🤝｜tech-support

[young bridge]

it didnt

[radiant stratus]

still

[ornate elk]

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

[young bridge]

@karmic crown it didnt fix it

[radiant stratus]

RX5700 and you gave me the link to the AMD specific branch

[karmic crown]

give me a second to try

[radiant stratus]

I'm on this page and my settings are

[ornate elk]

Right click the webui-user.bat and click edit

[radiant stratus]

I cant recall off the top of my head, where is the instal folder?

found it

c:/users/<not giving you my name>/stable-diffusion-webui-directml

[karmic crown]

how about git restore modules/textual_inversion/preprocess.py

[ornate elk]

The file is in that folder

[young bridge]

@karmic crown error: pathspec 'modules/textual_inversion/preprocess.py' did not match any file(s) known to git

[radiant stratus]

these are my options when I click the webui-user windows batch file

[ornate elk]

Show more options

Then edit

[ornate elk]

Okay, that explains your error

[karmic crown]

and if you delete that file and run git pull?

[young bridge]

i delete it, sure?

[karmic crown]

git pull
git status

[ornate elk]

@radiant stratus
At the line COMMANDLINE_ARGS=
you need to add: --medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling

Then save and relaunch the webui-user.bat

[young bridge]

do i delete it now?

[karmic crown]

if you didn't delete it, you can try with
git add .
git reset --hard origin/master

[radiant stratus]

this correct?

[young bridge]

git add

git reset --hard origin/master

?

[karmic crown]

git add . (do not forget the dot)

[ornate elk]

Correct, now you shouldnt run out of vram that fast

[young bridge]

On branch master
Your branch is up to date with 'origin/master'.

nothing to commit, working tree clean

🙂

@karmic crown

[radiant stratus]

is this no xformers note fine to have?

[ornate elk]

Check if your webui-user.bat is still edited with xformers

[young bridge]

no its gone

what is the command i forgot

[ornate elk]

Yes, Xformers is for nvidia only

[karmic crown]

yes, you need to redone that

[young bridge]

okay but what is the command

[karmic crown]

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

this one in webui-user.bat

[young bridge]

ok done

[quartz dock]

I have a really, really weird problem with openpose and SDXL in comfyUI. When I use other checkpoints they use the pose outline model control net fine.

When i use SDXL or anything based on it, it completely ignores the pose file and does a random pose.

Its working because it works on the other checkpoints but just not on SDXL. Any ideas?

[young bridge]

what is next?

can you tell me what was this error that i had?

[karmic crown]

run webui-user.bat

now you have uptodate a1111 (1.7.0)

and you can update anytime with git pull

[karmic crown]

somehow one file was untracked, which is totally unexpected, not it is fixed

[ornate elk]

Im off for now gn 😴

[karmic crown]

me too

1 pm

[young bridge]

okay guys thanks for the help

appreciated

[ornate elk]

No problem 🙂

[radiant stratus]

ok not so sure about this edit stable lol

[ornate elk]

Oh wait your on AMD

You need to add --no-half
To the webui-user.bat too
For inpainting to work.

[radiant stratus]

to the commandline args?

[ornate elk]

--no-half will increase the vram usage so remove it if you don't inpaint.

[uneven delta]

does using stable difusion consumes space on my disk?

[ornate elk]

After installing stable diffusion open up a cmd and type
Pip cache purge
That will give you 10gb back.
SD only takes space if you generate images with it or install extensions or download models etc.

[urban merlin]

hi i recently installed the openvino version of automatic1111 and it fails at generating on my arc a770 gpu with the following error which appears after spitting a ton of errors in the console window

BackendCompilerFailed: openvino_fx raised RuntimeError: ShapeProp error for: node=%self_norm1 : [#users=1] = call_module[target=self_norm1](args = (%input_tensor,), kwargs = {}) with meta={'nn_module_stack': {'self_norm1': <class 'torch.nn.modules.normalization.GroupNorm'>}, 'stack_trace': ' File "D:\\!AI\\Automatic1111-OpenVINO\\venv\\lib\\site-packages\\diffusers\\models\\resnet.py", line 691, in forward\n hidden_states = self.norm1(hidden_states)\n'} While executing %self_norm1 : [#users=1] = call_module[target=self_norm1](args = (%input_tensor,), kwargs = {}) Original traceback: File "D:\!AI\Automatic1111-OpenVINO\venv\lib\site-packages\diffusers\models\resnet.py", line 691, in forward hidden_states = self.norm1(hidden_states) Set torch._dynamo.config.verbose=True for more information You can suppress this exception and fall back to eager by setting: torch._dynamo.config.suppress_errors = True

strangely enough it will compile the model on my cpu and run there but on my gpu it decides to not work

idk if this is the right place to ask but i cant find information about this like anywhere

[radiant stratus]

hrm seems like it's doing worse on matching it to the area already there.

changing my prompt a little bit and decreasing the noise have helped keep it similar but

this just like looks the same as the original now

[verbal olive]

where are these 700mb options?

[tired copper]

Have you seen an automatic script that will run a prompt on a certain fine tune model, and then automatically switch the model, and then re-run the prompt… Like it runs through a list of fine tuned models?

[verbal olive]

https://huggingface.co/webui/ControlNet-modules-safetensors/tree/main

keep the old .yaml files in? @ornate elk

[wicked geode]

how would i be able to do a depth pass in batch?

[atomic ivy]

I accidentally dropped a Lora file in the models/Stable-diffusion folder and ran web-ui. It did a bunch of processing and had a progress bar totalling a large amount of space. Is there a way to remove whatever files it created? They won't be needed after I moved the safetensors file to the Lora folder.

[young bridge]

hey guys i have a question inside sd

if i generated a portrait photo of a person but would like to make it full body photo of that same person. how that can be done?

[heady stone]

UserWarning: MPS: 'nearest' mode upsampling is supported natively starting from macOS 13.0. Falling back on CPU. This may have performance implications.

