🤝｜tech-support

[ornate elk]

np, if your in the webui in browser, scroll to the bottom

[rich lynx]

i click on the webui in the folder but after that i just get that error. i cant get it to load in browser anymore

[ornate elk]

oh okay, when did you installed SD?

[rich lynx]

about 6-8 months ago

it wasnt until the last month that i really got back into it

[ornate elk]

okay do you have git pull in your webui-user.bat?

[rich lynx]

not sure what that means. i have ran that command in the cmd terminal or whatever its called though, and it says its up to date

not sure if im even making sense. using the cmd terminal and other command prompts is new to me btw

i get this if i run git pull. C:\Users\kironess\Downloads\SD\stable-diffusion-webui>git pull
remote: Enumerating objects: 105, done.
remote: Counting objects: 100% (65/65), done.
remote: Compressing objects: 100% (20/20), done.
remote: Total 105 (delta 50), reused 56 (delta 45), pack-reused 40
Receiving objects: 100% (105/105), 71.88 KiB | 1.94 MiB/s, done.
Resolving deltas: 100% (65/65), completed with 22 local objects.
From https://github.com/AUTOMATIC1111/stable-diffusion-webui

• [new branch] add-option-Live-preview-in-full-page-image-viewer -> origin/add-option-Live-preview-in-full-page-image-viewer
  334298d4..f92d6149 dev -> origin/dev
  b8040e4a..57428361 gradio4 -> origin/gradio4
  5ef669de..120a84bd release_candidate -> origin/release_candidate
  9a15ae2a..39ebd568 test-fp8 -> origin/test-fp8
• [new tag] v1.6.1 -> v1.6.1
• [new tag] v1.7.0-RC -> v1.7.0-RC
  Already up to date.

C:\Users\kironess\Downloads\SD\stable-diffusion-webui>

[ornate elk]

okay good now run
pip cache purge
then delete the venv folder

[rich lynx]

allright it looks like it deleted some things. im now removing the venv folder. i can try and reload after now?

and thank you btw. this problem has been driving me crazy.

[ornate elk]

yep

no problem, if it doesnt work youll need a quick reinstall of SD but i can guide you

[rich lynx]

this is strange , i have to webui files btw i dunno if that could be a prob

one shows this past month the other august

[ornate elk]

always run the webui-user.bat

[rich lynx]

ok got it

i got the same problem :(. RuntimeError:
CUDA Setup failed despite GPU being available. Please run the following command to get more information:

    python -m bitsandbytes

    Inspect the output of the command and see if you can locate CUDA libraries. You might need to add them
    to your LD_LIBRARY_PATH.

frustrating, one minute its working and im getting into it, really liking doing my gif art and now busted :(. I guess if ur free and can guide me to doing a fresh install i would appreciate it very much. otherwise ill youtube it later or something. i dont like to bother ppl too much.

[fading fable]

what do i do when it says not enough gpu video memory available?

[crimson leaf]

why does it try to install 1.6.1 even tho i used the 1.2.1 one?

alr i found out that it created in that folder another instance of sd

[twin pawn]

@jagged plover Hey there. I was asking about multi GPU on SwarmUI on Reddit. Would love to pick your brain when you get a sec. Thanks!

[jagged plover]

ohhai

[twin pawn]

Hallooo

So yeah. I just think i have everything set up correctly and my times aren't improving and now task manager isn't even showing activity like before.

Here are my debug logs

https://pastebin.com/PDE7zZCp

And I'm just running the first example on this page with a batch size of 9

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

[ornate elk]

whats your gpu?

[leaden night]

Hello! I am having issues installing stable-diffusion webui
more specifically, i cant run the webui-user batch file. Error code 128

[plucky marsh]

Good evening guys, I have a problem with stable diffusion and i dont know how to solve it. I downloaded fooocus and im running it locally on my laptop just fine, but whenever i go to the online version on https://colab.research.google.com/github/lllyasviel/Fooocus/blob/main/fooocus_colab.ipynb i can't run it cause it says ''runtimeerror: found no nvidia driver on your system''. I tried reinstalling nvidia drivers from scratch 3 times on my laptop so that didn't help 😦 IDK what I can do, everything was running fine yesterday, I remember running fooocus on my laptop locally and then i ran the one online simultaneously, im not sure if i broke it somehow in that way?

[ornate elk]

do you have a screenshot of the whole cmd log?

[leaden night]

Sure! gimme a min

[plucky marsh]

if i may phrase my question differently: can i add a command line to detect my gpu in slot 0 here? !pip install pygit2==1.12.2
%cd /content
!git clone https://github.com/lllyasviel/Fooocus.git
%cd /content/Fooocus
!python entry_with_update.py --share

[jagged plover]

that log looks like swarm was freshly started and you just did the first generation

the speedup from multigpu only comes into play after it's all booted and every backend has loaded the model you're using and then you queue up several images to generate

[kind hound]

i have installed this version "lshqqytiger/stable-diffusion-webui-directml" of stable diffusion with olive. now my question is can i install the d8ahazard /sd_dreambooth_extension on this version too? because i see only people use that with automatic1111. Some infos i use a amd gpu and the first version dont cause a memory leak.

[elfin dirge]

can anyone help me my extensions are not showing up even tho they are installedI have webuireactor downloaded and adetailer but only adetalier shows up

[proper pike]

Does anyone know if these settings are ok for training a custom model for sd 1.5?

[plucky marsh]

does anyone know why am i suddenly getting an error ''RuntimeError: Found no NVIDIA driver on your system.'' when running fooocus_colab in a browser?

[placid edge]

I am running Automatic 1111 and my prompts seem to get "stuck" at 98% but in the terminal it says it's going to 100%. Also the final processed image is not in my output folder. I did notice that this started happening since I downloaded controlnet, is this a known issue with controlnet and automatic 1111?

[ornate elk]

Make sure your webui is whitelisted in any browser adblocker

[ornate elk]

No you can't use it with olive or directml

[wet hull]

I have a question: Am I able to combine multiple LoRAs?

For an example... Let's say I have a Spongebob LoRA and a Patrick LoRA. am I able to use both to have an image of both Spongebob and Patrick?

[ornate elk]

Yes you can use both together but it would mix them up.
So you need an extension for prompt control like Tiled Diffusion or Regional Prompt Control

[wet hull]

I see, so without extensions it would be like if Spongebob and Patrick would be mixed into a blender.

Time to search the internet

Actually, while I'm here, are there any extension you reccommend me getting?

(and how would I install them to stable diffusion?)

[ornate elk]

Yes, the boorutag autocomplete extension is very handy

and for the two Lora get the Tiled Diffusion Extension

In the webui go on the extensions Tab. There click on available. Then on "Load from"
Now you have a big list. And a search bar

After installing go on the Installed Tab and click restart webui

[daring jasper]

hi! ) I'm still noob in coding, so may I ask for help with this) How can I fix it?)) How can install ffmpeg and what's wrong with a scripts.sam module? 2023-12-08 11:18:39,891 - ControlNetFastload - INFO - Load Controlnet Fastload Filter on isRemote=None and accessLevel=2
[FACEFUSION.CORE] FFMpeg is not installed
*** Error executing callback ui_tabs_callback for E:\Stable Diffusion and AI\stable-diffusion-webui\extensions\sd-webui-replacer\scripts\replacer_main_ui.py
Traceback (most recent call last):
File "E:\Stable Diffusion and AI\stable-diffusion-webui\modules\script_callbacks.py", line 166, in ui_tabs_callback
res += c.callback() or []
File "E:\Stable Diffusion and AI\stable-diffusion-webui\extensions\sd-webui-replacer\scripts\replacer_main_ui.py", line 141, in on_ui_tabs
from scripts.sam import sam_model_list, refresh_sam_models
ModuleNotFoundError: No module named 'scripts.sam'

full log

[mental locust]

Really sorry to ask such a basic question... is there a good tutorial for installing stable diffusion locally?
The youtube tutorial by matt wolfe worked for me back in May but then stopped being useful due to changes in python, or something. The FAQ and resources in here don't have install tutorials...

[karmic crown]

GPU and OS?

[mental locust]

4090, Win11

[karmic crown]

#🤝｜tech-support message

how much VRAM? 24 or less?

