🤝｜tech-support

[ornate elk]

cant help much with mac stuff, but @karmic crown can maybe take a look, sry for the ping 😅

[fair oxide]

anything helps

🙂

[karmic crown]

i am here

[fair oxide]

yay

[karmic crown]

https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/5461#discussioncomment-7557406

[fair oxide]

Ive been doin that

[karmic crown]

export COMMANDLINE_ARGS="--skip-torch-cuda-test --opt-sub-quad-attention --upcast-sampling --no-half-vae --medvram-sdxl"

[fair oxide]

still nothin

oh ok

[karmic crown]

this should be your command line

[fair oxide]

but thanks

ill see

gimme a min

[karmic crown]

i updated that a few times

🙂

[fair oxide]

this is my current file:

#########################################################
# Uncomment and change the variables below to your need:#
#########################################################

# Install directory without trailing slash
#install_dir="/home/$(whoami)"

# Name of the subdirectory
#clone_dir="stable-diffusion-webui"

# Commandline arguments for webui.py, for example: export COMMANDLINE_ARGS="--medvram --opt-split-attention"
#export COMMANDLINE_ARGS=""
export COMMANDLINE_ARGS="--skip-torch-cuda-test --opt-sub-quad-attention --upcast-sampling --no-half-vae --medvram-sdxl"

# python3 executable
#python_cmd="python3"

# git executable
#export GIT="git"

# python3 venv without trailing slash (defaults to ${install_dir}/${clone_dir}/venv)
#venv_dir="venv"

# script to launch to start the app
#export LAUNCH_SCRIPT="launch.py"

# install command for torch
#export TORCH_COMMAND="pip install torch==1.12.1+cu113 --extra-index-url https://download.pytorch.org/whl/cu113"
export TORCH_COMMAND="pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu"

# Requirements file to use for stable-diffusion-webui
#export REQS_FILE="requirements_versions.txt"

# Fixed git repos
#export K_DIFFUSION_PACKAGE=""
#export GFPGAN_PACKAGE=""

# Fixed git commits
#export STABLE_DIFFUSION_COMMIT_HASH=""
#export CODEFORMER_COMMIT_HASH=""
#export BLIP_COMMIT_HASH=""

# Uncomment to enable accelerated launch
#export ACCELERATE="True"

# Uncomment to disable TCMalloc
#export NO_TCMALLOC="True"

########################################### ```

it gives me this when launching:

Launching Web UI with arguments: --skip-torch-cuda-test --opt-sub-quad-attention --upcast-sampling --no-half-vae --medvram-sdxl
no module 'xformers'. Processing without...
no module 'xformers'. Processing without...
No module 'xformers'. Proceeding without it.
Warning: caught exception 'Torch not compiled with CUDA enabled', memory monitor disabled
Loading weights [62bb78983a] from /Users/kaare/stable-diffusion-webui/models/Stable-diffusion/epicphotogasm_lastUnicorn.safetensors
Creating model from config: /Users/kaare/stable-diffusion-webui/configs/v1-inference.yaml
Running on local URL:  http://127.0.0.1:7869

To create a public link, set `share=True` in `launch()`.
Startup time: 5.1s (prepare environment: 0.1s, import torch: 1.8s, import gradio: 0.6s, setup paths: 0.7s, initialize shared: 0.2s, other imports: 0.6s, load scripts: 0.3s, create ui: 0.3s, gradio launch: 0.4s).````

this? Warning: caught exception 'Torch not compiled with CUDA enabled', memory monitor disabled

is that normal

[karmic crown]

yes

we do not have nvidia

xformers and cuda are nvidia stuff

that is not an error

[fair oxide]

yeah

welll this is new:

[fair oxide]

a bit more filling output

says this on the site: 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.

[karmic crown]

go to your stable-diffusion-webu folder

and type this
git status

[ornate elk]

@lone wolf i now get the same error as you, never had that before with 512x512

[fair oxide]

Last login: Fri Dec 1 21:40:12 on ttys000
kaare@Kaare-sin-Air ~ % cd stable-diffusion-webui
kaare@Kaare-sin-Air stable-diffusion-webui % git status
On branch master
Your branch is up to date with 'origin/master'.

Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: modules/sd_disable_initialization.py
modified: webui-user.sh

Untracked files:
(use "git add <file>..." to include in what will be committed)
.DS_Store

no changes added to commit (use "git add" and/or "git commit -a")
kaare@Kaare-sin-Air stable-diffusion-webui %

[lone wolf]

Yeah, it seems to only happen to 512

everything else works very good

its weird af

[ornate elk]

so 768x768 works ?

[lone wolf]

Idk, but if I just increase height, it works fine

If its like a phone ratio

[ornate elk]

interesting

[karmic crown]

this is fine

[fair oxide]

okay..

[karmic crown]

try this

pip cache purge

then remove venv

[ornate elk]

hmm, i tried 512x768 and got also errors.
but now i saw that ublock was activated for the webui again. I whitelisted the webui in it and now it works!

[fair oxide]

kaare@Kaare-sin-Air stable-diffusion-webui % clear

kaare@Kaare-sin-Air stable-diffusion-webui % pip cache purge
zsh: command not found: pip
kaare@Kaare-sin-Air stable-diffusion-webui %

Should I do pip3

that usually works

[lone wolf]

How do you do that? why does it block? D:

[karmic crown]

venv/bin/pip cache purge

[ornate elk]

idk maybe ublock got updated.
if you use ublock you can click on the ublock icon and then hold ctrl and click on the blue Off/on button
that will whitelist the webui

[karmic crown]

easer than to activate venv 🙂

[lone wolf]

Ublock origin you mean?

[fair oxide]

kaare@Kaare-sin-Air stable-diffusion-webui % venv/bin/pip cache purge
Files removed: 1104
kaare@Kaare-sin-Air stable-diffusion-webui %

now?

[lone wolf]

Aaaah

[karmic crown]

remove venv and start webui-user.sh again

it will generate new venv

then just run sed -i '' 's/from torchvision.transforms.functional_tensor import rgb_to_grayscale/from torchvision.transforms.functional import rgb_to_grayscale/' venv/lib/python3.10/site-packages/basicsr/data/degradations.py

and it should work properly

[fair oxide]

alr ill try

[fair oxide]

full output:

[karmic crown]

ModuleNotFoundError: No module named 'torchvision.transforms.functional_tensor' is expected, they changed the name of module

run this

sed -i '' 's/from torchvision.transforms.functional_tensor import rgb_to_grayscale/from torchvision.transforms.functional import rgb_to_grayscale/' venv/lib/python3.10/site-packages/basicsr/data/degradations.py

[fair oxide]

kaare@Kaare-sin-Air stable-diffusion-webui % sed -i '' 's/from torchvision.transforms.functional_tensor import rgb_to_grayscale/from torchvision.transforms.functional import rgb_to_grayscale/' venv/lib/python3.10/site-packages/basicsr/data/degradations.py
kaare@Kaare-sin-Air stable-diffusion-webui %

done

now?

[karmic crown]

just run webui

and it should work

[fair oxide]

It does !!!

Im using a 2gb model thinfy btw

is that bad or good

[karmic crown]

you can use any model you want now

[fair oxide]

Yayyyyyy

[karmic crown]

even 6GB SDXL

but it will be slow

but it will work 🙂

[fair oxide]

so there isnt any like limt of how low

?

🙂

[fair oxide]

so no limt

[karmic crown]

i solved all the errors that i had 🤷🏻‍♂️

limit is time

SDXL will be slower than 1.5

i am using 512x512, 512x768, 768x512 for 1.5

for SDXL i am using 1024x1024, 768x1024 and 1024x768

for anything bigger user upsace, do not set bigger resolution

[karmic crown]

Launching Web UI with arguments: --skip-torch-cuda-test --upcast-sampling --no-half --use-cpu all --opt-sub-quad-attention --no-gradio-queue --listen 🤣

Time taken: 2 min. 53.9 sec.```

not bad for a VPS without GPU, right? 🙂

[onyx grove]

@ornate elk

hey

i've currently switched over to runpod for sdxl

however

rn my thing keeps crashing for some reason

cpu seems to always be spiking to 100%

[balmy tinsel]

Hi guys i just started messing around with SD I installed fooocus ui because it was the first one that was presented to me. I have a laptop with rtx 3050 4gbVram and 8gb normal ram , at first it was fine the images took a while to generate but they did eventually now it keeps on give the same error:

return torch.group_norm(input, num_groups, weight, bias, eps, torch.backends.cudnn.enabled)
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB. GPU 0 has a total capacty of 4.00 GiB of which 1.41 GiB is free. Of the allocated memory 1.72 GiB is allocated by PyTorch, and 74.91 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 would also like to know if there is a better ui that have more specifications for low vram models such as mine

[ornate elk]

Hey with that vram amount you need to make some changes to the webui-user.bat for better performance and less vram usage.
Edit the webui-user.bat with right click.
Then at the line COMMANDLINE_ARGS=
You add: --xformers --medvram --no-half-vae
Then save and relaunch

[ornate elk]

Does it run on GPU ?