[gentle arrow]

Does img2img sketch and inpainting just not work on Apple Silicon? Getting this error any time I try to do either. I've tried multiple models and I tried updating pytorch to 2.1

    NotImplementedError: convolution_overrideable not implemented. You are likely triggering this with tensor backend other than CPU/CUDA/MKLDNN, if this is intended, please use TORCH_LIBRARY_IMPL to override this function

[true sphinx]

hey guys im having a hard time getting SDXL to run with Vladmatic anyone willing to lend a hand

[kindred tinsel]

@karmic crown :)

[dark owl]

My memory cache isn't clearing after batches and is causing the next batch to get a memory outage error unless I re-open comfyui.bat. Does anyone know why? or how to fix it?

[hallow latch]

I tried a different model, same issue 🙂

[storm void]

Hi I am trying to use ControlNet with stable diffusion, but I get this error

OutOfMemoryError: CUDA out of memory. Tried to allocate 30.00 MiB (GPU 0; 6.00 GiB total capacity; 5.21 GiB already allocated; 0 bytes free; 5.31 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

Does anyone know a possible fix, thanks!

[karmic crown]

oh... I didn't know that. In that case, an upgrade is the most suitable way of solving the problem
is there any specific reason why you can not upgrade?

[karmic crown]

You tagged the wrong person. I haven't seen Intel GPU in years. I am using Macs and Linux servers 🤷‍♂️

[karmic crown]

whenever you get a memory related error, restart webui
you already know the implications of not doing that

[ornate elk]

You need to edit the webui-user.bat and at the line COMMANDLINE_ARGS=
You add: --xformers --medvram --no-half-vae
Then save and restart.

[karmic crown]

cu later, i need to work now

[storm void]

thanks so much for the suggestion, but sadly its still giving me the same error. Do you have any other possible solutions? Thanks again

[ornate elk]

Can you restart the webui-user.bat and make a screenshot of the cmd?

Also activate the lowvram option in ControlNet

And also use the 700mb controlnet models and not the 1.5gb ones

[reef smelt]

what the heck??

it's super slow...

but it says It's doing 32 iterations

or hmm

wait

oh...

lower is better?

[ornate elk]

It depends xD
Normaly you want
A high it/s (iteration per second)
And a low s/it (second per iteration)

[reef smelt]

I usually have... seconds per iteration...

woowies- actually more than one

3 is the highest I have gotten

[ornate elk]

What's your GPU again?

[reef smelt]

rtx 2060

[ornate elk]

It also depends on what your doing. upscaling for example takes longer

[ornate elk]

Okay and your using --xformers --medvram --no-half-vae ?

[reef smelt]

not --no-half-vae

actually I am

yes

[dark owl]

Batches matter too

[reef smelt]

batches?

It was only 1

[dark owl]

X1 should be fast

Yeah

[reef smelt]

I had 3 loras tho

I ran a benchmark using system info

[dark owl]

2060 is pretty old card though

4000 series has built in AI so the speed is insane

[brazen needle]

Hey guys, not sure if this is the right place to ask but i've been trying to setup Stable Diffusion on Google Colab for the last few days but just can't get it right. The closest i've gotten was the WebUI opening but unable to generate, now i always get some kind of error when even trying to start it

Does anyone know what the problem might be?

[reef smelt]

well money is not my strong suit

I'm gonna compare mine to another benchmark and see if I get the same

[karmic crown]

@ornate elk https://github.com/viking1304/a1111-setup/releases/tag/0.0.5 🙃

[reef smelt]

what's thta?

[ornate elk]

Niceee, thx for the mentioning ^^

Its an auto1111 nvidia installation script

[reef smelt]

is there an extension that auto refreshes things when you press on the selection tab? like for checkpoints, why make a refresh button if you could just do that with the selection box...

[karmic crown]

i will probably add linux support in the next version

[winged path]

Hello, I am trying to finetune Dreamboooth stable diffusion model from this google colab:
https://colab.research.google.com/github/sagiodev/stablediffusion_webui/blob/master/DreamBooth_Stable_Diffusion_SDA.ipynb#scrollTo=XU7NuMAA2drw
I am trying to finetune it by using 100 different images of real human faces with their textual descriptions saved as their filenames, and the filenames as their instance prompts. But during the "latent_caching" part of the forward process, it exceeds the system RAM limit of 12 gb after only 10%. Following is the modified part of the code i used to train the images :
Directory to store instance data
INSTANCE_DATA_DIR = "/content/drive/MyDrive/resized"

import os
List all filenames in the instance data directory
all_filenames = os.listdir(INSTANCE_DATA_DIR)

Create a concepts list based on filenames
concepts_list = [
{
"instance_prompt": filename,
"class_prompt": class_prompt,
"instance_data_dir": INSTANCE_DATA_DIR,
"class_data_dir": "/content/data/class"
} for filename in all_filenames
]

class_data_dir contains regularization images
import json
import os
for c in concepts_list:
os.makedirs(c["instance_data_dir"], exist_ok=True)

with open("concepts_list.json", "w") as f:
json.dump(concepts_list, f, indent=4)

import os
from google.colab import files
import shutil

for c in concepts_list:
# No need to ask for instance images, since we are using filenames as instance prompts
print(f"Using instance prompt {c['instance_prompt']}")

def clear():
from IPython.display import clear_output; return clear_output()

[idle stump]

Hi, I'm on 3090. Whenever my computer is left alone for a while and I generate something, the computer resets. I think my PSU is 700V. The shop that sold me the 3090 told me I need it. What might be causing the problem?

[brazen needle]

so i keep getting this error when trying to generate something. Does anyone know what to do? I've got no coding expierience btw so i'm completely lost

[karmic crown]

i can help with macs, and some other guys can help with linux and windows, but i honestly haven't seen anyone here who could help with Google Colab 🤷‍♂️

i think your best choice is to raise an issue on github
i hope this is the right one https://github.com/sagiodev/stablediffusion_webui

[brazen needle]

alright still thanks mate

[karmic crown]

i am planning to try colab when i find some time (next year)
maybe you will have more luck next time 🙂

[ornate elk]

Check the System logs for any crash indicators

[reef smelt]

medvram has an effect apparently?