[mental locust]

24GB

[karmic crown]

than just add --xformers to command line parameters

@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--xformers

call webui.bat

[mental locust]

That wasn't the question sorry. I want to install Stable Diffusion locally. It was easy and flawless when I did it in May, and used through until August without issue. After a break I tried reinstalling recently and had some kind of issue that I didn't understand. I might be being an idiot, but I was hoping for either a text or video tutorial I could follow along to ensure I install A1111 and everything correctly

Thanks for helping

[karmic crown]

i sent you a link #🤝｜tech-support message

just click on it, and you will get a windows guide for nvidia

[mental locust]

Epic, I am immortalised as the newbie. Thanks I shall do this now 🙂

[karmic crown]

if your python and git are already correctly set up (as in that guide), skip that

if they are working, just remove venv folder inside your stable diffusion

do git pull (that will do the update)

venv will be recreated on run, and that is what you want, to get all the latest versions of pyhton components

[austere needle]

I would like to know any post about making need check point for Stable Diffusion ? ，I have study about how to make new LORA to support create need image. but Still dont know how to make need check point file .safetensors

[karmic crown]

you can convert .ckpt to .safetensors

https://github.com/diStyApps/Safe-and-Stable-Ckpt2Safetensors-Conversion-Tool-GUI

if that is your only problem

lora trining tutorial is here

or this

[austere needle]

i train up a lora model and change it to check point ?? is this a seperate program of SD?

[karmic crown]

if you already has a ckpt, just use this tool to convert to safetensors https://github.com/diStyApps/Safe-and-Stable-Ckpt2Safetensors-Conversion-Tool-GUI

i am not sure if you can save it as safetensors directly, since i have never trained a lora myself

[austere needle]

I will like to do more detai of Leg / shoes Area. because much of them a focus on the top half body and face

[austere needle]

.safetensors in Lora and . safetensors apply for main check point is that the same? this is the point which i dont know.

[karmic crown]

.safetensors is just a format

as far as i know, you use a base model and you train your lora using that model, when you finish, lora goes to lora folder, models are always in stabele-diffusion folder

than you use your lora as any other lora from civitai or liblib

and you can use it with different models, not just the one you used when trained lora

[random pagoda]

after generating images i get this error

RuntimeError: min(): Expected reduction dim to be specified for input.numel() == 0. Specify the reduction dim with the 'dim' argument.

any advice?

[austere needle]

is that your output setting size too large ?

dimension

[mellow finch]

hi, I'm new here! I'm facing difficulty installing SD on my mac. I got stalled at the first step of installing homebrew. Is this the right community to come to for help?

[karmic crown]

hello, i can help with mac

you can follow my guide from https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/5461#discussioncomment-7738578

let me know if you have some problems

[spice nova]

Hi, I saw in the research paper about Stable Video Diffusion that motion loras can be applied. Are there any official weights for motions loras?

[spice nova]

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" this is the command you in to put in your terminal. If you get errors attach screenshot

[ruby lotus]

Hi anyone know anything about LORA training? I have some newbie question about how to train for EYES only

I have trained with 100 full faces with specific eyes, but when using it, it only seem to use the faces it has been trained with. I want to apply my lora to any type of person. Should i only include crops of the eyes in the training data or what is the best practice for training on specific subjects

[mellow finch]

Hi, thanks for your reply. I tried your command and still got failure:

[karmic crown]

@mellow finch what is your mac? intel?

and, have you ever installed brew before?

if that is silicon, do not use rosetta

[mellow finch]

yea intel mac 2020

[mellow finch]

running mac os sonoma 14.1

i tried installing brew today a couple of times

i also tried the .pkg installer but somewhere alongtheway i got errors

[rotund lava]

sorry but you cant generate anything on a UHD 630

that thing has the same performance as a GTX 260

[mellow finch]

oh no ok

[rotund lava]

but since you have an i9 you could probably run it on cpu through openvino

[karmic crown]

@mellow finch uninstall it first /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/uninstall.sh)"

i guess you had some messed up install

probably half-installed

[mellow finch]

exactly

let me try

[karmic crown]

when you finish install it again with /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

[mellow finch]

is this ok? can i try reinstalling now?

[karmic crown]

wait

remove those folders

this one /usr/local/Homebrew

remove it from finder

and those

all of them

Why do I have a feeling that you had run brew install with sudo at some point in time?

[mellow finch]

what's sudo?

[karmic crown]

command that should not be run unless is absolutelly necesary

have you removed those folders?

[mellow finch]

not yet

im trying to find these folders

[karmic crown]

ok, try like this than

rm -rf /usr/local/Caskroom

can it remove it or not? if there is a an error we will try "stronger" command

[sand trail]

why is that this controlnet model isn't working with my checkpoint

[mellow finch]

that command does nothing

[karmic crown]

if there was no error that it is fine

[karmic crown]

ok

repet with other folders like this

rm -rf /usr/local/Cellar

rm -rf /Users/aseem/Libary/Caches/Homebrew

rm -rf /usr/local/Homebrew

mentioned folders are now removed

now try to install it again

when you finish install it again with /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

[karmic crown]

ok

[mellow finch]

should i continue with this now?

[karmic crown]

wait

sudo rm -rf /usr/local/Homebrew it will ask for password

you will not see any response (like when you installing brew)

but, you can also try to find folder in finder

and to remove it from there

[mellow finch]

i have done this

[karmic crown]

try again

it should report nothing

if there is a permission denied you still have a problematic folder (with wrong user privilges)

[mellow finch]

exactly

reports nothing

[karmic crown]

great

install now

it hopefully should work

[sand trail]

error message

[mellow finch]

i think its succesful this time right?

[karmic crown]

yes, run those

when it finish, close terminal and open it again

and type brew doctor

it is a 1 second process

brew doctor should give you this

❯ brew doctor
Your system is ready to brew.

after that, you can follow my guide from here https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/5461#discussioncomment-7738578

if you already have stable-diffusion-webui folder, remove it first

[mellow finch]

i dont

ok let me try

thanks for your support

[karmic crown]

skip 1st step obviously

you just installed that

you will probably need to modify command line params, but that is not an issue at this moment

[ornate elk]

You can't use controlnet 1.5 models for sdxl

You need sdxl controlnet models

[mellow finch]

is this 64 bit intel cpu?

[karmic crown]

yes

what is happening?

[karmic crown]

give me a second

[mellow finch]

ok

[karmic crown]

actually wait

it is

close terminal, reopen it and type brew doctor

[karmic crown]

great

[mellow finch]

:))

[karmic crown]

beww is propery installed now

you can start with my guide

[mellow finch]

awesome

thanks again

https://tenor.com/view/thank-you-thanks-hey-bro-fist-bumb-gif-16262296206650437916

[karmic crown]

i will be here, let me know if it works

[mellow finch]

ill try your guide a bit later

[karmic crown]

your graphics card is potato, but... we will make it work 🙂

[mellow finch]

hahaha

[karmic crown]

you will probably need to use cpu only

[mellow finch]

i usually just play around with midjourney

but wanted to try sd

[karmic crown]

i have m1 and it slow

i have imac with radeon with 8gb that is even slower

[mellow finch]

right

i have that too

but figured macbooc would be a bit better

[karmic crown]

imac with radeon will work better

[mellow finch]

oh is it

[karmic crown]

if it is 2017 or 2019 5k with radeon with 8 GB

[mellow finch]

2019

[karmic crown]

you can try on both

[mellow finch]

but i think i have 16 gb on imac

i got extra ram later

[karmic crown]

vram

for graphics

[mellow finch]

ill check and get back to you

[karmic crown]

graphics part

[gray oyster]

not SD related but AI related:
Does anybody know what's causing this when using a trained U-Net?
I suspect it might overmatch certain weights so im just gonna add a mechanism that will punish the model for emphasizing the same weights over and over again and i will also try to complicate my synthetic dataset creator

though weirdly it looks like a pattern

[sand trail]

tried that but it got this

[gray oyster]

try adding set CUDA_VISIBLE_DEVICES=0

below your cmd line args

accidentally posted the wrong screenshot

now this is correct

hope that helps

by the way do you have a dual-gpu setup?

i had similar issues with language models when running a dual-gpu setup with two different gpus so they couldnt act like one unit

[crimson leaf]

does it make difference when using sd on ubuntu or windows?

[ornate elk]

If your on AMD then yes

[crimson leaf]

for nvidia?

no difference?

[ornate elk]

Not much

[crimson leaf]

is it a bit faster/better on support on ubuntu?

or depends?

[gray oyster]

i heard the linux drivers might be a bit shitty for Nvidia

[crimson leaf]

what way, incompatibility or optimization?

[gray oyster]

both kind of

i guess they are a little bit more outdated

since nvidia is mostly focusing on windows drivers

AMD is a little bit more active on linux

hence AMD cards run SD better on linux than nvidia

afaik

[crimson leaf]

hm, then i should also test sd on windows.. Thanks for the information

[gray oyster]

no problem

also yea on windows it should run better, atleast in theory

therefore of course you got higher ram usage unfortunately

would be cool if there is like a developer oriented windows version which is basically lightweight windows without any of the propriatary software like edge, cortana, microsoft store, auto updates, etc.