[balmy tinsel]

idk where that file is , I have webui.py tho

[ornate elk]

Its in the stable-diffusion-webui folder

Oh wait sry I didn't read properly

You installed foocus

[balmy tinsel]

yes

[ornate elk]

Can't help with that :/
Most users use Auto1111 or comfyui

[balmy tinsel]

im fine changing the ui i just need to know to wich one

comfy seems a bit agressive rn

auto1111 i can tweak thinks for the vram then?

[ornate elk]

Auto1111 would be the best for your GPU and usage:
Follow my install guide:
#🤝｜tech-support message

Then follow my advise from above (the file edit)

[onyx grove]

how would i know that?

this the spec btw

[ornate elk]

Idk much about runpod

[onyx grove]

damnn

is okay then thanks for your help

[balmy tinsel]

k thx

[onyx grove]

@ornate elk is this a good spec for SDXL?

[fading mural]

Hi, does anyone have a Comfy UI workflow that mimics the controlnet preprocessor colorfix+sharp from automatic111 ? I can't find anything anywhere.

[pearl crane]

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

[nocturne wraith]

i just cannot figure, why the hell resize canvas not work when using stable diffusion pose editor

fuck me

[sharp lichen]

what does controlnet is more important vs prompt is more important do?

[sharp lichen]

your gonna want more gpu ram i think

12+

but i think it will still work but slower

[onyx grove]

;-;

so vram plays a big part

doesn't matter the gpu

[sharp lichen]

yeah i think you want lots of vram for SD

[rotund lava]

amd sucks for SD unless u install linux

u prob should buy a new gpu instead of wastin money on the case+ the windows 10 key

[ornate elk]

Hey, 6gb vram with an older AMD card won't work

[ornate elk]

The biggest part for Ai stuff

[sharp lichen]

what is the ideal prepocessor resoultion for ip-adapter in xl?

can i select pixel perfect?

[little bough]

Hi, I keep getting an error with kohya hires fix, it worked fine when I installed it when it first came out but now it keeps throwing errors, it will generate an image for a few seconds and then stop. I've made sure a1111 is up to date and updated all my extensions and it persists, any ideas? it gives an error with a list of things and ends with this and then even when turning hires fix off I need to restart a1111 as it keeps throwing errors **" RuntimeError: The expanded size of the tensor (7488) must match the existing size (8748) at non-singleton dimension 1. Target sizes: [2, 7488, 320]. Tensor sizes: [1, 8748, 1]": **

[gray oyster]

@ornate elk sorry for ping but i contacted the guy again

if you want i can arrange a group DM

[ornate elk]

wow and he uses the directml fork of auto111 ?

[gray oyster]

i guess

he is using stock auto1111 he says

[ornate elk]

ah well then he is on linux

[gray oyster]

yea he is, he is on arch-linux

[ornate elk]

with rocm drivers you can run native auto1111

[gray oyster]

yea

but that fast?

god damn

[ornate elk]

thats pretty slow, it could do 20 it/s

[gray oyster]

though why is my card so underperforming

[ornate elk]

what was your gpu again?

[gray oyster]

4060 Ti

i can overclock it like a little bit more but i dont know whats more important if its memory speed or clock speed

[ornate elk]

how long does it take for an 1.5 model, 512x512 with 30 steps?

[gray oyster]

screenshot from yesterday

although thats 50 steps, lemme do 30

[ornate elk]

that looks good

12 it/s is fast

[gray oyster]

yea but 2.5it/s on SDXL seems slow to me

considering im not running out of memory

theres always like a 6gb headroom

unfortunately i cant just trade that vram for performance

[ornate elk]

it can only be a bit faster if your using comfyui with sdxl

[gray oyster]

yea but i cant use that from a different device, or can i?

[ornate elk]

only by using an remote desktop software

[gray oyster]

unfortunate..

[ornate elk]

or maybe with portforwarding

[gray oyster]

well if its on a local port only then i cant connect to it anyway iirc

[ornate elk]

it should be possible

[gray oyster]

not if its running on a 127 adress

thats only accessible within that computer

afaik

192 adress is accessible from within the network

or through port forwarding

guess i will just have to seethe with the poor performance for now

until TensorRT becomes available for sdxl

[ornate elk]

yep thats right

[ornate elk]

its not poor. xD on my gtx1080 an sdxl image takes 1:30

[gray oyster]

thats fair but i expected like atleast 3it/s especially with xformers

[ornate elk]

the 4060ti has a smaller bus size. maybe thats a factor here?

[gray oyster]

also the gtx is on the pascal architecture and doesnt have tensor cores iirc

[gray oyster]

could be

although

isnt that only for moving data between the cpu and the gpu?

not sure though

[ornate elk]

idk if it plays a big role in SD

[gray oyster]

does SD even use the tensor cores on gpus which have them?

maybe my cuda version isn't optimal for torch?

i got 12.1

[rotund lava]

it probably uses cuda cores since tensor cores cant do all math

idk why u gettin 2.5 its on sdxl on a 1024x1024 img but u should get like 7 its try to downgrade nvidia drivers if nothing else works

[gray oyster]

i also noticed that overclocking doesnt do much

[rotund lava]

yea thats useless

[gray oyster]

did like 0.2it/s

[ornate elk]

best supported is 11.8

[gray oyster]

imma try the studio drivers, they should be a little older

[gray oyster]

do i have to reinstall auto1111 if i switch the version?

because the torch installation would be invalid

[ornate elk]

you would need to delete the venv folder. then everything gets rebuild with the new torch etc

[gray oyster]

ah okay, thanks

imma try that

[gray oyster]

bychance do you know any nvidia driver settings i should change?

[rotund lava]

531.61 if u have enough vram and want speed, if u dont have enough vram and want to use ram then 536.99 but this is just me,im still using 536.99 cause i only have 6gb of vram so it swaps to ram to finish rendering even tho is slower

[gray oyster]

i have 16

so 531.61 should work

but driver versions are different between all card series's

[rotund lava]

yea give it a try if it doesnt work u can always go back to new version

[gray oyster]

yea

yea my card isnt listed

this looks interesting

in 546.01

i have the fallback enabled, i should maybe disable it

[karmic crown]

https://nvidia.custhelp.com/app/answers/detail/a_id/5490/~/system-memory-fallback-for-stable-diffusion

[rotund lava]

do it like this https://nvidia.custhelp.com/app/answers/detail/a_id/5490/~/system-memory-fallback-for-stable-diffusion

[karmic crown]

🙂

we posted the same link at the same time 🙂

[gray oyster]

i have it as a driver setting and i enabled it beforehand

i thought it was a ReBAR feature because just before i enabled ReBAR in my bios

but i also had to make a driver update so that would explain why its there

maybe disabling it would also not make my ram oom's in oobabooga happen anymore

🤔

[rotund lava]

only way to know is to downgrade to other driver versions and test

[gray oyster]

well its a system memory issue, not a vram issue

because the model should fit nicely but my ram is just short of a few mb

and a swapfile also doesnt work

[rotund lava]

well driver does matter thats why some ppl downgraded

idk what they changed tho

[gray oyster]

and upgrading ram i have to buy 2x16gb because i cant do dual channel because my mainboard is broken and for buying a new mainboard i would need to buy a new one because this one isnt even being produced anymore because the ancient greeks calculated their routes for trading with it and while im at it i gotta buy a new cpu too because the socket is unsuported

[gray oyster]

yea but they dont change the behaviour of how the ram works, only how the gpu is being interacted with

and the OOM Ram issue is because oobabooga loads the model into the ram and then into the gpu

but my ram is smaller than the model

eg i have 12gb free ram (after using 4gb for windows alone) and the model file is 13.5gb in size

but lets see if auto1111 profits from different cuda drivers

[rotund lava]

if it doesnt work you could try to download another auto1111 put it in another folder install it with just basics and try to run it to see how many its u get

[gray oyster]

still 2.15it/s

and cuda definitely works

[rotund lava]

if u install a1111 from scratch the components version (at bottom) should look like this

[gray oyster]

the python version is different

[rotund lava]

yea looks all good it works on python 3.10.6 to .11

[gray oyster]

wait a minute

https://tenor.com/view/the-rock-gif-25266750

maybe it is slowing down due to vram

wait no

during diffusion thats not an issue

[rotund lava]

idk whats causing it but try to downgrade driver version,after that delete venv and let it regenerate 🫡

[gray oyster]

alright

which version though

imma just try the oldest non-archived

do i need to redownload the cuda drivers too afterward?

probably not actually

[azure magnet]

Is the new update of xformers and torch stable?

if so how can upgrade?