1. medvram
2. no restriction
3. medvram

[ornate elk]

nope, xformers has an effect (not quality wise), medvram does nothing to the image

left is whithout medvram, right is with, (they are identical)

[reef smelt]

well for some reason it doesn't recreate the same thing when I don't have medvram

[ornate elk]

thats with xformers, (difference in ear, earring, head tilt)

[idle stump]

How do I go about finding these system logs?

[ornate elk]

its called Event viewer in windows

then on the left click on administrative, then on system, then check for the time and date the crash happens

[idle stump]

Thanks so much

[idle stump]

Doesn't seem too useful. Anywhere else I should look?

[ornate elk]

Check the logs before this event

[idle stump]

The previous event is 15 mins before it

Maybe next time

[ornate elk]

700w is a bit low for a 3090 I think. What's your CPU?

[rigid valve]

need some help, im getting Error occurred when executing KSampler: input must be 4-dimensional, when trying to do the stable difusion animate in comfy ui, i have an AMD 7900XT.

[urban merlin]

decided to try out sd.next and my intel arc a770 gpu isnt showing up anywhere in the webui and it defaults to using my cpu at all times

[karmic crown]

I suggest an upgrade of PyTorch version for A1111 on Macs, as well as a change of command line params
https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/14406

[urban merlin]

oh i think i forgot the --use-ipex param however trying to use it gives me the error ModuleNotFoundError: No module named 'intel_extension_for_pytorch'
i've tried installing it and it wont detect that its installed and it wont automatically install from what im aware of

[onyx ledge]

Anyone know why I can't install bitsandbytes?

[brazen needle]

GitCommandError: Cmd('git') failed due to: exit code(128) cmdline: git clone -v --filter=blob:none https://civitai.com/api/download/models/265938?type=Model&format=SafeTensor&size=pruned&fp=fp16 /content/gdrive/MyDrive/sd/stable-diffusion-webui/tmp/265938?type=Model&format=SafeTensor&size=pruned&fp=fp16 stderr: 'Cloning into '/content/gdrive/MyDrive/sd/stable-diffusion-webui/tmp/265938?type=Model&format=SafeTensor&size=pruned&fp=fp16'... fatal: unable to access 'https://civitai.com/api/download/models/265938?type=Model&format=SafeTensor&size=pruned&fp=fp16/': The requested URL returned error: 400 '

Anyone got an idea what's the problem here? It won't let me import models from civitai

[ornate elk]

whats your python version?

[ornate elk]

you cant git clone from civitai

[onyx ledge]

3.10.9

[young bridge]

Hey guys

if i generated a portrait photo of a person but would like to make it full body photo of that same person. how that can be done?

[storm void]

i got it to work now after downloading anaconda and cura but thanks I appreciate it and also could you please link me the 700 mb ones so I can download them if needs again, thanks!

[ornate elk]

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

[storm void]

Thanks!

[kindred tinsel]

Hello techsupport, (damn I feel like Im literally here every day), I tried using FaceDetailerPipe from ImpactPack for ComfyUI but I am getting this error

Something about tensors not matching the same size, this happened once during generation and since I havent changed the seed It happened again after I retried the same prompt, can anyone help me debug this?

[karmic crown]

@kindred tinsel you probably mixed sdxl with 1.5 stuff

check one by one and be sure that all of them are 1.5 or sdxl, and not a mix

[void kelp]

Guys, I'm literally going CRAZY. I'm trying to train Lora in Google Colab, but it doesn't work no matter how many times I try. Is there anyone who knows about colab? Thanks in advance

[kindred tinsel]

Turns out the error was about the SAM model using the GPU

It seems to only work with cpu

[kindred tinsel]

now its weirder tho @karmic crown , its finding the mask, I am hearing my cpu work lmao when it looks for the mask and somehow finds it but it isn;t changing anything, it has a prompt in there for the positive of what I want to change but nothing :c

[void kelp]

the problem is the following: every time I try to train lora for stable diffusion xl and try to insert an xl checkpoint, it always gives me this error. (ps: I use google colab pro):
/content/drive/MyDrive/CheckPoint/realisticStockPhoto_v10.safetensors
building U-Net
loading U-Net from checkpoint
U-Net: <All keys matched successfully>
building text encoders
Traceback (most recent call last):
File "/usr/local/bin/accelerate", line 8, in <module>
sys.exit(main())
File "/usr/local/lib/python3.10/dist-packages/accelerate/commands/accelerate_cli.py", line 47, in main
args.func(args)
File "/usr/local/lib/python3.10/dist-packages/accelerate/commands/launch.py", line 1017, in launch_command
simple_launcher(args)
File "/usr/local/lib/python3.10/dist-packages/accelerate/commands/launch.py", line 637, in simple_launcher
raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
subprocess.CalledProcessError: Command '['/usr/bin/python3', './sdxl_train_network.py', '--enable_bucket', '--pretrained_model_name_or_path=/content/drive/MyDrive/CheckPoint/realisticStockPhoto_v10.safetensors', '--train_data_dir=/content/drive/MyDrive/Elon_Musk/image', '--resolution=512,512', '--output_dir=/content/drive/MyDrive/Elon_Musk/model', '--logging_dir=/content/drive/MyDrive/Elon_Musk/log', '--network_alpha=1', '--save_model_as=safetensors', '--network_module=networks.lora', '--text_encoder_lr=5e-05', '--unet_lr=0.0001', '--network_dim=8', '--output_name=Elon Musk', '--lr_scheduler_num_cycles=1', '--no_half_vae', '--learning_rate=0.0001', '--lr_scheduler=cosine', '--lr_warmup_steps=50', '--train_batch_size=1',

'--max_train_steps=500', '--save_every_n_epochs=1', '--mixed_precision=fp16', '--save_precision=fp16', '--cache_latents', '--optimizer_type=AdamW8bit', '--max_data_loader_n_workers=0', '--bucket_reso_steps=64', '--mem_eff_attn', '--xformers']' died with <Signals.SIGKILL: 9>.