[crimson leaf]

That's not a problem, i was just annoyed by the gnome bugs on ubuntu.. Also the mass error..

gnome crashed sometimes when generating at full load

[gray oyster]

also if you want to free up a lot of ram on windows you want to right click the task bar and disable weather and news

[crimson leaf]

thanks for the information tho

[ornate elk]

That is available and is called Windows Amorilate

[gray oyster]

ooooo

[crimson leaf]

never heard of

interesting

[gray oyster]

same

cant find anything on google either for the short time i looked for it

[crimson leaf]

lmao me too

[ornate elk]

Sry, I didn't knowd the exact name but here it is:
https://ameliorated.io/

Its for ISO Customisation

[gray oyster]

ah alright, thanks

[crimson leaf]

thanks

[swift vigil]

hi, is there a way to auto launch sd on a different browser ie. not the default browser?

[unique urchin]

I'm trying to delete the stable diffusion folder so i can do a full fresh install, but its not letting me delete it, saying i require permission. But i have admin rights

[crimson leaf]

Is it true that on rocm you get slightly different results as on cuda?

i can't seem to generate the exact picture anymore

[ornate elk]

Restart the PC then delete

[ornate elk]

Yes

[crimson leaf]

and does --opt-sub-quad-attention --opt-split-attention-v1 decrease performance on nvidia

[ornate elk]

I wouldn't use it for nvidia

[rich haven]

Your account should be a member of the Everyone group. Try to use File Explorer (Hmm, I only know Windows stuff) to give yourself full rights to the directory, or take ownership of the directory.
https://youtu.be/QoVH8cMQ9Ns?si=SD2UcnLALS_l44y0

[crimson leaf]

i hate that issue, it is an extremely annoying issue on windows. Dam traumatized by this

Trying to delete microsoft trash

[rich haven]

How can I get support from stability.AI? I need a password reset, but it is not sending the password reset to my email address. The website had no problems subscribing this address to their newsletter. It seems to be just the password reset request that is not working.

[gray oyster]

honestly the performance cost atleast for me is negligable, literally no difference except that xformers uses a little less vram

[cedar depot]

Is there a way to fix the "Not enough memory" issue for AMD? I got 24gb VRAM but it's like it wont utilise it

[ornate elk]

yes, you need to edit the webui-user.bat
at the line COMMANDLINE_ARGS=
you add: --medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half --upcast-sampling
Then your able to use 512x768 and hires fix upscale by 2. (10 hires steps)
Then you need to get the Tiled Diffusion Extension and there activate the Tiled Vae option if you go for Full HD image like
540x960 upscaled by 2

[cunning badger]

hey everyone! tile_resample pre-processor seems to be missing for me in controlnet, any idea why this might be, or how to fix it?

i have control_v11f1e_sd15_tile.pth in the controlnet models folder

[cedar depot]

Ah okay so I take it I cant go up to 32:9 res ?

[cunning badger]

but no tile_resample comes up in the preprocessor dropdown

[ornate elk]

Yep not native, first use hires fix for better quality

But with img2img upscale script

[cedar depot]

Ah okay gotcha

Are blurred outputs also an issue with AMD?

[crimson leaf]

tf, that's weird, i generated with my rx580 8gb 2048x2048 images (--medvram)

[ornate elk]

Nope

[cedar depot]

Bruh tf be going on then

[cedar depot]

I tried --medvram and the outputs were super low quality

[crimson leaf]

Wrong inputs ig

at 2048x2048 you won't get something logical, for me it was just some random fading colors or sth

the models ain't trained for that

May you share prompt and settings?

[cedar depot]

I'm using dreamshaper 8, prompts are:

(masterpiece), (extremely intricate:1.3), (realistic), portrait of a girl, the most beautiful in the world, (medieval armor), metal reflections, upper body, outdoors, intense sunlight, far away castle, professional photograph of a stunning woman detailed, sharp focus, dramatic, award winning, cinematic lighting, octane render unreal engine, volumetrics dtx, (film grain, blurry background, blurry foreground, bokeh, depth of field, sunset, motion blur:1.3), chainmail

Negative: BadDream, (UnrealisticDream:1.3)

I am just testing out with an already made output from someone who used these

Output just comes out blurred no matter what

Lemme post the full size

Oh ffsd

[ornate elk]

Thats normal for an 512x512 when the face is this smal

[cedar depot]

You cant really see it, but the edges are blurry around the armour and such, easier seen when I enlarge the picture on SD

[ornate elk]

Try 512x768 and go for portrait tag

[ornate elk]

Dont use motion blur:1.3

That will blur

Use bokeh

Or depth of field

[cedar depot]

I mean its the same with or without tbh

[swift vigil]

what command args should i be using for nvidia 3080 10gb vram?

[ornate elk]

--xformers --no-half-vae and if an sdxl image takes longer than a minute add --medvram-sdxl

[swift vigil]

ty sir

[ornate elk]

Np

[wet hull]

I have a question: Does 'Draw full canvas background' affect the image generation by any means? I tested out Tiled Diffussion and I kept getting the different prompts in these weird picture frames, even though I want it all to be one connected image.

[ornate elk]

For better quality try hires fix

I have amd and no problems

[ornate elk]

You first need to select the resolution. Then press draw full canvas. Then start adding your squares

[wet hull]

Here is an example of what's going on. It's not supposed to be like this, it's supposed to be a normal jungle background.

[ornate elk]

What's your setting?

[wet hull]

Here, I'll use a better example:

-Without Draw Full Canvas Background (Image was four colors)

-With Draw Full Canvas Background (Image was fully white I think)

ik these are both awful generations

[ornate elk]

How does your regional prompt control looks ?

[wet hull]

It looks like this

There are 5 squares. The fifth one covers the entire image and is in the background.

[ornate elk]

Ans what's the resolution ?

[wet hull]

640x360

[ornate elk]

okay thats pretty smal

[wet hull]

I do small resolutions since i don't have a strong PC

Actually I think this extension optimizes that, right?

So, what's the issue?

[ornate elk]

Try only with two squares

[wet hull]

ight, I'll reduce the amount to two. I'll generate an image with these settings.

Here is the two square image. It looks a lot better, even with some weird lighting issues.

[ornate elk]

Also try lowering the lora strength

[jagged plover]

oh my god this is delayed but i went back and checked to see if you replied and just noticed this comment -- if you're using the Comfy tab that's not multigpu out of the box, especially batching, wherein the point of batching is doing several on one GPU. You'll want to use the Generate tab to do one workflow repeatedly across several GPUs. (You can do a multiended workflow on the Workflow Editor tab across several GPUs but that's more complex to set up)
ps should use #🐝｜swarm-ui so conversation doesn't get lost in this endless channel

[mossy folio]

What models and nodes do I need to run IP Adapter in comfy and where do I put the models

[wet hull]

Here is an image with Monika's LoRA strength at 0.75. If that's still too much then give me a good value to test out.

[ornate elk]

i think the resolution is still the problem too

[sharp lichen]

Is there a controlnet model that can pickup on color data?

or anyway for me to use controlnet and also get color data from the input image, other than img2img

[wet hull]

I can't do a bigger resolution, my 4 GB graphics card can't handle it.

[hardy canopy]

Is there a commandline arg for setting a custom path for ControlNet models? I've tried --controlnet-dir to no avail

[jagged plover]

Presuming you're on an SDv1 image, changing to exactly 512x512 might better but still fit your VRAM.
Also lowering CFG Scale might help (or using Dynamic Thresholding)

[jagged plover]

Depends what UI you're using - in Swarm it's right next to all the other model settings, in comfy there's a paths file if you want, in auto controlnet isn't native it's an extension so probably on the Settings tab -- when in doubt though everything always supports symlinks. Symlinks are the magic trick to redirect any folder any time for any program

[jagged plover]

img2img with very high denoise/creativity (eg 0.95 or higher even) will usually copy colors and nothing else

[plucky marsh]

hey guys, i have created an image of a girl with a yellow tshirt and the butterfly image on her shirt, how can i transfer that butterfly image onto second picture of a girl wearing the same yellow tshirt but in a different pose?

or if that's not possible, how do i remove the butterfly / text from the shirts so just that it stays a regular yellow tshirt with no symbols / images on them?

[hollow pike]

Photoshop

[stuck knot]

Hey has anyone got advice on this problem with SDNext the branch of Automatic 1111, I have installed SDNext from fresh two times, and I kept the Default settings, these are my System specs:
OS: Windows 11 Pro,
CPU: 13th Gen Intel Core i9-13900K,
Graphics Card GPU: NVIDIA GeForce RTX 4080, 16GB,
Memory: 32 GB ram

SD 1.5, and SD 2.1 work perfectly but I can't run SDXL 1.0 models I just keep getting errors, what could I be doing wrong?