[gray oyster]

should be implemented in 1.6.0 Auto1111

[azure magnet]

@ornate elk sorry for tagging do you have a minute for 3 questions?

[gray oyster]

even with older drivers, oldest one i could find was 537 but no difference in speed

[long lodge]

i added custom comfyui folder and this happend:
17:04:50.810 [Info] Self-Start ComfyUI on port 7825 unexpectedly exited (if something failed, change setting LogLevel to Debug to see why!)
17:04:50.810 [Info] Self-Start ComfyUI on port 7825 had errors before shutdown:
ComfyUI on port 7825 error: Traceback (most recent call last):
ComfyUI on port 7825 error: File "/home/berk/ComfyUI/main.py", line 69, in <module>
ComfyUI on port 7825 error: import comfy.utils
ComfyUI on port 7825 error: File "/home/berk/ComfyUI/comfy/utils.py", line 5, in <module>
ComfyUI on port 7825 error: import safetensors.torch
ComfyUI on port 7825 error: ModuleNotFoundError: No module named 'safetensors'
(i installed requirements)

[ornate elk]

Hey, yes

[azure magnet]

Is the new update of xformers and torch stable?
if so how can upgrade?

[karmic crown]

how can i paste a lot of markdown text here properly?

[ornate elk]

check your version at the bottom of the webui

[azure magnet]

python: 3.10.11  •  torch: 2.0.1+cu118  •  xformers: 0.0.20  •  gradio: 3.41.2

[ornate elk]

your using the latest torch and xformers that is currently supported

[azure magnet]

i see someone using torch: 2.1.0+cu121  xformers: 0.0.22.post7

[ornate elk]

where?

[azure magnet]

on a korean site

[ornate elk]

okay, maybe hes using the dev build of auto1111

[lapis oak]

Attach a text file

[pallid solar]

Is there a way to specify that I want stable diffusion ( I'm using fooocus) to use my NVIDIA RTX 3050 and not my integrated intel graphics?

[ornate elk]

are you on a laptop?

[pallid solar]

Yeah

[ornate elk]

then open up the nvidia control pannel, at the left click on 3D Settings, then on the right select "Nvidia high preformance device" in the top dropdwon

[karmic crown]

it must be markdown

[pallid solar]

ok thanks let me test it now

[karmic crown]

A1111 install on MacOS instructions (December 2023 version)

ok, done 🙂

[pallid solar]

It looks like it's using slightly more of the NVIDIA one, but it's still using a lot of the Intel one instead of the NVIDIA one

[ornate elk]

okay, i cant help much with fooocus :/ auto1111 should work for your gpu too, if you need help with that
For foocus you can open an issue on github

[pallid solar]

Okay thanks, GPU doesn't affect quality, right? Only speed?

[karmic crown]

speed is the main issue

i have a laptop with intel+nvidia (which i didn't turn on for years) and it was a mess

oh, yes, be sure that you are on power, sometimes i used intel just because i was on baterry

it might change by now (it was a few years ago)

before i stared using macs

btw. even when you use nvidia it will also use intel, since that nvidia cant work by itself

https://www.nvidia.com/en-us/geforce/technologies/optimus/technology/

i am not sure if something changed, but this was a case with my old laptop

[pallid solar]

Ah I see thanks. Maybe I'll just use my PC with a 3060

[ornate elk]

good point, some laptops need to be plugged in for the GPU to work

[ornate elk]

would be a much better option

[karmic crown]

i guess there is an option to fix that, but i cant remember

it was like 5+ more years ago

[ornate elk]

most likely a bios setting

[karmic crown]

https://docs.fedoraproject.org/en-US/quick-docs/set-nvidia-as-primary-gpu-on-optimus-based-laptops/

if you do not mind switching to linux

i am trying to remember if that was possible on windows, but i just can't 😦

[pallid solar]

I had ubuntu dualbooted for my CS class but I accidentally wiped my entire hard drive when trying to do a clean reinstall and I haven't properly reinstalled it since lol

[severe kindle]

Rust still required ? Are they still using that ancient version of tokenziers that requires being build from source ?

[karmic crown]

no compiling

as i wrote i took the first few steps from the original post (including rust as requirement)

if you scroll up a little, you will see that my original post started with changing webui-user

[gray oyster]

did a fresh installation of oobabooga, any idea how to fix this?

nvm im an idiot

i ran update instead of start

https://tenor.com/view/wow-gif-20411229

[tough plank]

DM bot not working help

Stable dreamer bot

[ornate elk]

you cant DM the bots

[tough plank]

Its not working When i want to dm my generated image by clicking "DM A or B " button bellow generated images

[ornate elk]

ah okay. i tested it. for me its working.
Make sure to allow messages from "not friends or everyone" in the discord settings

[karmic crown]

what i want to be a friend with a bot? 🤣

I have a stupid question: how can I disable upscale? I mean, when I use it once and then reduce steps to 0, it uses the same number of steps I used for generation. but i have no idea how to disable it. i do not want to restart just for that

[ornate elk]

you have to click on the little arrow on the right 🙂

to collapse it

[karmic crown]

collapse will disable it?

"very intuitive" - NOT

but ok

[ornate elk]

your an apple user, it should be normal for you 😛

[karmic crown]

i am a webdeveloper and i know what my colleagues from UX team would say

[pearl crane]

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

[karmic crown]

I am also Samsung user, I had almost every phone from Samsung S series since S3 - my main phone is S23 Ultra, but yes I have an iPhone too, but just as a testing device for my work 🤣

[ornate elk]

I had an S3 too, such a good phone. Switched to Google Nexus 5X, then OnePlus 6T and now Google Pixel 7

My next phone needs an 8 in its name

[karmic crown]

S28? 🙂

[ornate elk]

Maybe 😄

[karmic crown]

this "coalpse to disable works for eveything"?

dynamic prompts for example

[ornate elk]

Only for hires fix and refiner

Not extensions

[karmic crown]

there is a checkbox

forget it

so i need to set fixed seed as i did

all clear

[ornate elk]

Why fixed seed?

[ornate elk]

Ah for dynamic prompt

[karmic crown]

also the main one

i am experimenting with lcm

tring to find best values

since it is very tricky with some loras

[ornate elk]

Okay, good to know. Didn't had the time to try it

[karmic crown]

i really need we need separate section just for LCM

[jolly gorge]

has anyone figured out how to get DAT upscale models like 4xFaceUpDAT working in automatic? They either give you errors or do nothing, but look like the best for photos of people in most tests

[plucky sun]

I'm trying to install SD, but I constantly get a problem, I reinstalled git, reinstalled python, deleted and downloaded the archive, can you help me?

[tranquil mountain]

I installed steable diffusion but I don’t know where to start the application

Pls help me

[ornate elk]

@tranquil mountain Follow my install guide:
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.

[tranquil mountain]

thank youuu so muchhhhhh

[plucky sun]

did not help

[ornate elk]

What's your error ?

[plucky sun]

the same error as above, every time you run the bat, the error changes, for example, at first it cannot install Torch, when you run it again, it installs but goes to another dependency, and stops at the error that is higher on the screen

[ornate elk]

Did you installed from scratch via my guide?

Can you share a screenshot of the error ?

[plucky sun]

yes

moment @ornate elk

[ornate elk]

Okay you need to delete the venv folder and the repositories folder.
Then go inside the stable-diffusion-webui folder and click in the explorer bar. Type cmd and hit enter.
Then type
Pip cache purge
Then hit enter.
Then type
Git pull
And hit enter again.
Then relaunch the webui-user.bat

[rotund lava]

531.61 works on 4060,it works on 4090 and its 3x the card that 4060 ti is

[gray oyster]

but it doesnt list the 4060Ti in the compatability tab iirc

[plucky sun]

if I now run the batch file again, it will install the stick but move on to the next error, and then everything will be the same

[ornate elk]

What's your GPU and python version?

[plucky sun]

3.10.11, Nvidia 3070

[ornate elk]

Hmm okay.
My last guess is that you need a VPN for the installation

[plucky sun]

Which country should you prefer to connect to? I have Finland and Kazakhstan

[ornate elk]

Finland would work

You need to run
Pip cache purge
Before the installation

Also you need to install python again with active vpn

[plucky sun]

@ornate elk Eh it didn't help 😦

[fair oxide]

Hello, i need help with these errors

[narrow fjord]

Hello! I do not know if this is the right Discord server to ask this but, here goes:

https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-NVidia-GPUs

In this link, I followed the instructions to install StableDiffusion but I am encountering this error.

I am on Arch and I ran the command provided for Arch.

I do not know if this error is Python version related but, the install command installed the 3.11.6-1 version, instead of 3.10.6

I did install the 3.10 version but it is still running the 3.11 one.

So I do not know what happened here.

[rich zodiac]

how much VRAM do you have?

in your graphics card

[narrow fjord]

I have a 3070 8GB.

