ornate elk]

Are you using sdxl or 1.5?

[cursive cobalt]

A pony SDXL merge

[ornate elk]

That won't work with the 1.5 Controlnet

You need an sdxl Controlnet openpose model

You can find them here:
https://huggingface.co/lllyasviel/sd_control_collection/tree/main

[cursive cobalt]

https://civitai.com/models/136070?modelVersionId=155270 ?I do mainly dabble in the 2, 2.5D space

[ornate elk]

Then use that openpose model

Should work too

[cursive cobalt]

https://tenor.com/view/pleasestandby-lol-gif-24264557

(Simply edit for testing)

[inland oracle]

HI, I was using the webui and I tried to generate the second pic I face this error:

OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 7.75 GiB of which 7.06 MiB is free. Including non-PyTorch memory, this process has 7.73 GiB memory in use. Of the allocated memory 7.21 GiB is allocated by PyTorch, and 376.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

I am using linux and rtx 4060 laptop

[cursive cobalt]

Didn't quite seem to work, unless i obviously did something wrong I pressed the "Generate" button

It does generate the stickman, but not doesn't seem to "force" the character into that pose

[worthy nymph]

Anyone knows if its possible to put pictures upon Loras or embedding for when selecting ? Having a hard remembering all they do exactly when I have to choose them with their weird names....

[cursive cobalt]

Is there an option for ComfyUI to do this (pic), but with promts, like Auto1111 can?

[regal ridge]

i just got this error today while opening SD: "Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check". It was working last night but not sure why now it isn't. any help would be appreciated 👍

[formal trout]

once you have the line dummy, you don't need the reference image and you don't need to run the pre-processor, set it to none

[waxen pebble]

I have been having some great success training a comic style dreambooth... except alot of the images are coming out cropped:

im guessiung this is due to bucketing, i havent paid too much attention to it, and have quite a few bucket sizes in my training dataset.

anyone got ideas to what might be going wrong?

*i have tried this with wider aspect ratios. some things it just seems to want to cut off:

[safe basin]

gets stuck on this

[ornate elk]

hey, let it load, its not stuck

[safe basin]

I've let it load for 10 minutes with no progress.

I got past this on my first try, but I had some other kind of error that I tried fixing

[ornate elk]

okay, let it load it can really take a bit

wlaos whats your gpu?

and how much RAM

[safe basin]

rx 6950 xt and 32gb

but i got past this before but i had this error instead
RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check.

so I followed some tutorial to fix that and somewhere along the way I prob messed something up

[ornate elk]

then you have installed the wrong webui it seems

AMD need a different guide

[safe basin]

yeah I followed the AMD guide, python and git. then I cloned it to a folder on my local disk. ran webui-user.bat and then It always just gets stuck on: Installing collected packages

[ornate elk]

did you followed my AMD Zluda guide?

or the directml one?

[safe basin]

https://github.com/lshqqytiger/stable-diffusion-webui-amdgpu is what I used

[ornate elk]

Thats the right source but you need to follow an exact guide

I made an Zluda guide. Its in the pinned messages of this channel.
If you follow that you will get it to work

Direct link to the guide:
#🤝｜tech-support message

[safe basin]

Im on it

[ornate elk]

I'm off now as its pretty late here (4am)
I can help you tomorrow

[safe basin]

Yeah still doesn't work, stuck at this now

python3 -m pip install git+https://github.com/mlfoundations/open_clip.git@bb6e834e9c70d9c27d0dc3ecedeebeaeb1ffad6b --prefer-binary

got me this

but when I reopen cmd its still on open_clip

[subtle quarry]

does anyone know what this means?

[ornate elk]

Open up a cmd and type
Pip cache purge
Then delete the venv folder again and relaunch the webui-user.bat

And then let it load until you get an error.
Also make sure to have 10-15gb free space on C drive

[safe basin]

Well I fixed it by running webui-user.bat with admin through cmd and disabling my integrated graphic drivers. But If I try to start it without admin I get an error.

But my new problem is that it crashes: There is not enough GPU video memory available. It uses all of my 16gb vram when I try to generate hi res images

Zluda made it so that I couldn't even render an image but I got it to work with directml.

[cursive cobalt]

What is this now 😢

[ornate elk]

Yea directml isnt recommended to use please use zluda and don't start the file as admin

Better show me the error.
Show me your environment path settings too

[cursive cobalt]

Mornin @ornate elk How i fix the error code above?

[warm otter]

Stability api still broken, no way to contact support. Using them in production app. I will consider blogging about this failure.

How can a multimillion dollar company fail 1/10th of their product and not realizing?

I lost real money, because of that.