[ornate elk]

Do you get any error on the cmd ?

[vapid sluice]

Why can’t I load SDXL base model and SDXL VAE

No errors displayed, just “press any key to continue

Should I reinstall the webui for the 15th time ?

I use nvidia geforce 3060

[ornate elk]

You installed your webui in a onedrive synced folder. That can cause issues

[vapid sluice]

Idk why it moved automatically to one drive. I hate onedrive

How to fix ?

Idk how to move out of the onedrive

[ornate elk]

first you need to make a new folder on the C: drive and name it Ai (not on desktop, downloads, documents)
then you move the stable-diffusion-webui folder to that new folder.
After you moved your webui you won't be able to update it anymore.
To fix this go into the stable-diffusion-webui folder and click in the File Explorer bar 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
After that relaunch the webui-user.bat

[vapid sluice]

Thanks a lot

[ornate elk]

no problem 🙂

[chrome rampart]

cs1 is the best

[chilly sandal]

Hi! I've been trying to download the "MixTape /// Blues" model from Civitai, but it does not seem to work... I place both the checkpoint and VAE in their respective folders, but I cannot load any of them, whenever I click them, it loads for a few secs and then comes back to the old model.
I am not sure what error to look for in cmd, cause it's a mess there from what I've seen. Anyone else had this issue or any idea what to do?
(also, if I reinstall SD from scratch, do I have to backup everything before? the models, loras, outputs, etc?)
Ty for any scratch of help! 😄

[ornate elk]

hey, do you have an screenshot of the cmd with the errors?

and yes if your reinstall you need to save the outputs folder and the models folder

[karmic crown]

I have an extremely stupid question - if I use a1111 API, will the generated image be available in txt2img-images on the "server" machine as well? and if I use a different model than the one currently selected in UI, will it also be changed in UI?

[stuck knot]

This is the error I am getting in SDNext:
Error: model not loaded
Time: 7.05s|
GPU active 2 MB reserved 0 | used 1350 MB free 15026 MB total 16376 MB | retries 0 oom 0

[ornate elk]

hmm thats strange, is sd next updated?

[karmic crown]

do not use 640x640 for XL models, use 768x768

or i just misread the log?

[stuck knot]

Yes to the latest version build

[karmic crown]

wait, i think i saw a similar problem when changing models from xl to 1.5 while xl vae was still active. try to switch vae to automatic, and then change the model. if it sitll doesnt work, try to restart the whole A1111. it might sound stupid, but that actually helped

[cedar depot]

Tried with and without, same issue

Could it be the lack of prompts that make it blurry/

Seems like setting Hires,fix to none makes it more detailed but still blurred. I am really not sure whats going on

[spiral comet]

Why when I exceed the number of prompt 75 the SD shows the number of prompt in red and when I exceed it again in white?

[ornate elk]

i have the same gpu, so download this image and put it into the png info tab, then click on Send to txt2img. then generate.I used dreamshaper v8 and 84000-mse-vae.
You should get nearly the same image.

[ornate elk]

if it stays red that means there is an open ( that hasnt been closed by )

[karmic crown]

@ornate elk do you know the anser to those questions? for some reason it points to somone elses message when i click the link

[spiral comet]

but also why there is 73/75? why is the number 75 important what happens after 75?

[jagged plover]

these are both questions that you can find the answer to by just trying it

but - no it doesn't save the image by default, and yes the model change is shared by default

[karmic crown]

thanks alex

[jagged plover]

that's token count, SD's input is in batches of 75 tokens. Once you exceed that it has to use a larger batch size (ie slightly more VRAM and it will interpret the prompting a bit differently)

a "token" is approximately 1 word, though uncommon words sometimes take 2 or more tokens

(if you're curious about token details, use Swarm, there's a tokenizer under Utilities that will show you how your tokens break down)

[spiral comet]

if I have a GF 4080 I can not worry about it, does it matter because above 75 for example the results are much worse?

[jagged plover]

not worse, just different

give it a try and see what happens

[cedar depot]

Does the VAE get automatically triggered or ?

[ornate elk]

Nope it has to be selectet.
Best is to go into Settings. User Interface, Quicksettings and there add SD_vae
Then hit apply and reload ui
Then select the VAE at the dropdown

[cedar depot]

Done, attempting now

Oh fuck thats so much better

[spare swallow]

How is it possible that SDXL models are so much inferior in anime or NSFW style? I know 1.5 has been around a lot longer but SDXL has also been around a while and still there are no models to match the visuals of 1.5. Why? SDXL is not suitable for anime or NSFW? I used a lot of promps and models, still not even close to 1.5.

[ornate elk]

Looks good

[vapid sluice]

I’ve actually reinstalled the whole thing into my d drive cuz onedrive wouldn’t let me move the folder cuz there was “no space left”

And again, the same error

Very strange

[ornate elk]

do you use xformers?

[vapid sluice]

Yes

xformers is working

[ornate elk]

okay, how much RAM do you have?

make also sure to have at least 15gb free space on C

[vapid sluice]

Oh I don’t have

[vapid sluice]

Why tho ?

[vapid sluice]

You mean VRAM ?

[ornate elk]

because your PC has something called a Pagefile. That offloads data (cache) when the RAM gets to full. this file is hidden and located on C. its dynamicly managed by windows and can be larger or smaller. if you dont have any space left on C you block this file to get bigger. SDXL needs more RAM to load so the pagefile needs also be a bit bigger.

[vapid sluice]

Oh i see

[vapid sluice]

Sorry, how to block the Pagefile?

[vapid sluice]

Wait i have

But it still doesn’t seem to work

[ornate elk]

how much RAM do you have?

[vapid sluice]

16gb

Just got this error

[ornate elk]

okay then check the following:
for changing the pagefile
launch sysdm.cpl from the Start menu search or run box (Win+R) and navigate to Advanced –> Settings –> Advanced –> Modify
Set it to "system managed" for c: and turn it off for other disks
Then restart the PC

[vapid sluice]

Ok

Oh i think it’s working

Thanks a lot

[brisk heath]

is it normal to have multiple instances of this process? it started with only one, but automatic1111 extension tab's "Check for updates" button caused it to multiply/added 6 more (coincidentelly i have 6 extensions installed)

[ornate elk]

perfect, no problem 🙂

[ornate elk]

no you only should have one

[brisk heath]

okay, do you have a clue on how i can fix this?

here's how it looks like on the top process (in task manager)

[ornate elk]

i testet this rn with check for updates. and yes then you get multiple git tasks but they vanish after the check is done

if they dont stay then its okay

[brisk heath]

oh mine stays, weird...but i guess it's not a big deal eh

[ornate elk]

if they found an extension update you have to click apply and restart. maybe then they vanish

[brisk heath]

nah it just says everything was 'latest' and that's it, but i will try later on when i finish the current prompt i'm doing

thanks

and also do you know if everything here is necessary? can i do without any of them?

[ornate elk]

they are not necessary if you dont use the features they provide.
but for the best experience i wouldnt deactivate them

[brisk heath]

thank you 👍

[ornate elk]

no problem 🙂

[faint moss]

my ipadapter doesnt seem to work anymore...
the model loader can't find the model files - anyone had this with last updates?

[sleek birch]

Hi guys,
totally new to stable diffusion(and generally ai). Just installed it yesterday but since then I couldn't get past this error. I tried fixing it with the no half command line etc but just couldn't get it to work. Looking through the internet I found that it isn't that uncommon, hope someone can give me some tips.
thanks in advance
I'm working on a MacBook Air 2020(M1 Processor) if that's relevant.