I dualboot with Windows and use RVC Mangio over there.,

And it works fine with making AI covers.

But I wanted to try out this on Arch.

And it makes no sense that I encounter CUDA errors.

Unless my drivers are wrong?

Which would not make any sense since I game on Linux perfectly fine.

Here is the entire output that happens when I run webui.sh

Also getting this error as well.

Wait, I guess it worked?

I got the webGUI running here.

Nvm, I seemed to have fixed the issue somehow?

[hollow schooner]

Anyone know how to not use the second progress bar

Sometimes the images come out glitchy with red a blue everywhere

It’s usually when this bar (red marked) comes up and finishes and kinda just messes up the image

Like this

[tame yoke]

can u like downgrade torch to the version when my gpu was supported?

[tame yoke]

i think its the high res fix

disable it

[hollow schooner]

Tried that the bar still shows up

And it can sometimes still come out a little distorted like the image

[ornate elk]

Delete the venv folder so it gets rebuild with python 3.10

[ornate elk]

Can you show your txt2ing settings?

[ornate elk]

Hey, open up a cmd and run:
Pip cache purge
Then delete the venv folder and rerun the webui-user.bat

[vale prism]

Hey everyone, real noob here 👋

Im looking for recommendation for the first model I should try to use. Anyone have an advice ?

[ornate elk]

Hey, Dreamshaper v8 is a pretty good starter model

[vale prism]

Ty !

[hardy thorn]

I need help with this, can't launch automatic

[ornate elk]

What's your python version?

[hardy thorn]

3.10.6

[ornate elk]

Okay, install python 3.10.9
Then open up a cmd and type
Pip cache purge
Then delete the venv folder and rerun the webui-user.bat

[graceful folio]

Hi, could you please help me. Do u have experience running SDXL on Macbook? Im considering, whats better to buy? Air M2 24 GB, or Pro M3 16 GB. Do you know how long it takes to generate pictures there? thank you

[hardy thorn]

still got the error

[ornate elk]

Okay then uninstall python and reinstall it

[hardy thorn]

reinstall which version?

[ornate elk]

Uninstall both and install 3.10.9

[hardy thorn]

i did, the issue still there

[ornate elk]

Have you deletet the venv folder?

After that

[hardy thorn]

yes

[ornate elk]

Is your C drive full by any chance ?

Also restart the PC

Delete the venv folder again

If nothing works make a new folder and follow my install guide:

#🤝｜tech-support message

[hardy thorn]

I store them at drive E

[hardy thorn]

I made a new folder, and still get that error lol

[ornate elk]

okay, but is the C drive full?

there are temporary files tored

[hardy thorn]

no

[ornate elk]

Okay what's the full cmd log?

[gray oyster]

Try opening a cmd window and type pip install numpy, i wanna test if my theory with pipy is correct

Because at the end of your errorlog it says it has a "module not found" error

[karmic crown]

I am using SDXL on M1. It works, but it's slow. It takes a few minutes to generate a picture. But, with some "tricks" like using 1.5 models with upscale, LCM sampler and Lora, and similar "witchcrafts," you can get a decent image in fewer steps, which means a shorter time. Check out this one I generated last night #🏞｜general-with-images message

give me full specs for both macs, and I will tell you which one I would choose, it is not all about RAM

[grim moon]

Hi, Is coil whine normal while generating an image?

[karmic crown]

generating images needs more GPU power, which means that the cooler needs to spin faster to cool down the GPU, so i guess it is nothing unusual if it is in the same range as when you play some GPU-intensive game

if it is abnormal, check if your GPU cooler is dusty and clean it

if everything is clean and noise or vibrations are severe (above normal range), go to service and ask them to check it

[hardy thorn]

Already up to date.

exit code: 1

stderr:
Traceback (most recent call last):
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in run_code
exec(code, run_globals)
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\site-packages\pip_main.py", line 29, in <module>
from pip._internal.cli.main import main as _main
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\site-packages\pip_internal\cli\main.py", line 9, in <module>
from pip._internal.cli.autocompletion import autocomplete
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\site-packages\pip_internal\cli\autocompletion.py", line 10, in <module>
from pip._internal.cli.main_parser import create_main_parser
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\site-packages\pip_internal\cli\main_parser.py", line 9, in <module>
from pip._internal.build_env import get_runnable_pip
File "C:\Users\LEGION\AppData\Local\Programs\Python\Python310\lib\site-packages\pip_internal\build_env.py", line 23, in <module>
from pip._internal.utils.temp_dir import TempDirectory, tempdir_kinds
ModuleNotFoundError: No module named 'pip._internal.utils.temp_dir'

Launch unsuccessful. Exiting.
Press any key to continue . . .

[graceful folio]

Thank you for the recommendation

[fair oxide]

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

[lilac scaffold]

hey can someone help me i am getting this error " Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument index in method wrapper_CUDA__index_select)" i have intel 13900k i9 and rtx 4090. which means i got 2 gpus intel integrated one nd rtx 4090, how do i fix this error, i did a fresh install of sdxl on my pc just moments ago and launched the ui and selected the default safetensors file

[sand trail]

any idea why is this showing up

[lilac scaffold]

what is

[fickle pewter]

hi guys i need tech support

[sand trail]

im trying to run svd on comfyui through google colab but its not working

[fickle pewter]

i tried using another model to generate image and i felt it was ugly, i switched back to old model and it affect my generations

[fickle pewter]

even my old model generation has that ugly effect now, i tried restarting webui and deleteing old model still not work

[grim moon]

Does anyone know how to fix this?

I mean i have 16GB of vram, is that not enough?

[fair oxide]

I did this, the error earlier is now fixed but now this showed up

[lilac scaffold]

add skip-torch-cuda-test to web_user.bat command_line

[lilac scaffold]

use medvram

[lilac scaffold]

ya man 16 gb is not enough,

[grim moon]

whaaaat

Damn

Ok i guess

[lilac scaffold]

i am running 24gb rtx 4090 and idk how much it uses honestly.

its very intensive

just go with --medvram

[grim moon]

ok

i put it here?

@echo off

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

call webui.bat

[lilac scaffold]

@grim moon i generated this for you though

[grim moon]

(Sorry im kinda new to this)

[lilac scaffold]

your prompt, this is from the default model

[grim moon]

Nah, i was just testing

Not really the thing im aiming for

[lilac scaffold]

commandline_args= --medvram

[grim moon]

Thanks

[lilac scaffold]

i think its use medvram or what, i don't know, i didn't use it, better read about stable diffusion medvram

[grim moon]

And do you know how what resoloution I could generate images?

[lilac scaffold]

no, i did instal sdxl like 30 mins ago, i am very new

[grim moon]

Like, should i use upscaling or just go 1024 by 1024

Ok

Thx tho

Yeah, It works now thanks

Nevermind.

[lilac scaffold]

can you add --xformers if you have nvidia gpu

[grim moon]

And if i do not?

[lilac scaffold]

in commandline_args = --xformers and run the webui.batch file

--xformers significantly optimizes sdxl

[grim moon]

(I have a RX 6900 XT)

[lilac scaffold]

hmm, i guess it won't work then

[sand trail]

Any pros over here are free to help setup my comfyui on Google colab?

[narrow fjord]

Hello! After getting SD to work, I tried testing out other models.

However, I am getting this error and I do not know why.

RTX 3070 8GB model is what I have.

Using RVC, there was an option to reduce the batch size.

But I can't seem to find it in the settings option.

What is it called?

Ok, apparently I am blind. I found the Batch Size bar 😄

But working with this one still gives me Cuda out of memory.

[ornate elk]

Hey, you need to edit the webui-user.bat at the line COMMANDLINE_ARGS=
You need to add: --xformers --medvram-sdxl --no-half-vae

That will save a lot of vram and gives better performance

[narrow fjord]

I am on Arch Linux so .bat files do not work here.

I need to use .sh

[static moat]

Does anyone know how projects like this https://photoai.com/ train the images of the person then to generate photos?
I'm kinda new to diffusion and loras, and would love to learn more about it regarding face training etc.

[narrow fjord]

I found it in the .sh. I'll give it a go.

Ok, that fixed it. Thanks 🙂

[grim moon]

Btw are you sure it is nvidia exclusive?

[sand trail]

how to solve this

[lilac scaffold]

YES SIR

[obtuse yacht]

Hey there. I probably just missed something--an update or the like--but is there a way I can stop SD from generating "animated" images? This is happening seemingly at random with no prompting on my part. One prompt I get still images, another prompt I get animation. Thanks.

[ornate elk]

With the Discord bot?
Use the format: tag

[ornate elk]

Can you show an example?

[ornate elk]

What's your GPU ?