[karmic crown]

all i can tell you is that launch.py can't launch and just dies
SIGKILL: 9 = terminate process

[shrewd talon]

I really need help on this
I am using (trying) Fooocus and am downloading everything properly but in the instructions it says to use run.bat to start but I dont have that no matter how many times I download it, I tried fooocus mre and it had it and it worked

[tight olive]

Can you or someone recommend a current, up-to-date, guide for installing stable-diffusion-webui, rocm, pytorch, and etc on linux for AMD? I'm a beginner at linux so baby steps would help. I installed Ubuntu today and tried to get things up but I'm having trouble.

I'm getting some errors in trying to run webui.sh. UPDATE: newer version of pytorch installed and issue went away

[timber scroll]

im having the same problem

[kindred tinsel]

@ CS1o could help you both with that but they are not online, probably ping them to see if they might come around and help out !!

[ember crescent]

Когда пишу промт и использую символы, то почему то в строке ввода например символ " \ " у меня один, а когда нажимаю ввод, то пишет что у меня четыре символа " \\ " или если я каждый раз ввожу промт с символами " \ " или " % " то они начинают умножаться, если символов " % " у меня было два, то если я введу ещё раз этот промт, то их станет уже четыре. В итоге в строке ввода один промт, в чате другой, а в промте к картинке третий промт, и у каждого разное количество символов " \ " " % " интересное наблюдение

When I write a prompt and use symbols, for some reason in the input line, for example, I have one symbol ' ', and when I press enter, it says that I have four characters ' ' or if I enter a prompt with the symbols ' every time ' or ' % ' then they begin to multiply, if I had two ' % ' symbols, then if I enter this promt again, there will already be four of them. As a result, there is one prompt in the input line, another in the chat, and a third prompt in the picture attached, and each has a different number of characters ' ' ' % ' interesting observation

Введите например промт: !"№;%:?*()_+/ white white red red gold golden black black и тогда у вас начнут появляться дополнительные символы " \ " " % " а дальше они будут умножаться, это только у меня так?

[burnt badger]

Can anyone help me understand why my refiners are causing my images to come out digitally pixelated and color reduced? Probably the wrong words but look at the image.

image looks great until the very last step and delivers a crappy version

[sonic ferry]

Looks like one of those images you gotta cross you're eyes to see something pop out lol

[kindred tinsel]

that looks like a VAE decoding problem, most cases your model has either baked in VAE and it messes up at the last iterations or it is the method you are using, but it depends, what are you using for upscale, direct latent upscale in hires.fix or extra 4xultrasharp adn the like upscale?

[burnt badger]

4x ultrasharp

thing is that is works fine for a while

maybe i change refiner, then is starts doing this

are certain refiners and models incompatible? (yes I am new to all this)

[kindred tinsel]

I wouldn’t think so, does it give you any errors in the console? Anything about out of memory? It shouldn’t happen since 4x literally upscale the already finished image not the latent image

Unless you are using 4x during hires. Fix steps which could cause this issue. I had this when working with comfy but just to be sure @karmic crown do you know about this issue?

[burnt badger]

honestly it seems like its only when i use Dreamshaper_8 model, as refiner or model

[kindred tinsel]

Wait, you have dreamshaper as refiner? In the sense in the refiner tab ?

[burnt badger]

i can use it in both, as Automatic1111 lets me

[kindred tinsel]

Look I think it is an issue not either the vae but with lack of steps, how many steps is your refiner running on?

[burnt badger]

anywhere from 20 to 80

dreamshaper works fine as model with no refiner

[kindred tinsel]

For instance in comfy my refiner runs the same amounts of steps as generation after initial generation is finished

Maybe it’s the model then, some models haven’t been trained for having a refiner like that.

[burnt badger]

the steps is equal to teh base model steps

[kindred tinsel]

Weird lemme see your webui set up

[burnt badger]

refiner is off right now, but its just on default settings

[kindred tinsel]

Damn I do not see anything else that could be wrong, the denoising strength is okay for the way you are using the hires fix. I honestly had hoped Viking would have joined up to give you some help but man idk. Keep asking later tonight and see if he or other mods connect up

[burnt badger]

Thanks for the input at least

I am thinking the model just doesn't want to play with refiners

[reef smelt]

is token merging bad for image quality?

or is it overalll good

what's the side effects?

[sonic ferry]

Anything I should know about running ComfyUI from an external drive? I'm just running on a small office PC in CPU mode, only have a 250GB SSD in this PC and I'm running out of space. 512x512 taking about 7 minutes, would it be slower off of external media? Might be an HDD actually to top it off, don't remember what I had in the external drive case.

[worn magnet]

Hi, I'm banging my head against a wall trying to install stable diffusion. I keep getting the "No module named pip" and "RuntimeError: Couldn't install torch" errors. Seems somewhat common but can't figure how to fix it with google searches. Can anyone help please?

[sonic ferry]

You got pip installed?

[worn magnet]

I installed python 3.10.6, do i need to install pip separatley?

[sonic ferry]

Internet says 3.4 should have it, so guess not.

[burnt badger]

anyone know where to get the latest control net models? I am getting version errors

[worn magnet]

is there another place I should ask this question?

[proper pike]

Can anyone help me figure out why img2img is essentially just doing nothing other than making my image uglier? I'm trying to make Jerry look like a real mouse, no matter my prompting sampler denoise strength etc.. it just looks like a discolored original. Left original, right is just him getting darker and smoother

txt2img I can make a realistic mouse no problem

[jaunty cedar]

Anyone know why all the color gets taken out of my generation?