[cosmic raptor]

is it possible to use controlnet outpainting with different modules like depth? I tried to do it, but because of depth image sd filling blank space with... textured blank space

[karmic crown]

do not use --no-half, I posted instructions here They were tested on both Silicon and Intel Macs

start from "existing install" and go to the end

let me know if you have any problems

[sleek birch]

I tried now with the yaml file but now I get a different error. I have had this one before but after restarting it normally went back to the "NansException" error. Now It keeps saying connection errored out

[karmic crown]

show me the console

also, which broser is that? do you have adblocker?

in your console you should have this

Python 3.10.13 (main, Aug 24 2023, 12:59:26) [Clang 15.0.0 (clang-1500.0.40.1)]
Version: v1.6.1
Commit hash: 4afaaf8a020c1df457bcf7250cb1c7f609699fa7
Launching Web UI with arguments: --skip-torch-cuda-test --opt-sub-quad-attention --upcast-sampling --no-half-vae --medvram-sdxl

also this Running on local URL: http://127.0.0.1:7860

[sleek birch]

this is what i got

[karmic crown]

please send your webui-user.sh

[karmic crown]

you had to change it

you didnt do all the steps

start with existing install, and go through the end

but do not skip any steps

do not miss this part

you need development version of torch, for some things in my command line params

[sleek birch]

Am I blind or are these steps not in the link you sent? This is what it says under "Existing Install"

[karmic crown]

you are reading the original posgt

not mine

[sleek birch]

ah that's why😅. sorry

[karmic crown]

just follow the steps and it should be fine

i am using sdxl models without any issue (except speed) 🙂

Loading weights [a88bfb6ffa] from /Users/viking/stable-diffusion-webui/models/Stable-diffusion/hephaistosNextgenxlLCM_v10LCMBF16.safetensors
Creating model from config: /Users/viking/stable-diffusion-webui/repositories/generative-models/configs/inference/sd_xl_base.yaml
Applying attention optimization: sub-quadratic... done.
Model loaded in 18.9s (create model: 0.6s, apply weights to model: 11.3s, apply half(): 4.1s, apply dtype to VAE: 0.2s, calculate empty prompt: 2.6s).

[karmic crown]

i just removed the sentence with a link to the original post, since it might be a little misleading

[sleek birch]

I managed to get to here. But I'm unsure wheter the "sed -i..." fix worked. I just copy pasted it into my terminal. is that correct?

[karmic crown]

it is a simple command that just replace text in one file, there is no output

[sleek birch]

so technically my sd should work now?

[karmic crown]

please do this

git status
and show me the output

[sleek birch]

in the stable-diffusion-webui folder?

[karmic crown]

yes

ok

but

have you done other sed command?

[sleek birch]

wait that's german haha. can you see what you need to?

[karmic crown]

from step 7

[sleek birch]

no

[karmic crown]

please do

it should be like this

[sleek birch]

ok done

[karmic crown]

    modified:   webui-user.sh`

do git status again

[wet hull]

Hey I'm having an issue. The thing's crashing my graphics card and is saying that it can't collect the query data.

[karmic crown]

and finally, do this to check if first sed was done properly
grep functional venv/lib/python3.10/site-packages/basicsr/data/degradations.py

[karmic crown]

it should return this
from torchvision.transforms.functional import rgb_to_grayscale

then just run ./webui.sh

[sleek birch]

it did return this yes

[karmic crown]

does it work now?

[sleek birch]

I got the process bar for the first time so i guess!

omg yes it did! thank you so much definitely would have not been able to solve that without you!

[karmic crown]

du bist willkommen 🙂

i can only understand some basic german 🙂

[sleek birch]

love it😁

[karmic crown]

now, go to https://civitai.com/ and find some nice models

and learn more about loras and other stuff

this model is nice realistic 1.5 model, that should work at a decent speed on your mac
https://civitai.com/models/84728/photon

sdxl models will work, but it will be painfully slow

[ornate elk]

what did you changed?

[wet hull]

I'm not sure, it just started suddenly. I'm trying to use 512x512 instead of 640x360

[ornate elk]

does it work after a PC restart?

[wet hull]

No

[ornate elk]

whats your GPU again?

[wet hull]

Radeon RX 560 Series AMD GPU

[ornate elk]

is your gpu driver uptodate?

[ornate elk]

if your gpu is dusty that error can appear too

[somber bison]

does someone know what exit code:1 no module named pip mean or how to fix it?

[karmic crown]

full log please, you should not get that error in normal enviroment

[somber bison]

how to do that

[karmic crown]

screenshot or copy paste the text with error

[karmic crown]

whole cmd (black) screen

most important things are above

[karmic crown]

have you check add to path when you installed python?

[somber bison]

yes

[karmic crown]

as far as i see that is amd gpu

check this tutorial

#🤝｜tech-support message

if you run pip --version what do you get? unknow command?

this looks like some kind of script, better follow that tutorial and do all manually

i cant help you more, since i am on mac

[somber bison]

i did all the steps

in tutorial

[karmic crown]

close that window

open cmd from start menu

and try again

[somber bison]

still not working

[karmic crown]

check again later, some of windows guys will help you for sure

[somber bison]

alright thank you

[karmic crown]

i can help with macs and linux servers without gpu, but not too much for pcs 🤷🏻‍♂️

[somber bison]

no problems its all good

[karmic crown]

python3 --version
works, right?

[somber bison]

nope

[ornate elk]

Install python 3.10.9 or 3.10.11

Then delete the venv folder if there is one

That looks indeed like an install script

[somber bison]

i downloaded 3.10.11

how do i check for venv

[ornate elk]

Look for a folder with that name in the stable-diffusion-webui folder

If there is none, launch the webui-user.bat

[hardy gulch]

can somebody help me with preview images not displaying for my extra networks? i have the images in the same folder as the model and is named the same thing.

it shows a tiny missing image in the top left and clicking on it shows
{"detail":"Not Found"}

[somber bison]

i can copy the files so i dont have it in the folder

cant*

[ornate elk]

If your from Asia you need a VPN for the install

[somber bison]

im from europe

[ornate elk]

Okay then check that you have a stable internet connection

[ornate elk]

Because the error could be due unstable connection or something blocking it like a proxy, VPN, etc

[somber bison]

i checked

the connection is good

[hardy gulch]

the link when i open the tiny missing image has the correct path and name for the stored image. this happens for all my loras too.

im running with --xformers --autolaunch --update-check --share --api

clicking the info icon also shows "there was an error getting metadata"

[rough locust]

is there any video to guide me to create perfect settings for my prompts

[karmic crown]

try to rename lora to something without brackets

[ornate elk]

Then try the git clone again

[hardy gulch]

ok ill try. it happens on my loras that dont have any brackets or special characters too, though, so i dont think that will fix it.

[karmic crown]

i just remembered something. 1.5 loras are not visible if xl model is selected

only compatible loras are visible

[hardy gulch]

i dont have an xl model selected, and no matter what model i have selected, preview images for all loras are never displayed.

[karmic crown]

loras work, but you do not have images?

[hardy gulch]

i do not have the preview images. i can generate the images perfectly fine.

[karmic crown]

copy paste and remove .preview

ghibli_style_offset.png

i am using civitai browser plus to download models and model organizer to set preview images and similar stuff

[hardy gulch]

i use civitai browser plus aswell and all the models i download dont have the preview images

i did this and refreshed the lora page and i still dont see the previews

[rough locust]

which model do u prefer

[hardy gulch]

meinamix

[karmic crown]

it shoud work now

try to refresh

[hardy gulch]

should i refresh my whole webui?

[rough locust]

is it good for hyper realistic

[hardy gulch]

no

[karmic crown]

refresh button inside lora shoud do it

[rough locust]

then ishould check civitai

[hardy gulch]

it did not work

[somber bison]

i tried 3 times still the same result

i unisntalled and installed again

[karmic crown]

@hardy gulch

i am using this one, and just set preview image location

[hardy gulch]

ok ill install that extension

[rough locust]

how your ui is like that

[karmic crown]

also in browser plus, you have same images

[karmic crown]

https://github.com/lobehub/sd-webui-lobe-theme

[rough locust]

thx

wow looks so hot

[somber bison]

ok i found it

i already had it dowloaded

[ornate elk]

For Lora's the .preview is not needed
Loraname.png is all you need

[rough locust]

whats controlnet i checked their website but didn't get anything

[karmic crown]

you can change the pose of your subject

[somber bison]

it says the same thing no module named pip

[rough locust]

ahh for that i was using open pose editor

[hardy gulch]

ok i tried removing those ones and refreshed like so and refreshing still has no effect on the preview images not displaying.

[karmic crown]

i am using this https://civitai.com/models/140117/latent-layer-cameras mostly

[ornate elk]

Okay, is your webui updated?

[rough locust]

ah you can rotate face for me its not allowed

[hardy gulch]

yes

[karmic crown]

this refresh button

[rough locust]

can u simply explain what is lora and checkpoints

[hardy gulch]

when i open the missing image, i get this link: https://c1a06631808eed0247.gradio.live/sd_extra_networks/thumb?filename=C%3A/Users/jeffs/Desktop/homework/stable-diffusion-webui/models/Lora/Ghibli_v6.png&mtime=1700772052.2240708

[rough locust]

i know them as
checkpoint: create car
lora: edit car

[karmic crown]

model - used for creating pictures
lora - used to style a picture (like a specific face or cloth)

clear enough?

[rough locust]

which means, first i have to create model
then i have to edit with image to image with lora ?