[ornate elk]

With that GPU you need
--medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling

Then you can generate images like 512x768 upscaled by 2 or 540x960 upscaled by 2 with highres fix.
Also you should install the tiled diffusion extension and there only enable tiled VAE option
That can prevent running out of vram

[static raft]

Hey people, recently reinstalling A1111 after having to wipe my computer because of malware, can anyone tell me the best Command_args for my 6700 XT? Can't seem to find anything definitive online

I also remember having to change something because i was on an AMD card

But don't remember what it was

[vale prism]

Trying to install stable diffusion

Took Torch 2.01
Took proper version of Python

I have an AMD GPU (which apparently is a problem)

If anyone could help 🙏

[rose rune]

I'm currently running Fooocus. I originally installed it on an HDD, but deleted that install after finding that the HDD bottlenecked generation. I deleted everything and reinstalled it onto my SDD, but for some reason every time I get to "Moving model to GPU" stage, my HDD ramps up to 100% and it takes several minutes for the model to start generating. Any ideas about why this is happening and how I can address this bottleneck?

[grim moon]

Could you tell me Please:

• How to upscale and
• How to enable tiled VAE, i already installed tiled difusion, but i cant find it

[ornate elk]

hey, set a default resolution to 512x768
for upscaling you need to enable hires fix.
then select:
Hires steps: 10
Denois: 0.5
Upscale by: 2
Upscaler: REsrgan4x+

thats just an example.
Tiled vae needs to be enabled when the default resolution is to high like for 540x960
you can set it to this:

[ornate elk]

thats mostly because you have an active windows pagefile set on your hdd.

[grim moon]

And how do i know what to put in the upscaling?

[rose rune]

you're right, just turned that off, let's see if this works.

[ornate elk]

you should only have it active for your C drive

best would be on "system managed"

[rose rune]

Done, restarting to see if it's working

[ornate elk]

you need to restart the pc after changing pagefiles

[rose rune]

Maybe now it won't take 10 minutes to move to GPU

[ornate elk]

for hires steps stay between 10 - 15
denois 0.5 or less

[grim moon]

Ok, and how do I know what upscaler does what?

[ornate elk]

testing ^^

[grim moon]

Ok thanks

Andd. what if my pc crashed while generating an image?

[ornate elk]

Latent ones can add stuff to the images (details) Esrgan based upscalers are more commonly used to smooth an image or get sharper resolution

[ornate elk]

then your psu could be to weak for your gpu.
but it could be also something different

[rose rune]

Moved from,10 minutes to 54 seconds. That's acceptable.

[ornate elk]

sounds good

[agile vector]

I'm almost completely new to all this stuff, but I got everything working locally, tried a few different models that I found, and I haven't really been able to get anything of quality. I was trying various prompts that I found, is there something I'm doing totally wrong here or is it just really hit or miss?
https://itsarat.world/1a948946.png

[ornate elk]

hey, your using the default model. while its good for testing its not good in quality. Checkout Civitai.com for the Community made models.
For example get Dreamshaper v8, its a good starting model that can generate nice images.
The next thing is to change the resolution. for example for a portrait try 512x768

[agile vector]

Figured it was something like that. tyvm

[grim moon]

For Clearly (and only reserved for) educational purposes, is there a way to turn off the ||NSFW|| filter?

[ornate elk]

there is no nsfw filter in the local webui.
also the models are more or less capable to do that

[grim moon]

Thank you very much good sir

[ornate elk]

no problem

[grim moon]

Do you know any browser for stable diffusion models?

[ornate elk]

Civitai.com is the largest model database 🙂

[static raft]

Hey people, installed A1111 and it launches, but now i am getting "RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'

[ornate elk]

hey, whats your gpu?

[static raft]

6700 XT

[ornate elk]

did you installed the directml version on windows or the default one on linux?

[static raft]

On windows

[ornate elk]

okay, whats inside your webui-user.bat?

[static raft]

@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS= --autolaunch --skip-torch-cuda-test --medvram --no-half-vae
git pull
call webui.bat

[ornate elk]

is your folder called
satbel-diffusion-webui or stable-diffusion-webui-directml ?

[static raft]

The folder leading to stable-diffusion-webui?

[ornate elk]

no the folder itself

[static raft]

I have it called A1111

[grim moon]

Really?

[static raft]

Then inside of it i have stable-diffusion-webui

[ornate elk]

thats the wrong webui, AMD cards need a different one

you need this:
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs

[grim moon]

Is this just false or what? Bc I have seen like 5+ articles while looking for models?

[static raft]

Shitty, alright thanks for the help

[ornate elk]

nope, there isnt an invisible watermark to my knowledge. Also the nfsw filter only applies to website services like Dreamstudio

[grim moon]

Oh, okay

[ornate elk]

if your local you wont have anything to worry

[grim moon]

Im just beeing sceptical cuz i saw like 5+ articles telling you have to put sth in the code

f.e https://pinter.org/archives/14569

[ornate elk]

wow xD they dont even explain in which local webui (there are a lot) xD

but thats just false

nothing gets flaged or censored localy

[grim moon]

Okay since i read ur bio i believe you entierly so thx

[ornate elk]

no problem 🙂

[ornate elk]

no problem, feel free to ask if you have any install problems

[static raft]

Ok the install was successful and it seems to run fine, but pretty slow. Anything i can do to speed up gens?

[ornate elk]

yes, with that GPU you need
--medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling
In your webui-user.bat
Then you can generate images like 512x768 upscaled by 2 or 540x960 upscaled by 2 with highres fix.
Also you should install the tiled diffusion extension and there only enable tiled VAE option
That can prevent running out of vram

[static raft]

Got it, i'll try that out thank you

[static raft]

Out of curiosity, what does tiled diffusion do?

[ornate elk]

tiled diffusion is fore tiling an image into squares to lower the amount of vram used for a direct full resolution.
Tiled vae does that only for the last step (the vae apply)

[static raft]

Got it

[static raft]

So i download it as a zip file and upload it into the extensions folder on my computer?

[ornate elk]

no you can go into the extensions tab in the webui and then click on available, then on load from, then search for tiled

then install and restart

[static raft]

Ah that makes more sense

Got it, it worked and it's installed. Thanks again for the assistance

[ornate elk]

nice np 🙂

[ornate elk]

you need to use an other upscaler like Latent bicubic or Resrgan4x+

[hollow schooner]

Oh okay thankssss

[civic shoal]

Hi!, I'm training a own model using autotrain dreambooth collab and I have a doubt about what a refiner does in the code after the training model

[karmic crown]

https://github.com/CompVis/stable-diffusion/blob/21f890f9da3cfbeaba8e2ac3c425ee9e998d5229/scripts/txt2img.py#L88

a1111 has that code in this extension https://github.com/AUTOMATIC1111/stable-diffusion-webui-nsfw-censor

[fair oxide]

Oh, I dont have one. Is a GPU really needed, or can i use the CPU instead?

[karmic crown]

what are your machine specs?

it is possible, but it would be painfully slow

#🤝｜tech-support message

my VPS has 10 cores and 60 GB of RAM, with fast SSD

about 3 min for a 512x512 image with 20 steps, but only because i have 10 cores and a lot of RAM

[fair oxide]

Uhh, which web ui file?

Oh nevermind I got it

[humble moon]

Hey guys, Im watching this video where he teachs how to get highier quality results but he says I need to pick this processor type on controlnet which mine (latest version) doesnt have

]

Is it that the tile_gaussian is now tile_resample?

[fast mirage]

Hi! I downloaded SD from github and i generated a few images, but it was painfully slow

I have a RX 6600 gpu, is that why is it taking too long?

Like 3~5 minutes
I see People on the internet generating images crazy fast

[devout dirge]

Hi,Why can't I find all the pictures I generated yesterday?

[devout dirge]

为什么我昨天及以前生成的所有内容，在收件箱里都没有了

[formal trout]

hey @ornate elk I'm currently trying "best of both worlds", generating in 832x1216 and hiresfix x1.23, in Latent (now I'm trying other latents, but no much idea)

So I generated in "full" resolution and resize in Latent to the max

@ornate elk I'm seeying some artifacts here and there, that's why I'm trying other Hiresfix methods (i.e., in the recycling sign)

[grim moon]

Yoo, I have a technical question: What does it mean for ai image generator to be "trained"? Is stable diffusion "trained"?

[formal trout]

all AI models are trained in the sense that they learn by being fed with data that they work out with. So certain models are more trained in photo-realistic images, others on anime, etc.

[grim moon]

So bigger model=better trained=more accieate image

[formal trout]

yes I guess most of the time

It depends also if the model is well trained or not

[hardy thorn]

I found the solution, instead of install python with recommend installation, go to custom install then choose "install for all users"

[ornate elk]

Bigger models contain more information. But the community found a way to reduce the size by removing not necessary stuff out of the models.
These models are "pruned" and I recommend using those over their unpruned versions. There is no big difference

[ornate elk]

Ahh nice! Thanks for the update

[ornate elk]

Hey, the images look nice. Did you also tried Latent Nearest Exact?