[kindred tinsel]

Try changing the VAE if you are using 84000 try moving to kf-18 anime or see the one that the model recommend( if you still get bad colors try changing to the model recommended sampler and try it out. But from what has happened to me VAE might be the culprit

[kindred tinsel]

If you are installing python by the installler found in their page, like in Python.org you can check a box in the advanced installation that says “Add PIP to path” which makes the system recognize the pip install commands that are used in stable diffusion. To change this after installation you can rerun the installer and check on modify installation and add pip to pass checkbox should show up

[livid jetty]

when i try to use Stable Diffusion it just turns into a grey image, but somthimes it will be what i wanted.

[kindred tinsel]

Can you reproduce the error and send an image of your generation data prompt, and model data?

[kindred tinsel]

Control net models usually are downloaded along with the installation of control net extension in webUI. If you are trying to use control net with sdxl you need the exact models as well as 2.1 models if you are using that version: if you feel like the extension doesn’t have any models which would show up in ( extensions/ controlnet/ models) is empty or is giving you outdated errors check the stable diffusion version you are running: or if you want to download the models yourself here is the link:

(SDXL)

https://huggingface.co/lllyasviel/sd_control_collection/tree/main

(1.5)

https://huggingface.co/lllyasviel/ControlNet/tree/main/models

There are other for 1.5 but I can’t find them rn

[kindred tinsel]

If you are running confuUI I would strongly suggest you do not run anything that has to do with external or mechanical drives. If it’s an SSD you should be fine or even run it from your boot drive.

Also your run time might not be long because of python but because of you runnnf it on cpu. CPU is always slower for this since it’s not optimized for the kind of operations that AI generative models do in the GPU. I would recommend you use services like collab or civit AI generation where you pay for a good GPU and use it as you wish

Comfy for my own AMD card brings temps up to 91 C and it might kill your computer and it wouldn’t be good.

[livid jetty]

uh... ignore the prompt lol

[kindred tinsel]

Okay ( I don’t know who that is either way) you might not have a VAE selected since the tab is mot visible in your stable diffusion. For that head over to settings and to user interface and in the section where it says Ui tabs or smth press on it and begin to type vae and it will show the option for it

https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/9703

There is guide for it there too ^^

[strange thunder]

Has there been a recent way/breakthrough in being able to use loras for specific parts of an image so that they don't all blend together? I am using an old method and was wondering if a new one had surfaced yet.

[burnt badger]

Thank you so much, i will see if these work

[kindred tinsel]

Np ^^ hope the error goes away, also do check your installation folder and or the extension tab in web ui and update extensions to check for changes in versions that might fix the error

[sonic ferry]

Hadn't even thought of temps, been using the fan spinning up and down to know when it's done. I'm 75c right now watching an anime with ComfyUI generating in another tab, that temp isn't too bad, right?

[kindred tinsel]

Depends, if you are running your cpu at constant high temps; for hours at a time ( like I am doing on my GPU) constant high temps will injure the hardware and degrade it faster. I would recommend either getting better ventilation or using online solutions

[sonic ferry]

Oh, I do see it tickling 80c.

[kindred tinsel]

If your computer is a eork computer it might be best to not damage it through stable diffusion besides you would get faster results and much more creative liberty through that or buying yourself a pc

[kindred tinsel]

Yeah if it gets too high for too long. You’ll have to change your thermal paste more often and get better ventilation and so on

[sonic ferry]

Just something I bought off of eBay as in-between till I can get actual hardware, my other PC is 10 years old.

[kindred tinsel]

Yeah it will hold up until you get better hardware just don’t overdo it haha

[sonic ferry]

Yeah, gonna have to stick to playing around with 7 min 512x512 gens

[tight olive]

I had to install a newer version of pytorch.

[void cairn]

Hi all,
Got a kinda technical question I hope anyone from here can help me 🙂
On the moment I have Stable diffusion 1.5 with automatic 1111, I want to go to SDXL because I see such good results.
How can I remove the version of SD I have now from my harddrive, and is it neccesary to remove it before I get SDXL?
(Because it saves space on my drive)
Thanks in advance for the help and advice ❤
Greetings and happy holidays❤

[karmic crown]

A1111 can be used with 1.5, 2.x and sdxl models

just add some sdxl model(s) to your current a1111. sdxl models are around 6-7GB each
you can clean up your models and remove the ones you do not need later

in your stable-diffusion-webui folder you have a folder called models and inside that folder is the one with models called Stable-diffusion

[karmic crown]

this usually happens when you are using too many steps, that is the same thing as when you are baking something in the oven and forget to turn it off

[tight olive]

I'm now running WebUI! Woohoo! However, I'm getting an error and can't generate. The error is "RuntimeError: mat1 and mat2 must have the same dtype, but got Float and Half"

Someone suggested on github.com: "I was having that issue, and I fixed it by using --precision full --no-half"

Where would I put those arguments so I can see if that fixes it?

[karmic crown]

what are your gpu and os?

[tight olive]

Running AMD Radeon RX 6800XT 16GB on Ubuntu Studio 23.10

[karmic crown]

in that case webui-user.sh

[karmic crown]

find this part

# Commandline arguments for webui.py, for example: export COMMANDLINE_ARGS="--medvram --opt-split-attention"
#export COMMANDLINE_ARGS=""

and show me what you have there

[tight olive]

It's commented out or empty

Commandline arguments for webui.py, for example: export COMMANDLINE_ARGS="--medvram --opt-split-attention"

#export COMMANDLINE_ARGS=""

[karmic crown]

you haven't touched that file yet?

ok

[tight olive]

nope, just installed

[karmic crown]

add this, restart and try again

# Commandline arguments for webui.py, for example: export COMMANDLINE_ARGS="--medvram --opt-split-attention"
#export COMMANDLINE_ARGS=""
export COMMANDLINE_ARGS="--medvram --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --upcast-sampling"

lines that start with # are comments
lines without # at the beginning are commands

[tight olive]

RuntimeError: mixed dtype (CPU): expect parameter to have scalar type of Float

Throwing a different error now

[karmic crown]

try with --no-half instead of --no-half-vae

just remove -vae from that line

[tight olive]

It's trying now, waiting for it to finish

Worked!