See reddit in the next hour for a write-up

[ornate elk]

Do you have git installed?

[cursive cobalt]

Git... As in the client? No

Didn't need the client to install these 2, so didn't think I'd need it at all @ornate elk

[ornate elk]

If you try to git clone something, like in the top screenshot, then you need the Programm Git

Install it and try again

It should work

[cursive cobalt]

Even when using the "INstall via GitURL" in the manager?

[ornate elk]

Yep

[cursive cobalt]

Right. With Github installed and logged in, i just try again?

Nop @ornate elk Didn't seem to fix it

Hol

I c 1 issue

Sec. Lemme retry

Ok. If i weren't such a dummy, I should've realized... I already have the impact one installed xD

https://tenor.com/view/ralphwiggum-smart-simpsons-gif-5270199

[cursive cobalt]

AHA i'm BACK! What r this?

[vocal burrow]

https://tenor.com/view/futurama-math-mathematics-we-need-math-we-need-to-use-math-gif-3486402

[cursive cobalt]

I barely passed that stuff. Maff is hart

[vocal burrow]

can t tell with a screenshot so croped

[cursive cobalt]

Might have made a slight mistake importing a, what seems like a fairly complex workflow, cause damn son, this be producing errors

[vocal burrow]

Looks pretty self explanatory, you connect whatever numerical input you want to a,b and c. Type whatever math expression you want to use them with and it will spit out the result on the other hand.

[cursive cobalt]

Is the errors in question

Then it throws a purple outline around the same steps note

[vocal burrow]

yeah cause the expression is empty

so it can t output anything

[cursive cobalt]

I am a Ralph, when it comes to anything python related xD Slow learner

So, this is the group its near. Not quite understanding what the first box needs

[vocal burrow]

no idea what it wants, that depends on the whole workflow :p

try passing just a for now

[cursive cobalt]

https://comfyworkflows.com/workflows/f5a43fc5-625b-4fa7-9dfe-aeb19b3051de Workflow link xD

[pseudo prairie]

I am very new to this Ai thing

i am not sure what to do

[vocal burrow]

it s not empty

you probably erased it

[cursive cobalt]

Windows key + Shift + S, Feather. Select area

[cursive cobalt]

Nah fren. A lot of stuff were missing some connections

Including these 2, but I assume they're just not needed xD

[vocal burrow]

it s telling you what to do at the very top of that screenshot of yours.
Install a compatible python (anything from 3.10.6 to 3.10.13), delete your stable-diffusion-webui/venv folder, launch again, read logs if there is any errors

[vocal burrow]

those are not connected on the workflow you ve sent

but that a / 2 was there

[cursive cobalt]

So is safe to simply ignore the error its outputting? (I assume the red outline = invalid input/output).

[cursive cobalt]

I c. I might deffo return later, once another unknown error, my smoof brian can't quite understand, pops up 😄

[vocal burrow]

try running it and see what happens

it s not gonna self explode :p

or is it...

[cursive cobalt]

Based on your novelty status... I'm afraid

[pseudo prairie]

ok im an idiot i dont know where to find webui/venv folder

[vocal burrow]

should be right next to your webui-user.bat

unless you downloaded the .zip from github (I d advise against that method)

[pseudo prairie]

i followed this tutorial

[vocal burrow]

then it should be in one of the folder such as webui or something like that, I don t have access to it atm

[pseudo prairie]

and it did have the github thingy

[vocal burrow]

"this" ? which one ?

[pseudo prairie]

https://youtu.be/Po-ykkCLE6M?si=Gum5-rEB-5EMeP7X

this guy

[cursive cobalt]

It seems once it reaches this area... (Expressions) it just... Stops. It doesn't seem to output a result to these

[pseudo prairie]

now i don’t even have python

im so dumb sorry

[cursive cobalt]

Friend plz. Take screenshots not phone 😄 Makes it easier to read (Shift + Windows key + S, on Windows)

[vocal burrow]

wow he recommends downloading sd 1.4... why... it s so outdated...

[pseudo prairie]

im lost

im dumber than a rock l

[vocal burrow]

you probably didn t check the "add to path" checkbox during python install

[pseudo prairie]

what should i do 🥲

[pseudo prairie]

i did actually

[vocal burrow]

it s still trying to find your previous python 3.12

that s because you did nt delete the venv folder

[pseudo prairie]

how do i delete that silly folder

[vocal burrow]

delete this

[pseudo prairie]

this what?

[vocal burrow]

right click the folder, shift + click on delete

[pseudo prairie]

i can’t find that folder

[vocal burrow]

it should be at the adress I just croped from your screenshot