[sharp lichen]

Hellko, how much faster would a 4090 be than a 3080, in auto

For SDXL

Also how much faster is comyfy for sdxl than in auto?

[solemn moon]

heloo friends, looking for an Auto masking tool o0 {asking for a friend}

[ornate elk]

Hey, idk if it helps but take a look at this:
https://github.com/continue-revolution/sd-webui-segment-anything

[solemn moon]

Thanks @ornate elk

[potent wraith]

@ornate elk

(yes i cut out part of the controlnet image)

[ornate elk]

Okay try again after adding the commands to the webui-user.bat

[potent wraith]

works!

[ornate elk]

Controlnet requires xformers for better performance and less vram usage

[unique anchor]

I get this error when trying to use KohyaSS to train my LoRA model, how to fix it?

https://hastebin.com/share/egikanuquk.swift

I have bitsandbytes installed globally.

[merry whale]

any guides on running the ai on archlinux with an AMD GPU?

[ornate elk]

Here:
https://github.com/AUTOMATIC1111/stable-diffusion-webui#automatic-installation-on-linux

Make sure that you have the Rocm drivers installed

[merry whale]

I tried that
clicked on generate and nothing happened
also some buttons don't work like the interrupt one and the random generator on the seed thing
I'll try again when I have time

[ornate elk]

Make sure to check the terminal for any issues.
Also whitelist the webui in any browser adblocker

[merry whale]

well when I click some buttons it prints some errors
I'll see tonight

[long pumice]

I'm trying to download from this: https://github.com/apple/ml-stable-diffusion#readme and speed up SD on mac but have no idea how to do it

does anyone know how to do it??

[sly plinth]

Hello guys, when using the SDXL api to generate an image-to-image, my code is catching a Non-200 response error which looks like a Cloudflare problem, anyone else is facing something like this?

[karmic crown]

you need to build it with Xcode, they do not provide binaries

i am not sure how much benefit will you actually get with it to be honest

what are you using now and how fast (slow) generation is?

[long pumice]

honestly im not sure what im doing but as of right now im wiling to learn how to do it

im using m1 pro 16gb

[karmic crown]

you can follow my guide for A1111 installation on Macs i published a few days ago https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/5461#discussioncomment-7738578

[long pumice]

generation is pretty slow, roughly 800x800 takes 10~15 minutes with 40ish steps

oki

i think thats what i did

one thing im confused about

[karmic crown]

show me your runtime params, please

[fair oxide]

why is the local url not working?

[long pumice]

let me figure out how to get that and ill come back

[karmic crown]

open your webui-user.sh

copy and paste the complete uncommented line with COMMANDLINE_ARGS

also copy and paste the line with TORCH_COMMAND

[long pumice]

wait actualy

i think i only did "new install"

can you tell me how i follow this step

4. Remove venv folder with rm -rf venv or using Finder

[karmic crown]

open terminal

then cd stable-diffusion-webui

and then do the rest

[karmic crown]

but wait

yes that

[long pumice]

and than ./webui.sh?

ah i see

[karmic crown]

Existing install:

this should be done first

but ok

[long pumice]

its installing bunch of stuff

[karmic crown]

its fine

let it finish

[long pumice]

ah i see

[karmic crown]

and then do this
git pull

[long pumice]

should have followed instructions

[karmic crown]

that will update the a1111

[long pumice]

okay so i got the error ModuleNotFoundError: No module named 'torchvision.transforms.functional_tensor'

i do the git pull before "sed -i '' 's/from torchvision.transforms.functional_tensor import rgb_to_grayscale/from torchvision.transforms.functional import rgb_to_grayscale/' venv/lib/python3.10/site-packages/basicsr/data/degradations.py"?

[karmic crown]

remote: Enumerating objects: 510, done.
remote: Counting objects: 100% (490/490), done.
remote: Compressing objects: 100% (244/244), done.
remote: Total 510 (delta 303), reused 402 (delta 242), pack-reused 20
Receiving objects: 100% (510/510), 307.33 KiB | 2.52 MiB/s, done.
Resolving deltas: 100% (308/308), completed with 14 local objects.
From https://github.com/AUTOMATIC1111/stable-diffusion-webui
   8aa51f68..883d6a2b  dev               -> origin/dev
 * [new branch]        gradio-4.x-test   -> origin/gradio-4.x-test
 * [new branch]        gradio4           -> origin/gradio4
   5ef669de..81105ee0  release_candidate -> origin/release_candidate
   40ac134c..f5f89780  test-fp8          -> origin/test-fp8
 * [new branch]        torch210          -> origin/torch210
 * [new tag]           v1.6.1            -> v1.6.1
 * [new tag]           v1.7.0-RC         -> v1.7.0-RC
Already up to date.```

like this

please slow

[long pumice]

done

oki

[karmic crown]

venv/bin/pip cache purge

this will remove all cached python stuff

[long pumice]

oki

[karmic crown]

it should saye something like ___ files purged

for example 425 files purged

[long pumice]

yup i see file removed: 1147

[karmic crown]

great

now remove venv again

rm -rf venv

[long pumice]

got it

[karmic crown]

does your webui-user looks like the one in tutorial?

[long pumice]

ye i did that part

[karmic crown]

export TORCH_COMMAND="pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu"

this is to download development build of torch

some things doesnt work with public version

[karmic crown]

do steps 5. to the end

you do not need to run that command

it has to be in your webui-user.sh

[long pumice]

oh i see ye i have that

wait so what exactly are these steps achieving

the common steps part

[karmic crown]

line 6 changes the file venv/lib/python3.10/site-packages/basicsr/data/degradations.py

so you do not have to it manually

that is a fix for this error ModuleNotFoundError: No module named 'torchvision.transforms.functional_tensor'

[long pumice]

okay

so when i run that fix it doesnt start loading or anything

[karmic crown]

line 7 changes modules/sd_disable_initialization.py and fixes this error TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. Please use float32 instead.

it just do a quick search and replice

[long pumice]

okii

[karmic crown]

takes the same time as when you replace on string in any text editor

it literally do that

[long pumice]

ohh got it

[karmic crown]

torchvision.transforms.functional_tensor -> torchvision.transforms.functional (step 6)

[long pumice]

ahh i see your note on the instructions

okay so ./webui.sh now opens the webui

[karmic crown]

generate something and show me the speed

somewhere in webui find this

version: v1.6.1  •  python: 3.10.13  •  torch: 2.2.0.dev20231125  •  xformers: N/A  •  gradio: 3.41.2

and check if it is the same

your torch will have todays date

[long pumice]

version: v1.6.1  •  python: 3.10.13  •  torch: 2.2.0.dev20231204  •  xformers: N/A  •  gradio: 3.41.2  •  checkpoint: cc6cb27103

[karmic crown]

what do you mean? which local url? for webui?

generate something and show me the speed

[long pumice]

thanks for the help

i would have never done that by myself lol

[karmic crown]

generate 512x512 with 1.5 model or 768x768 with SDXL

[long pumice]

so what is https://github.com/apple/ml-stable-diffusion#readme for??

[karmic crown]

as far as i understand, it converts stable diffusion to apple's ml

btw. you are using metal now

[long pumice]

i see

what is metal

[karmic crown]

something like apples opengl or directx

[long pumice]

does that mean this is running like locally on my computer

[karmic crown]

yes

[long pumice]

ahh i see

[karmic crown]

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 817807654, Size: 512x512, Model hash: ec41bd2a82, Model: photon_v1, VAE hash: 735e4c3a44, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Version: v1.6.1

Time taken: 1 min. 19.4 sec.

Intel iMac with Radeon

Time taken: 44.5 sec.

M1 MacBook Pro

[fair oxide]

the local url it gives is not working

[karmic crown]

you have added --listen

do you want to access that from anoter computer?

[fair oxide]

no?

[karmic crown]

i have that on one of my computers, so i can access that one from another

[fair oxide]

i just have to remove --listen?

[karmic crown]

if you do not need that, just remove --listen

[sly plinth]

anyone knows how to fix this please? im working on a project that requires this

[karmic crown]

even with --listen (or without it) your webui can always be accessed by http://127.0.0.1:7860/ from a same computer

[karmic crown]

520 maybe?

[sly plinth]

520?

[karmic crown]

http error 524, sorry

[sly plinth]

Oh no, I get a cloudflare error from "host", the following:

Host
Error
What happened?
There is an unknown connection issue between Cloudflare and the origin web server. As a result, the web page can not be displayed.

What can I do?
If you are a visitor of this website:
Please try again in a few minutes.

and yes ur right its 520

[karmic crown]

520 is unknow, 524 is timeout

520 is probably worse

[sly plinth]

Web server is returning an unknown error Error code 520

[karmic crown]

what is your setup?