Thanks so much for your help!

Not the most elegant "dog", but he's there. hahah

[karmic crown]

i am not sure if those are most optimized parameters for you, so please show us your generation time and settings (copy paste the text below the image)

[tight olive]

dog
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2735423496, Size: 512x512, Model hash: ad2a33c361, Model: v2-1_768-ema-pruned, Version: v1.7.0

Time taken: 1 min. 50.6 sec.

[karmic crown]

wait

do not use 512x512 with 2.x models

it needs to be 768x768 at least

my advice do not use 2.x models

use 1.5 or sdxl

[tight olive]

Ok, I've got the SDXL in there

[karmic crown]

for 1.5 use 512x512 (or larger)

[tight olive]

I'll try it

Not 512x512 for XL either then?

[karmic crown]

for sdxl you can use something like 1024x1024 or more, but they work with 768x1024

or 1024x768

[karmic crown]

you got this ugly thing because you used wrong (too small) dimensions

[burnt badger]

Is there a way to increase the amount of steps that work?

I mostly work around 50 to 80 steps

[karmic crown]

more steps doesn't always mean better quality

[burnt badger]

Gotcha

[karmic crown]

see this #🏞｜general-with-images message

[burnt badger]

8 steps, wow

[karmic crown]

exactly

but i used a few "tricks" there lcm and sdxl renderer

[burnt badger]

Lcm?

[karmic crown]

lcm reduce the number of steps needed to generate image and sdxl rendered is a lora that helps get better images with 1.5 models without refiner

[burnt badger]

It's just called lcm?

Reducing steps is a good idea

[karmic crown]

https://civitai.com/models/195519/lcm-lora-weights-stable-diffusion-acceleration-module

[burnt badger]

I'm running sdxl turbo, do I need sdxl renderer?

[karmic crown]

https://civitai.com/models/171159?modelVersionId=236130

i am still sleepy, i was talking about sdxlrenderer

sdxl turbo is for speed, but not for qaulity

i need coffe

[burnt badger]

I have regular sdxl

As well

[karmic crown]

oh, yes

you need 1.5 lcm lora for 1.5 models
and sdxl lcm lora for sdxl models

[burnt badger]

Copy

[karmic crown]

sdxl rendere is 1.5 lora, do not mix with sdxl models

it is itended to make 1.5 images more like sdxl

[burnt badger]

Lcm is fine with XL?

[karmic crown]

to get LCM sampler install https://github.com/continue-revolution/sd-webui-animatediff

it will popup in the dropdown below unipc

[burnt badger]

Okey dokey

[karmic crown]

btw, there are some lcm models that already have it baked, so you do not need to use lora (i love those)

[burnt badger]

Thank you

[karmic crown]

https://civitai.com/models/200061?modelVersionId=256258

i love this one

[burnt badger]

That looks great too

Is it wise to merge checkpoints I like?

[karmic crown]

some things should not be mixed, but you can try and see

i need to go, i need to buy coffee, it is still early
there is too much blood and not enough coffee in my veins
🤣

[tight olive]

I'm an espresso guy, I've been roasting my own the last couple weeks

[void cairn]

So as I understand I dont need SDXL but instead I can use the models from it because I have 1111, right? thanks for your help , very much appreciated 🙏

[ornate elk]

If you want to safe space use these 1.5 controlnet models:
https://civitai.com/models/38784/controlnet-11-models

[ornate elk]

The problem here is, that your trying to use a Lora as model. That won't work.

Loras go into models/lora
Models are always bigger than 1.98gb and go into models/stable-diffusion

[ornate elk]

Auto1111 is the tool to work with stable diffusion models.
It supports the 3 different models and loras etc

You just need to download the model you want to use and put it into models/stable-diffusion

Then select it in the dropdown

[ornate elk]

What are your settings in img2img?

[ornate elk]

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

[void cairn]

when I downloaded an XL version I got this error when trying to make an image, any ideas? "NotImplementedError: No operator found for memory_efficient_attention_forward with inputs: query : shape=(1, 9216, 1, 512) (torch.float16) key : shape=(1, 9216, 1, 512) (torch.float16) value : shape=(1, 9216, 1, 512) (torch.float16) attn_bias : <class 'NoneType'> p : 0.0 cutlassF is not supported because: xFormers wasn't build with CUDA support flshattF is not supported because: xFormers wasn't build with CUDA support max(query.shape[-1] != value.shape[-1]) > 128 tritonflashattF is not supported because: xFormers wasn't build with CUDA support max(query.shape[-1] != value.shape[-1]) > 128 triton is not available requires A100 GPU smallkF is not supported because: xFormers wasn't build with CUDA support dtype=torch.float16 (supported: {torch.float32}) max(query.shape[-1] != value.shape[-1]) > 32 unsupported embed per head: 512 ""

[reef smelt]

this is scary

is there a way to stop stable diffusion from using too much vram?

I rather have it take more time than crash

I keep getting </3

empty_cache
    torch._C._cuda_emptyCache()
RuntimeError: CUDA error: device-side assert triggered
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

I get it from using animatediff

[ornate elk]

What's in your webui-user.bat?

[ornate elk]

Animatediff is very resource hungry. You can only try using --lowvram but I don't think it will help

[reef smelt]

hmm

people say

[ornate elk]

Yes use that

[karmic crown]

do not use 512x512 for sdxl
use 1024x1024, 768x1024, 1024x768

A100 GPU? it should require --xformers as the only parameter, i guess

[karmic crown]

@ornate elk I think I finally optimized 1660 Super

Steps: 20, Sampler: DPM++ 2M Karras, Size: 512x512, Model: cyberrealistic_v40, Version: v1.7.0, Batch Count: 4

GTX 1660 SUPER - Time taken: 53.7 sec.
M1 MacBook Pro - Time taken: 3 min. 30.1 sec.