[karmic crown]

you can use both at the same time

[rough locust]

or i can do all in one prompt

oh koay

like niji_wushi style samurai ?

[karmic crown]

for example realistic model + specific person face lora

[rough locust]

oh okay

[ornate elk]

Make sure the webui is whitelisted in any browser adblocker

[karmic crown]

like https://civitai.com/models/84728 with https://civitai.com/models/3700/jennifer-lopez

[hardy gulch]

Ok I disabled all of my adblockers in my browser and refreshed the lora page and the webpage and still cannot see the preview images.

[hardy gulch]

this happens on all of my different browsers too and on different devices aswell

[karmic crown]

can you send an image

any of them so i can try to use it with some of my loras

but copy paste from your folder

[hardy gulch]

yeah ofc

[karmic crown]

blah, it is converted to webp

i wanted to see if those are real png files

[hardy gulch]

how are they converted to webp?

[karmic crown]

discord converted an image

[hardy gulch]

ohh nvm then

[karmic crown]

maybe if you zip it?

[hardy gulch]

yea ill try

ill send it over dm

[karmic crown]

@hardy gulch as i suspected

❯ file ghibli_style_offset.png
ghibli_style_offset.png: JPEG image data, JFIF standard 1.01, resolution (DPI), density 96x96, segment length 16, Exif Standard: [TIFF image data, little-endian, direntries=1], baseline, precision 8, 920x1376, components 3

that is JPG

rename it to jpg and it will work

[somber bison]

@ornate elk I checked there is no venv file and it says the same thing

[ornate elk]

Yea the promplem is the connection from your PC to the github where the webui is hosted

[somber bison]

so its beacuse of the internet connection?

[hardy gulch]

tried it with both jpg and jpeg, and it still did not work

[karmic crown]

the image you sent me is jpg, named .png

[ornate elk]

Yes or maybe something blocking it.
Try restart your PC for example then open up a cmd and type
Pip cache purge
Then make a new folder and git clone again

[hardy gulch]

im seeing it as a png. maybe the rar compressed it or something? idk honestly

[karmic crown]

anyway it works

[somber bison]

When i typed that in the cmd it says Pip' is not recognized as an internal or external command

[hardy gulch]

now i think my issue is something other than the name of the file. maybe a flag i have set or something?

[karmic crown]

do you get an error like this one?

> pip
> Traceback (most recent call last):
>   File "<stdin>", line 1, in <module>
> NameError: name 'pip' is not defined. Did you mean: 'zip'?

[hardy gulch]

or maybe something in my webui settings

[somber bison]

yes

[karmic crown]

that is python, not windows cmd

you need to run windows cmd

[somber bison]

I did

it says

pip' is not recognized as an internal or external command,
operable program or batch file.

when i typed Pip cache purge

[karmic crown]

https://www.howtogeek.com/235101/10-ways-to-open-the-command-prompt-in-windows-10/

[somber bison]

yes this is what i launched

[karmic crown]

when you lauch it, all of those are required to work
run them one by one and post a result screenshot

python3 --version
pip --version
git --version

[hardy gulch]

i disabled share, and i still cant see the previews. my launch flags is only --xformers now

[karmic crown]

what is your browser?

[hardy gulch]

opera gx, but it doesnt work on any browser

[karmic crown]

try with edge or chrome in incognito

[hardy gulch]

this is in chrome incognito, still does not work

here is my cmd window when i run it

[karmic crown]

even after you click refresh here

[hardy gulch]

yes

[ornate elk]

That means your python installation is broken too

[karmic crown]

username with strange characters can cause many problems if you installed python just for yourself, and not for all users

i would suggest to make a new windows user with engish letters only, and to install both for all users

[somber bison]

I installed it for all users

It is english letters

[karmic crown]

be sure to add to path

[somber bison]

I did

[karmic crown]

you edit the image, i see now

i show it was something else

my screen is small (13")

[karmic crown]

git is correctly installed

[ornate elk]

Python is not installed correctly

[karmic crown]

but python is not added to path

[somber bison]

I added it to path tho

So do i reinstall?

[karmic crown]

https://datatofish.com/add-python-to-windows-path/

check this

or just reinstall

[ornate elk]

Run the python installer again, click modify and then "add to system variables"

[rough locust]

torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 MiB (GPU 0; 5.79 GiB total capacity; 5.46 GiB already allocated; 13.81 MiB free; 5.65 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

its happening when i try to change model

but i can change models in foocus etc.

[karmic crown]

your command paramteres are not correct

[rough locust]

how

[somber bison]

Where

[karmic crown]

which gpu do you have?

[rough locust]

rtx 2060

nvidia

cuda

[karmic crown]

VRAM?

[rough locust]

6

[karmic crown]

what are your command line parameters?

[rough locust]

just ./webui.sh

[ornate elk]

click next

then it shows

[karmic crown]

you need to edit ./webui-user.sh and to add parameters for your card

[rough locust]

can u guide me or do you have link or do you have parameters already in your sh and can u share

[hardy gulch]

should i give up and just reinstall a1111?

[somber bison]

It says add python to enviorment variables

is that the one?

[karmic crown]

@ornate elk is nvidia master, he will tell you, i am mac guy 🙂

[rough locust]

ahh okay i hope he will see this ping

[ornate elk]

--medvram --xformers --no-half-vae

[somber bison]

I did it

[rough locust]

i'll try

[somber bison]

But it didnt fix it

@karmic crown

Is this ok?

[karmic crown]

like this
set COMMANDLINE_ARGS=--medvram --xformers --no-half-vae
in webui-user.bat

than run that one

[hardy gulch]

i have a 4070ti, should i use --no-half-vae?

[karmic crown]

VRAM size?

[ornate elk]

If you use sdxl models or anime vae files then yes

[hardy gulch]

12

cool 👍

[karmic crown]

this looks good, and it should work, relaunch cmd and try pyton3 --version again

[somber bison]

still the same

[karmic crown]

python

th

python3 --version

[karmic crown]

type quit()

you are in python shell

[ornate elk]

Your on windows 32 bit ????

[karmic crown]

somehow he ended inside python shell

[somber bison]

Im on 64

i did quit()

[karmic crown]

python3 --version

you need to see this

#🤝｜tech-support message

c:\

and the rest

and then type python3 --version

does if finally work?

[somber bison]

still dont work

[karmic crown]

that is still python shell

close that

open cmd

[safe inlet]

run from CMD, not clicking a python shortcut or not from start menu

[karmic crown]

if you see this

[karmic crown]

that is the right window

[somber bison]

but it still doesnt work

[karmic crown]

does python --version works?

[somber bison]

yes

[karmic crown]

and pip --version?

[somber bison]

nope

[karmic crown]

how about python -m pip --version

[somber bison]

It said no module named pip

[karmic crown]

what pyton --version retuns? 3.10.11 ?

[somber bison]

yes

[karmic crown]

and if you do this

python -m ensurepip --upgrade

does it install pip and does pip now works?

[somber bison]

It says this

[karmic crown]

this is your problem

[somber bison]

so theres a virus?

[karmic crown]

do you use some antivirus program?

[somber bison]

yes

[karmic crown]

it looks like it stops python to work correctly

you need to resolve that problem with their support first

we cant help you here until you have all 3 working python, pip and git

[somber bison]

so if i unisntall the anti virus it will work?

[karmic crown]

better ask them for help

what av is that?

maybe there is something somewhere rearding that problem

[somber bison]

I think its McAfee

[karmic crown]

i cannot find a solution to that problem 🤷🏻‍♂️

you can ask mcafee support for help or disable it (which i would not advise)

[ornate elk]

Wow I wouldn't use that, Windows defender is better

[somber bison]

I unisntalled it

Well it still doesnt work it still says that antivirus blocked it

[karmic crown]

have you restarted pc after uninstall?

[somber bison]

yeah

[karmic crown]

when i see this i am glad that i use windows like 10 times a year

[somber bison]

ok i did it

But it still shows me this

[lost crescent]

WAS Node conflict - anyone any idea how to resolve? Conflicted Nodes:Image Save [ymc-node-suite-comfyui], Save Text File [ymc-node-suite-comfyui]

[quaint leaf]

Hey i have a question

Is It possibile to run SD on a PC with 4gb vram without using the lowram option?

Cause its very slow 😦

[solar scarab]

says wrong python version but i installed 3.10

rx 7900xtx

@ornate elk

[ornate elk]

you need to delete the venv folder

then relaunch the webui-user.bat

[solar scarab]

i did

just did it again, same error

[ornate elk]

Uninstall python 3.12

Then delete the venv again

[solar scarab]

aha looks like its pushing through this time

thanks @ornate elk 👍

[ornate elk]

no problem 🙂

[quaint leaf]

Is this the right Place for this question?