[sly plinth]

django python, i send a request same as the one they show in their documentation

[karmic crown]

is Django on a public url (behind CF)?

[sly plinth]

no no, im still doing it locally, but the normal generation works fine, only issue is in image_to_image

[karmic crown]

wait, you are using api that is not from your server?

i thoug you are running a1111 on your server at first

than that you have your app on your server

[sly plinth]

no no, im using the public API from stability https://platform.stability.ai/docs/api-reference#tag/v1generation/operation/textToImage

the text_to_image is working great, but the image_to_image is giving me the 520 error

[karmic crown]

image size?

try the code from example

[sly plinth]

1024x1024, same as recommended, i did exact code from example, and thats when i got hit with the cloudflare error

im not sure if theres a problem from me, or stability and cloudflare not reaching each other

[karmic crown]

maybe it is some temporary error, try again later

[sly plinth]

alright will do, thanks mate!

[fair oxide]

it works now, thank you

[crystal gate]

New to comfyui.
i just installed comfyui and i get this in terminal-

got prompt
model_type EPS
adm 2816
Using pytorch attention in VAE
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
Using pytorch attention in VAE
missing {'cond_stage_model.clip_l.text_projection', 'cond_stage_model.clip_l.logit_scale'}
left over keys: dict_keys(['cond_stage_model.clip_l.transformer.text_model.embeddings.position_ids'])
Requested to load SDXLClipModel
Loading 1 new model
Requested to load SDXL
Loading 1 new model
loading in lowvram mode 3126.826922416687
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:26<00:00, 1.33s/it]
Requested to load AutoencoderKL
Loading 1 new model
You shouldn't move a model when it is dispatched on multiple devices.
Prompt executed in 34.37 seconds

What does it mean by "You shouldn't move a model when it is dispatched on multiple devices" and is it normal 'the other things which comeup in terminal when i click queue prompt it compared to auto1111 which has only one line in terminal'.

[formal wadi]

Good evening, if i may, how do i ensure that there is a safety filter on to prevent NSFW generation?

[ornate elk]

Hey, Where?
In a local webui or here in the bot channels?

[formal wadi]

Local webui please

all i can find online is steps to disable the NSFW filter, i kind of want to enhance it lol

[ornate elk]

Then you need to add tags like nudity, nude, gore, blood, nsfw
Into the negative prompts box.
Also you can add an extension called nsfw filter.
You can find it under the extensions tab, click on available, then on load from. And search for nsfw, install and restart.
It will make nsfw images Black

[fair oxide]

Guys, which one ROCm I should use for rx580?

I'm reffering to docker images

[formal wadi]

amazing, thank you very much

[pale plover]

Hello! Sorry to bother you. I've had a prompt pending for several minutes and it won't load. I keep getting the error message 'you're still dreaming!'

[ornate elk]

Hey, can you try if an other bot channel works for you?

[neat bay]

hey guys did anyone experience this problem with comfyui custom node face detailer pipe

[sturdy cedar]

Does anyone know what this means, and how I can fix it?

[lapis raven]

Are the boot options below correct when generating CPU-only on a PC with AMD's Ryzen and Radeon?
--use-cpu all --no-half --no-half-vae --opt-sub-quad-attention --skip-torch-cuda-test
Also, is there a way to change the amount of main memory allocated at that time?

[karmic crown]

I am a little confused. why would you use cpu only if you have Radeon? It will be painfully slow on that machine

[lapis raven]

I only have an 8G GPU, so I wanted to see how fast it would be if I generated it on a CPU.
Main memory is 64G

[ornate elk]

Even AMD Vega 8 onboard GPU can use directml

[karmic crown]

#🤝｜tech-support message

i have SD on my VPS and it is sooo slow

60GB or RAM and 10 cores

[lapis raven]

Does this mean I should use this option?

[karmic crown]

if you want to try cpu only, you can use what i have there

[lapis raven]

If I had to be greedy, I would be grateful if I could use both GPU and main memory.

[lapis raven]

thank you

[ornate elk]

You can only use one at a time

I would suggest you to use the 8gb vram

[mental elbow]

Hey, I followed instructions from stable difussion github page for installation on my ubuntu pc, i had less storage so the installation did not complete, how do i undo the complete installation?

[fair oxide]

Hi, can you help me with something? When I try to use any AbyssOrange2 model my SD crashes

[ornate elk]

Open a terminal and run
Pip cache purge
Then delete the stable-diffusion-webui folder

[mental elbow]

pip cache purge removes everything from pip cache folder right, cant i remove only the ones that got dowloaded while installing stable diffusion? I know its all cached files but just checking if its a possibility,

[alpine yoke]

When i generate images I just see noise. This has happened twice. The first time I deleted my stable diffusion folder and started over and it was working fine again. Now it's happening again and i don't know what is causing it. Any advice?

[karmic crown]

just purge everything, it will not take long if you have a decent internet speed

[karmic crown]

copy paste info about sampler, steps, cfg and other settings used

[fair oxide]

Can you help me with my problem? ;~;

[alpine yoke]

Hmm... I had closed it out and so I just reopened it to try generating an image so i could copy my settings and now it's working again. I don't know what the problem was because I am using more or less the same settings nearly every time. I guess it's a relief that I don't need to delete everything and start over again to fix it.

[fair oxide]

I tried to use the Hassaku model, but halfway through the generation it crashes

Get to this point and then crash

[sturdy cedar]

Bump. Could use some help here. Not sure what exactly is mismatched or what I’m doing wrong, but I can’t seem to get IPAdapter working correctly 🤔

Error occurred when executing IPAdapterApply:

Error(s) in loading state_dict for ImageProjModel:
size mismatch for proj.weight: copying a param with shape torch.Size([3072, 1024]) from checkpoint, the shape in current model is torch.Size([3072, 1280]).

[karmic crown]

what is your image size?

[ornate elk]

What's your GPU and what's inside your webui-user.bat?

[fair oxide]

I use a google colab

[azure magnet]

Can someone help me with controlnet extension?

[ornate elk]

You should describe what you trying to do

[ornate elk]

Oh, can't help with collab sry

[azure magnet]

nothing i'm using controlnet model and i got this warning "Warning Unable to determinate version for controlnet model controlnet11models_openpose"

[azure magnet]

And a last question where i can find the last version of controlnet models?

controlnet work well but i maybe there is something to fix

[ornate elk]

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

[ornate elk]

Thats an error from using xformers

[azure magnet]

ok, this is where download it

[fair oxide]

Ohhhh thank you very much

[azure magnet]

And about the warning?

[quartz sinew]

is the bot borked ?

[eternal elbow]

#1047610792226340935

[tough merlin]

Do you guys know how to upload photos on here and then use that to generate new versions?

[ornate elk]

Check that the model has the same name as the .yaml

[azure magnet]

Yes

So why SD don't read the version of model? Maybe i need to add to the filename?

[ornate elk]

Hmm nothing. Is Our controlnet updated?

[azure magnet]

Check the update now controlnet is update to the last version 1.1.419

[wet hull]

Hey! I'm trying to run Stable Diffusion Locally but I'm getting this error.

[ornate elk]

hey, you installed the webui the wrong way:
use my guide instead:

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.

[wet hull]

It still fails

[ornate elk]

can you show me the new error?

[wet hull]

TypeError: AsyncConnectionPool.__init__() got an unexpected keyword argument 'socket_options'

[ornate elk]

the whole cmd log pls

[ornate elk]

wow you didnt even followed my guide

wrong python version, and wrong installation method

[wet hull]

Sorry. Currently installing Python Version 3.10

Going to do a fresh install instead of adding a new one to the old one

Okay it worked I think I'm in the website

[ornate elk]

nice, whats your gpu btw?

[wet hull]

I'm using AMD, not NIVIDIA

I assume I can only use this on NIVIDIA?

[ornate elk]

nope, but for AMD you need a different webui xD

[lapis raven]

It is also possible to generate it on AMD.

[wet hull]

o

Where can I find that?

[ornate elk]

ill edit my guide

[ornate elk]

hey, you installed the webui the wrong way:
use my guide instead:

Here is a quick Guide to install Automatic1111 Directml AMD 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/lshqqytiger/stable-diffusion-webui-directml && cd stable-diffusion-webui-directml && git submodule init && git submodule update
4. Press enter and after its done you can close the cmd
5. Edit the webui-user.bat (right click), At the line COMMANDLINE_ARGS= You add: --medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling
   After that save 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.

In fact Step 3 and 5 are now changed compared to the nvidia guide

[lapis raven]

Is there a setting to not save the prompt after each generation?

[ornate elk]

yes, in the settings you can disable the automatic save option

[lapis raven]

I will try to find it

[wet hull]

Command's not working, says the '&&' symbols aren't valid statement seperators

[ornate elk]

thats strange, can you show a screenshot of the cmd?