"COMMANDLINE_ARGS": "--xformers --opt-sdp-attention --no-half --medvram-sdxl",
"TORCH_COMMAND": "pip install torch==2.1.2+cu118 torchvision==0.16.2+cu118 --extra-index-url https://download.pytorch.org/whl/cu118"

[ornate elk]

you cant use xformers AND opt-sdp-attention, it will only use one of them at a time

[karmic crown]

ooops, you are right

[ornate elk]

also did you tried upscaling or controlnet too? because without --medvram it could get ouf of vram

[karmic crown]

i will try later when i get a chance. my son is playing LOL currently instead of preparing for his C exam 🤣

when you have time, try to test and compare timings with torch 2.1.2 on your pc

most 2.1.2 release notes are mac related, so i do not know if it has any impact on nvidia on windows

it is posible that they had some optimizations in between

i didn't check

[ornate elk]

it could be faster as i could install xformers 0.0.23 too

[karmic crown]

btw, check this https://www.reddit.com/r/StableDiffusion/comments/1379y2m/optimal_installation_of_pytorch_20_xformers_or/

some guys says that their 4090 works faster without xformers ?!

or that was the case when 2.0 just came out

i installed that too

[ornate elk]

Yea outdated info

We've done some tests with xformers vs opt-sdp-attention and it turns out they are the same speed but xformers use less vram

[void cairn]

I try again with a different checkpoint model and will get back to it, thanks for your answer 🙂

[void cairn]

I also used 768x768, but that didnt work either, the thing is, I am not that technical...and that I have installed everything I have now is a Gods miracle 🙂

[kindred tinsel]

@karmic crown mind jumpjng in here? They have a problem with the image coming out grey, I thought it was because of no vae decode but idk

[karmic crown]

i am 99% sure it is because they need to add --no-half, since i have no other information

[kindred tinsel]

Ah I see, so if VAE doesn’t work then it might be webui bat file 📝📝

[karmic crown]

vae can be a problem only if it is 1.5 vae with sdxl or 2.x model

in every other case vae is not a problem

(unless it faulty one)

overcooked images are almost always related to over use something (steps count and cfg in most cases)

noise and gray/green/black images are almost always related to --no-half

if someone with more expirence disagree, i am really willing to listen, but based on my exprience so far (which is very short - just a little more than a month) all i just wrote is correct

[fair oxide]

https://www.youtube.com/watch?v=onmqbI5XPH8 6:57

I can't find "webui-user.bat"

[kindred tinsel]

Okeee than you, I tried helping out last night to some ppls questions. Here hence I wanted to learn

[karmic crown]

this one

[kindred tinsel]

Yeah

If you enable extensions column in the list viewer it will say .BAT File

[fair oxide]

Does this still work?

Oh

[karmic crown]

what is your gpu? let me give you short and easy guide made by our freind CS1o

nvidia or amd? (and i guess windows)

[void cairn]

it says this, used dreamshaper XL turbo: NotImplementedError: No operator found for memory_efficient_attention_forward with inputs:
query : shape=(1, 12288, 1, 512) (torch.float16)
key : shape=(1, 12288, 1, 512) (torch.float16)
value : shape=(1, 12288, 1, 512) (torch.float16)
attn_bias : <class 'NoneType'>
p : 0.0
cutlassF is not supported because:
xFormers wasn't build with CUDA support
flshattF is not supported because:
xFormers wasn't build with CUDA support
max(query.shape[-1] != value.shape[-1]) > 128
tritonflashattF is not supported because:
xFormers wasn't build with CUDA support
max(query.shape[-1] != value.shape[-1]) > 128
triton is not available
requires A100 GPU
smallkF is not supported because:
xFormers wasn't build with CUDA support
dtype=torch.float16 (supported: {torch.float32})
max(query.shape[-1] != value.shape[-1]) > 32
unsupported embed per head: 512

[karmic crown]

@void cairn please find and show us webui-user.bat

[fair oxide]

I edited the notepad but stuck here at the next step

[karmic crown]

do you have amd or nvidia?

[karmic crown]

models? which one is btter?

[void cairn]

this is everything I see now, can you see the text?

[karmic crown]

go to your C:\Users\HWRem\stable-diffusion-webui\ and find a file named webui-user.bat and post it here

[karmic crown]

@void cairn

the one above webui-user.sh (it migh not show extension)

[karmic crown]

in that case it is probably ryzen + nvidia

[fair oxide]

What do I do

[kindred tinsel]

What were the changes on the webui user file that you did

[karmic crown]

remove this folder

[viscid badge]

the .bat

[karmic crown]

uninstall pyton

[viscid badge]

the launcher is the "web-user.bat"

this

[karmic crown]

and do this #🤝｜tech-support message

[kindred tinsel]

That’s the launcher too

I said webui user file did you also need the .bat at the end what are you correcting me on i don’t get it lmao

[fair oxide]

I highlighted and backspaced, terminal closed

[viscid badge]

my bad I read the wrong message lol

I have a question, but let me provide some context:

I'm starting to create small games on Ren'Py (a game engine primarily using drawing), and I wanted to know if it's possible with StableDiffusion to: generate a full-body character and use the same character in different positions, change its facial expressions while keeping the base character?

(It's all a bit complicated...)
(repost message I had already sent)

[karmic crown]

close the cmd window
remove folder with windows file explorer
uninstall python
and save and run script, it will install everything (including python) for you
and it will work
then we can work on custom parameters for better optimization

[fair oxide]

remove what folder?