[ornate elk]

yes its possible

whats your gpu?

[quaint leaf]

Nvidia 1050 ti

I check

[ornate elk]

you need --medvram --xformers --no-half-vae

[quaint leaf]

Is that possibile also if i am using comfyui

Cause the arguments are different

[safe inlet]

--normalvram --force-fp32 and xformers should launch default i think, i like --auto-launch also

[quaint leaf]

When i try to make an image It puts me in lowram

Now ill try with --force-fp

[safe inlet]

you might be working in to big a resolution also, work small, figure out how to upscale separately

[quaint leaf]

This Is the resolution i used

Maybe im using the wrong lora

[safe inlet]

i saw SDXL in the screenshot somewhere, if you are using that 504x504 might not be good for it, google SDXL resolutions

[quaint leaf]

And what should i use

[safe inlet]

i dunno its trained on 1024x1024 so any variation of that listed from your google search

[quaint leaf]

So i also Need to change browser?

Or ill Just try with 1024 1024

[safe inlet]

yes thats what i would try as thats what its good at

[quaint leaf]

But its strange that he puts me in lowram without telling me that there Is an error or something like that

[safe inlet]

probably can see it happening in the console, and its trying to help you 🙂

[quaint leaf]

What browser do u suggest?

[safe inlet]

shouldnt really matter if the page works, not gonna fix anything else

[quaint leaf]

I tryed but its still in lowram

[safe inlet]

can ctrl-alt-del/task manager see if anything is using GPU, shut them off if possible, can somehow shut off photo preview in comfy to lower vram use also, beyond that i cant really add anything, tapped out my knowledge

[quaint leaf]

Ok now ill see

[safe inlet]

i dont really know how diffrent models affect vram usage, maybe there is something there, i run off a 2080ti /shrug

[quaint leaf]

Yes there Is something that uses my GPU even when i finished the image making

Look

[safe inlet]

click on processes tab, far right colum sort by GPU use, see if there is anything like a normal program you can shut off, might just be normal windows stuff

[ornate elk]

windows and pytorch

[unique urchin]

I'm getting this everytime i try to install an extension, how do i fix this to stop it from happening in the future

[modest pier]

Hi yall, I am trying to install comfyui-reactor-node but keep getting an error. I have tried updating my python, reverting to 3.10.0, and git cloning the files. Kinda lost for what to do next. Here is the error info too.

[spark tusk]

Hey everyne,, im prob not in the right place for this but I ws wondering if there ws anyone tht might be able to help me finish setting up my SD,. Im hving problemsetting up -venv drives so I cn hve more then 1 verson of Python nd not interfer with the others?

[fair oxide]

hello guys im back with another error
please help

[arctic zinc]

I am having problems after I upgraded to Mac Sonoma

and python crashes

[formal trout]

hi, any guide on how to run img2vid on Automatic1111? I had switched to the last branch candidate

I'm downloading one of these models https://huggingface.co/stabilityai/stable-video-diffusion-img2vid/tree/main

[solid lion]

Hello, running into the following issue of "Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases." when trying to set up virtual environment. I have installed 3.10.6 Python and also added path to the environment variable but still running into the same issue. Any suggestions? Thanks!

[karmic crown]

i posted updated instructions for Mac here you should start from "Existing install" part. Please follow my instructions and not the ones from the original post. Let me know if you encounter any problems

[karmic crown]

remove the folder tmp inside your stable-diffusion-webui

[lost crescent]

Sounds like (as it says) the target directory/folder already exists and is a full folder. Rename the troublesome folder, and try again

[lost crescent]

https://stackoverflow.com/questions/8949252/why-do-i-get-attributeerror-nonetype-object-has-no-attribute-something

[lost crescent]

Type where python

[karmic crown]

that is totally irrelevant. it just explains what can cause that error in your python code. since he is just a user of a1111, it is not helping at all

[karmic crown]

even i am not sure what all those parameters are, i noticed that you have 500, 500 for some of them. if those are dimensions they should be 512, 512 as above

[ornate elk]

Right click and edit the webui-user.bat
In the line Python=,
you add the path to the 3.10.x python.exe

Then delete the venv folder

[ornate elk]

Hey, what you try to install ?

[idle fable]

Hello Gs when running my ksampler I got this error, does anyone know why is that?

Error occurred when executing KSamplerAdvanced:

free_upper_bound + pytorch_used_bytes[device] <= device_total INTERNAL ASSERT FAILED at "..\c10\cuda\CUDAMallocAsyncAllocator.cpp":540, please report a bug to PyTorch.

[fair oxide]

thanks, i dont get it

[lyric vector]

Got a problem with multidiffusion-upscaler-for-automatic1111 When I click generate I'm getting this error "RuntimeError: The size of tensor a (111) must match the size of tensor b (256) at non-singleton dimension 3" when ever I enable Noise inversion it gives me this error but if I uncheck it it works, Im doing this inside img2img

[stuck knot]

Same, I'm looking for Stable Video guide for Automatic 1111 to!

[lost crescent]

Where can I find Kandinsky Channel?

[fair oxide]

yea your right, i dont understand a single thing there

[cedar depot]

Got a weird thing going on, when opening SD, I have prompts saved from a few days ago that I used for testing and they always appear when re-opening SD. They arent the prompts I used like yesterday for example. Is there a way to change this?

[hallow latch]

How can I get a better way to Scale up images?

[fair oxide]

so uhh how do i fix this?

[fair oxide]

Hello, I would like to do a garment transfer with Stable diffusion. Has anyone done it before? I think a mix between Inpaint + Openpose would be a good solution, what do you think?

[feral mango]

hi everonbe. I need help with anaconda and autoamtic. Anaconda only offers me this options
Please the question is around if 3.10.13 is as comaptible as 3.10.6 not if I should jsut install python and not use ANACONDA. I am using ananconda because i need to have different versiosn of python.
Should I install just 3.10.6 from command line anaconda... or can i use 3.10.13?

[karmic crown]

a1111 uses venv, so you will basically use 2 different virtual environments. i recommend installing 3.10.6 outside of conda. but yes, 3.10.13 should work (i use that one on mac)

[karmic crown]

as I already wrote, I am not sure what you are doing, but I noticed that you have two parameters with a value of 500. Start by checking if you have dimensions 500x500 anywhere and change that to 512x512. if that is not a height and width, i have no other ideas for now 🤷🏻‍♂️

those two

[fair oxide]

oh sorry... i was training a hypernetwork, image size is 512x512 by default for me and i do not change that

[fair oxide]

these two must be the 500 you were talking about

[karmic crown]

so, those are not related to size, in that case forget what i wrote

check this https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/5490

[halcyon tiger]

where can I ask a question about hardware...?

[thorny horizon]

can anyone tell me how to install playground v2 on Foocus UI ?, on the huggingface page i find many files other thatn the main safetensor files

[solid lion]

I am trying to generate images using the code so to do that I am trying to set up virtual environment so then I can install required libraries.

[chilly sandal]

Sorry for the very late reply, heres the ss of the cmd. As a reminder of my issu, I cant load a model (mixtapeBlues). It basically just sais it fails...

[ornate elk]

Is your webui updated?

[chilly sandal]

yep, updated it last night

[ornate elk]

okay, do other models work?

[chilly sandal]

yeap, yeap, I've been working with a few other models, this is the forst one to cause issues

[ornate elk]

Have you tried redownloading it?

Also what's your torch and xformers version?
You can check it at the bottom in browser

[chilly sandal]

and yea, tried to redownload the model like 4 times

is the fact that versiom is 1.6.1 and the model is for SD 1.5 a problem?.... didnt notice that before

(the others I used are for 1.5 usually too I think... Im still a newbie at this, sorry)

[ornate elk]

No that's fine. The webui version isn't the model version

[formal trout]

Really don't know 😅 what are people actually using for SD video? It's seems there are several approaches, don't know which is best. It seems there is a model for img2vid, but also AnimateDiff, Deforum, I don't know if they combine those or what models they use or what.

[ornate elk]

It depends what you want to create

They are for different purposes 😄

@odd fractal
Here is a quick Guide to install Automatic1111 Webui (Stable Diffusion)

You need to install Git 64bit: https://git-scm.com/download/win
and Python 3.10.11, 64bit: (any python above or 3.10.6) No 3.11, 3.12 https://www.python.org/downloads/release/python-31011/
And check "add python to path" when installing Python.