[wet hull]

I'm using powershell, do I have to use cmd?

[ornate elk]

thats not a cmd, thats a powershell window

[wet hull]

Ok yeah that was the issue

[ornate elk]

oh before i forget

run
pip cache purge
as first command

before launching the webui-user.bat

that will delete the nvidia files that got downloaded from the first try

[wet hull]

Where do I put that? In the bat file or cmd?

[ornate elk]

in cmd

[lapis raven]

I didn't know where it was...
Also, what should I do if I get an insufficient memory error during generation with Radeon?

[ornate elk]

whats your gpu?

you need this:
Edit the webui-user.bat (right click), At the line COMMANDLINE_ARGS= You add: --medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling

[lapis raven]

RX57000XT

[wet hull]

I'm getting another error

[ornate elk]

enough free space on C ?

[wet hull]

Is 143 GB enough for you?

[ornate elk]

okay then delete the venv folder.
open up a cmd and run
pip cache purge
again
Also make sure to have a good internet connection
And you need 8gb Ram or more

[wet hull]

I tried it again, didn't work. I deleted the venv folder and purged the cache.

[ornate elk]

do you have a full screenshot?

would like to see where it fails

[ornate elk]

it still uses the wrong python version

[wet hull]

I did install Python 3.10 btw, idk where it was installed at tho

Python 3.11 is still on my PC. Do I need to delete it completely?

[ornate elk]

yep uninstall it

then do the venv delete and the pip cache purge again

[wet hull]

It seems to be going smoothly so far

[ornate elk]

you may need --no-half too if you get any errors

[wet hull]

It worked! I made my first image prompt to test it out. What do you think?

[lapis raven]

When I put it in, it was successful, but the second time I got an error.

[ornate elk]

haha perfect 😄

[ornate elk]

play around with the settings (steps,cfg, resolution), then checkout Civitai.com for better Models/Checkpoints
For example Dreamshaper v8 is a good allrounder model

[wet hull]

I'd like to make my own models sometime

Oh snap I ran into an error

My GPU only has 4GB of ram. I assume the program is able to support that amount?

[ornate elk]

oh if you have only 4gb than you need to change the --medvram to --lowvram in the webui-user.bat

[wet hull]

kk

[sand trail]

I'm having troubles figuring out why SVD model isnot loading into my ui, im using google colab

[sturdy cedar]

Does anyone know how I can resolve this error? I'm attempting to use IPAdapter.

[formal trout]

No, I will. With certain styles like realistic I found it a little too grainy, at least with Latent, anti-aliased and upscaler. It is pretty noticeble when using adeailer on face. For illustration works pretty nice.

[lapis raven]

After all, when I try to generate it from the second time onwards, I can't do it due to a memory error.
The usage rate is displayed as 100% in the UI.

--medvram --opt-sub-quad-attention --opt-split-attention-v1 --upcast-sampling --no-half --disable-nan-check --precision full

This is the option I'm using

[ornate elk]

Don't use --disable nan check
Also remove --precision full

[ornate elk]

Yea sometimes you can see a differences with adetailer.
Also you may need to use a different VAE if its to grainy

[lapis raven]

It didn't work

[ornate elk]

You may need to use --lowvram instead of --medvram

Because no-half needs more vram

I'm sure that will work

[lapis raven]

It's done!

thank you!

[ornate elk]

No problem 🙂

[ornate elk]

Tip for the future:
For upscaling images you need to use the SD upscale script in img2img.
Because hires fix in txt2img would get you out of memory

[lapis raven]

understood.

[elder hawk]

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

[fast mirage]

Could anyone help me on how to generate images faster? I have a AMD GPU and it takes 5min for a bad quality img to appear 💀

[azure magnet]

Hi, i'd installed Adetailer but i don't have deepfashion, how can i fix it?

[formal trout]

hey I tried Nearest exact, it's perfect, thank you. No grain, almost same texture as adetailer.

[shy crest]

Anyone else suddenly having issues with Automatic1111 shutting it's UI down and giving the "Out of Memory" error in Chrome?

Everything has been fine for weeks. Didn't do any changes, didn't add any extensions etc. It just suddenly started doing this

And I've got --no-half-vae --medvram in my bat file

And my GPU is MSI GeForce RTX 4070 Ti - so it should be able to handle opening up the UI without even doing anything

So basically Automatic1111 is shutting itself down before I even try to generate anything. About 10 seconds after launching it

I tried the trick of reinstalling the /venv/ folder. (which usually helps in these cases) but now it didn't help

Also tried going back to a previous version of Automatic1111 but that didn't help either

Now I'm cursing myself for having "git pull" in my bat file. I should have removed it so it doesn't automatically update. groans

Ok tried disabling all "extra" extensions and that seems to have done the trick. So I wonder if the problem was that a new Automatic1111 version automatically updated and perhaps one of the extensions "broke things" if it hasn't had time to get an update for the new version? Must be it

[vocal lichen]

How come all of my prompts fail? It says there's content blocked by the server, but even one-worded prompts don't work. Am I banned for some reason?

[ornate elk]

Yep, you need to update your extension in the extension tab.
You can then enable them one by one and check which one causes the error

[ornate elk]

What's your GPU?

[wind grove]

So, i have questions about working with sdxl models in a1111, i dont know if it is offensive to ask them here? -> in any case if people know where i can go and ask questions about developping services for a1111 running in --api mode, let me know

-> can anyone link a json payload they send to a1111 for image generation using sdxl w. a controlnet where that payload generates an img similar to what they get when using the UI?

[shy crest]

Thank you CS, will do!

[ornate elk]

Hey, I think your question would better fit in #1042896447311454361 or #1011228667659178055 🙂

[wind grove]

i will move it -> should i delete or let it stay up or create thread here? ty for help

[ornate elk]

You can let it stay. Maybe someone reads and reply to it here who knows 🙂

[fair oxide]

@ornate elk i have an issue with animatediff which im reposting here..

i got to install the animatediff extension and donwloaded the mm_sdxl_v10_beta.ckpt into the animatediff model folder, but when i run the prompt i get the following message.. altho i see it generated 16 images but no animation.

*** Error verifying pickled file from C:\StableDiffusion\stable-diffusion-webui\extensions\sd-webui-animatediff\model\mm_sdxl_v10_beta.ckpt
*** The file may be malicious, so the program is not going to read it.
*** You can skip this check with --disable-safe-unpickle commandline argument.

[karmic crown]

is there a safetensors version?

[fair oxide]

i dont see it

got that from here https://huggingface.co/guoyww/animatediff/tree/main

[signal sigil]

Hey guys.
Can anyone tell me the difference between using the Inpaint ControlNet and SD Inpainting models?
I'm not sure which method is best to use. Because for me at least, they seem to give different results.
I have been using Inpaint Controlnet with a none inpaint trained SD 1.5 model, up until now.
But i recently used a SD inpainting model without controlnet, and it worked very well but gave me different results.
(This is both for Img2img inpainting btw)
Just want to know if there is anything extra i should know?..

[karmic crown]

do not use --disable-safe-unpickle, unless you are 100% sure that file is safe

[fair oxide]

i didn't

[karmic crown]

https://huggingface.co/guoyww/animatediff/commit/3d375c164f01b43a79cb9dd1ac6537af5121ae31

[fair oxide]

oh sweet, let me download taht

thanks going to try the safetensor

how do i download that?

i dont see any clickable link there

[karmic crown]

https://huggingface.co/guoyww/animatediff/tree/3d375c164f01b43a79cb9dd1ac6537af5121ae31

now? 🙂

[fair oxide]

yes perfect thanks, and downloading

lol these github pages and their link kind of confuse me how they are arranged

[karmic crown]

you do not have developer spirt in you my friend 🙂

[fair oxide]

i know some web designing with basic css,xhtml 🙂 but not in the caliber of a full blown coder

[karmic crown]

xhtml? Which year is it now? 2003? 🤣

[fair oxide]

haha thats what i did

ok here goes the animatediff, lets see if this generates now

so far so good, no error except for this message AnimateDiff - WARNING - Missing keys <All keys matched successfully>

[karmic crown]

i made the first website in the last century (while the main browser was netscape 3) 🤣

[fair oxide]

and its working now

[fair oxide]

wow that goes back in time 🙂

[karmic crown]

recently i found my business card from 1999 (with official job position "webmaster") 🤣

[fair oxide]

lol that used to be cool i think its still cool, kind of like smart geek 🙂

[karmic crown]

webmaster = developer, sysadmin, devops... all in one

[fair oxide]

but yea imagine how things have changed now

ai is gonna do all the coding for us lol

[karmic crown]

and of course full stack

[karmic crown]

no, but it helps a lot to speed up

[fair oxide]

its getting there ;p

i wrote some fun javascript w/o knowing how javascript works when chatgpt came out lol

like animation

oh damn, i actually did it ...