1. Make a new Folder on your drive (not on Desktop, downloads, documents, Programms, Onedrive) and name it Ai for example:
   C:\Ai\
2. You go into the folder you created in this case Ai, then click in the File Explorer bar (not searchbar) and type cmd then press enter.
3. Then you copy and paste this command:
   git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
4. Press enter and after its done you can close the cmd and launch the webui-user.bat inside the stable-diffusion-webui folder.
   After the Installation that can take a while. Youll get an URL http://127.0.0.1:7860
   Thats the webui you open in Browser.

[formal trout]

just some video

It seems they are various methods and Comfy workflows but I want to do it on Automatic1111

hey also what about ControlNet, it is very tedious that at certain point after some error it stops working all along, I have to restart SD

or after hooking two IP-Adapters it stop working

[sharp lichen]

what is a good desnoising strength for latent upscale?

[proper pike]

Can anyone help me figure out why a 420 frame animatediff / sdxl takes over an hour and a half to render?
https://youtu.be/KgskT-98vww

[subtle quarry]

Hey does anyone know why this is happening?

workflow in case someone needs it

plugging in a vae and seeing if that does anything...

I had the same lora twice, sorry folks im a dumbass

[fluid canyon]

hello guys!
is there a way to get from a Photo A to Photo B in Deforum ?

[odd fractal]

what does it mean when its just stuck on this

[arctic zinc]

@karmic crown thank you so much I spent days trying to look for solutions and I'm very greatful. I know Mac generally takes longer to generate compared to windows. Is there anything that could help with speed?

[formal trout]

Hi, I 'm trying to inpaint in Automatic1111 with an IP-Adapter ControlNet Unit, loaded with an independent control image, but it doesn't work well.

I noticed it delivers this after, so for some reason it is cropping/zooming the reference image. I tried "pixel perfect" and other pre processor resolution but it is always the same.

With "Reference" it deliver the processed image full, but it doesn't seem to help inpainting what I want

With "Revision", same as IP adapter, zoomed

I discoverd that it works inpainting "Whole picture", and the processed image is delivered in the right size, ,but I want to inpaint "only masked"

For some reason the image resolution affects the processed reference image, here I inpainted (only masked) in 16:9 and it delivered this

[ornate elk]

Hey, you need to have "add python to path" selected when installing python

You can run the python installer again, click modify, next, and then select "add python to system variables"

[formal trout]

oh I found someone with the same problem, maybe he describes it better, no answer xD https://www.reddit.com/r/StableDiffusion/comments/1740b61/prevent_controlnet_image_from_being_zoomed_in_and/

I tried manually cropping the image but now the inpaint seems to be broken, and my head hurts

[ornate elk]

Your on AMD right? If you use inpaint in img2img then you need --no-half

But inpaint with controlnet should work without

[graceful loom]

why am i getting this error Error: 'invalid literal for int() with base 10: ' 3000.5''. Before reporting, please check your schedules/ init values. Full error message is in your terminal/ cli.

pls help

[formal trout]

ohh I'll try that, with my old amd GPU I think I have to use that. But inpainting works without it, but I'll try it anyway

[formal trout]

well the problem is this https://www.reddit.com/r/StableDiffusion/comments/1740b61/prevent_controlnet_image_from_being_zoomed_in_and/

exactly

[formal trout]

It actually work inpainting with whole picture, but the inpaint is awfull as it needs more zoom. With "only masked" it do what the reddit post says

BTW I keep trying to run this on Automatic1111 https://huggingface.co/runwayml/stable-diffusion-inpainting

no dice

I read someone in civit.ai claiming that it works now on Automatic1111.

So I tried and it gave me out of memory, tried a lower resolution and it gave me "use --no-half", and then Nans problems etc

[karmic crown]

that is the fastest you can get with A1111 and your hardware. the command line i posted is actually optimized as much as possible for macs

[mellow finch]

@karmic crown i tried the steps for running web UI on macOS

while running git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui there were errors

@karmic crown can you help?

@karmic crown

can anyone help ⏫ ??

[vocal burrow]

network error during the process

you probably got disconnected or something

try again with a proper connection (no wifi for instance)

[summer tinsel]

Hi guys, we are using your API and for some presets, when selecting 2 or for images to be generated at a same time, those images are mostly blurred. The output looks like this: https://prnt.sc/SZTiuCm_wihl

Here is the used prompt: red haired slim woman on Starboard carbon paddle board in rough sea, lightning in the sky, foam tipped wave around

Prompts like "Photo of a giant steampunk mecha fighting cthulhu in the ruins of a city. The city is half submerged in murky water." work with no issues.

"style_preset" is not selected.

Could you let me know hwat is causing this and how to fix it? Thank you.

[fair oxide]

thank you
i just had to run
git clone https://github.com/deforum-art/sd-webui-deforum extensions/deforum

[vocal burrow]

might be some NSFW filtering, try something less likely to trigger nude to confirm

[mellow finch]

thanks a lot

the install has worked

SD is up and running now

but i'm getting a different error

[mellow finch]

Can you help with this error: 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.

ok i just fixed it in settings

but results are terrible

[craggy tusk]

hello I have done this tutorial many times ( https://www.youtube.com/watch?v=Hmk2lYPK_Bc ) without any problem.. I just recredited my account and I got this error " RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument mat1 in method wrapper_CUDA_addmm) " I don't understand.. can you help me?

[vocal burrow]

are you using a laptop ?

Regardless you can force auto1111 to use a specific hardware with the --device-id xxx (x being your gpu id, you ll find it in task manager) flag

problem happens because part of the pipeline seems to be working on the cpu while another part is on the gpu

[craggy tusk]

thank you for the answer I don't use it locally but on https://diffusionhub.io/

[vocal burrow]

same answer then

use the --device-id xxx command line argument and see if it helps (with xxx being 0 or 1 most likely)

[craggy tusk]

ok I try this thank you ..I am a beginner where do I type? the argument line

[vocal burrow]

Command line arguments explained like I'm five :

• First off, golden rule; .bat files are for windows, .sh for linux/mac, do NOT modify webui.bat/webui.sh. The only file you should be modifying ever is webui-user.bat/sh
• What is a command line argument/arg ? Those are words you use when launching a program to specify how it should work/trigger a specific behavior. They can look like this --xxx or --xxx y (y being a specific value associated with the -xxx command line argument)
• How do I set those ? Edit (with notepad or whatever text editor you want) your webui-user.bat/sh file and then :

• for windows users : find the following line set COMMANDLINE_ARGS= and set whatever command line arguments (and its associated value) on the same line after the =. (eg : set COMMANDLINE_ARGS=--xformers --no-half-vae --medvram-sdxl)
• for mac/linux users : find #export COMMANDLINE_ARGS="", remove the # to uncomment it and then do the same as windows users

• Is there a list explaining what those arguments do ? What values I should use with such and such argument ? Check it here https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Command-Line-Arguments-and-Settings#all-command-line-arguments

but because you re using someone else`s computer you ll have to find a way to modify their webui-user.bat/sh on your own

(wow they didn t even setup proper licensing / about us / troubleshooting / etc page on their website.... it goes straight up to "here s a the service" page with little to no details. And I m guessing the "here s how much it costs" page goes up the moment you give them your google account)

[craggy tusk]

Yes no contact on the site page.. I will try to contact them by email because I do not see how to interact on the site itself to carry out manipulations since it is a collaboration.. thank you for your help

[vocal burrow]

I see they have a discord link in the footer

might be faster to get information that way

That s the problem with "All in one/Ready to use solution" most of the time you don t actually own anything nor have proper access to things. Good luck.

[craggy tusk]

thank you for your time buddy

[karmic crown]

those are unexpected connection problems, try again

[daring jasper]

Hi!!! I have stoopid question) Well I tried to used Extras with deoldify option ) and it shows me this: TypeError: 'NoneType' object is not subscriptable
Time taken: 2 min. 47.7 sec.

A: 4.31 GB, R: 5.05 GB, Sys: 3.0/3 GB (100.0%)

[lilac cobalt]

I'm trying to download stable diffusion but I don't know how to do it.

[karmic crown]

what OS are you using and what is your GPU?

[lilac cobalt]

SO: windows 10 GPU: Intel(R) UHD Graphics

[karmic crown]

you can follow this guide -> #🤝｜tech-support message

[lilac cobalt]

I have done all the steps but I get this

@karmic crown

[fallen nymph]

same ;c

i use amd just sayin

[karmic crown]

you need to fix your command line parameters, give me a second

[karmic crown]

if you have radeon, you need another guide

[karmic crown]

this one is for you -> #🤝｜tech-support message

[fallen nymph]

amd rx 6700xt is what i got

ya but i get this

okay nvm, for some reasom my terminal doesn't work but cmd does 💀

[karmic crown]

you can actually do one by one (without combining them with &&), but yes cmd will work 🙂