[kindred tinsel]

That’s probably generating one image and using img2img with control net and impainting to generate same character but Change clothing and position etc

[fair oxide]

Also I can't uninstall python

[viscid badge]

why

[fair oxide]

I would like to know myself

[viscid badge]

show the message

what he sent when u try

[fair oxide]

I pressed uninstall and it brings me to settings, then the windows close

[viscid badge]

screenshot

[karmic crown]

guys, it is very problematic to get a proper answer when 10 of us write at the same time

[fair oxide]

I can't screenshot a process

[karmic crown]

so please have a little patience

[viscid badge]

no messages of error ?

[viscid badge]

yes

[karmic crown]

lets first solve chips problem

you installed stable diffuser webui (automatic1111) on your deskop, remov that installation, since it migh make problems

uninstall python, since i gave you a script that will install it using winget (windows package magerer that exists in windows 10 and 11)

[fair oxide]

sd?

[karmic crown]

stabile diffuson webui

if you want step by step guide, i can give you one

[karmic crown]

but script is faster and easier

#🤝｜tech-support message

click there for a step by step guide

[fair oxide]

When I press uninstall it brings me to settings

Then everything closes

[karmic crown]

ok, run settup again and click And check "add python to path" when installing Python.

[viscid badge]

Windows Settings ?

[fair oxide]

Yes

[karmic crown]

then follow the rest of the guide

[viscid badge]

this window but in your language ?

[fair oxide]

setup?

[karmic crown]

pyton install setup, run it again

[fair oxide]

I can't find it in my dowloads

[karmic crown]

that is why i told you it is better to run script

[fair oxide]

So I redownload python?

[karmic crown]

if you want to follow manual steps, yes

[viscid badge]

He said he couldn't uninstall Python?
He doesn't have access to that page through the installer if Python is already installed?

he need to uninstall in first and launch the installer of python for hving this window of choice

[karmic crown]

he can install it again with right params (when he download it again)

[viscid badge]

mhh ok

[karmic crown]

and he doesnt need to uninstall it first

[viscid badge]

oh yes I see

modify

[fair oxide]

Oh

Okay

[karmic crown]

repair

[viscid badge]

repair it

[karmic crown]

just in case

modify should work, but lets be sure

[viscid badge]

all the pyth vers work for stable ?

[karmic crown]

3.10.6 is fine

just do not use 3.11 or 3.12

[viscid badge]

ok ok

[fair oxide]

I pressed repair, what do I do next

[karmic crown]

wait

[viscid badge]

what is ur pyth vers ?

[karmic crown]

i pasted the wrong thing

#🤝｜tech-support message

[fair oxide]

https://www.python.org/downloads/release/python-3106/

64 bit

[karmic crown]

add to enviroment

[viscid badge]

good

[fair oxide]

How do I get here

[karmic crown]

what you get after repair?

[fair oxide]

It loaded and then closed

[karmic crown]

and if you run ti once more and clikc modify?

do you get that window from screenshot?

[viscid badge]

@karmic crownif u can check in #📝｜prompting-help maybe u can help me or have an idea on what happen
check when u have time ! ty

[karmic crown]

click next

and you should get the screen i posted

[karmic crown]

install

[viscid badge]

no need to isntall for all user ?

[fair oxide]

Done

[karmic crown]

ok

open cmd

and type pip --version

[karmic crown]

not really, on windows i have it aslo installed for my user only

[fair oxide]

?

[karmic crown]

remove the space

pip --version

like this

[fair oxide]

done

[karmic crown]

works?

[karmic crown]

huh

wait

was your cmd opened before you reinstalled pyton?

[viscid badge]

If you're getting the error that "'pip' is not recognized" in the Command Prompt, it means that pip (Python's package installer) is either not installed or not added to your system's PATH.

[karmic crown]

if yes, close cmd adn run it again

[fair oxide]

no

This thing hates me

[viscid badge]

dont worry, I had all ur problems too

It take me a long moment before I install it

[karmic crown]

try this

1. open notepad
2. copy paste this script

@echo off
winget install -e --id Python.Python.3.10
winget install -e --id Git.Git
winget install -e --id mbuilov.sed
mkdir c:\ai\
cd c:\ai\
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui
cd stable-diffusion-webui
sed -i "s/set COMMANDLINE_ARGS=/set COMMANDLINE_ARGS=--xformers/" webui-user.bat
call webui-user.bat

3. save as install.bat and run it

it will do everything for you

[viscid badge]

cheat code lol

[karmic crown]

this is very simple code, with just a few lines, but it works

[viscid badge]

yes, @echo off is always good for do some code automaticly

[karmic crown]

mac users can use my script which has many advanced things compared to this one

[fair oxide]

[karmic crown]

all files

[viscid badge]

just type name.bat

[karmic crown]

this

[viscid badge]

all files = program
.bat is the launcher of program

[fair oxide]

Done

[viscid badge]

launch it

[fair oxide]

How

[karmic crown]

double click the file you just created

[viscid badge]

if u have this icon on ur file, double click on

more little, more little my screenshot lol

[karmic crown]

yes

y

y enter probably

waith wait

give me a second to find that script

your winget is outdate

there is a script that someone made that update it

give me a second to find it

i had te same problem

[viscid badge]

"winget upgrade --force"
winget upgrade

??

[karmic crown]

it is more complicated

it must run from powershell

[viscid badge]

i see

"winget upgrade" work on powershell for me

[viscid badge]

aaah for update it okk

[fair oxide]

[viscid badge]

Chips VS Stable Diffusion, infinite fight

[karmic crown]

https://github.com/microsoft/winget-cli/issues/3068#issuecomment-1763402494

i found it

[viscid badge]

gg

[karmic crown]

open powershell

@viscid badge please help with this since i currently have no access to pc

@fair oxide and copy paste that code

it will reinstall your winget

then you will be able to install ANY software you want with winget

or to update your apps

From the Start menu, click Start, click All Programs, click Accessories, click the Windows PowerShell folder, and then click Windows PowerShell

[viscid badge]

tell me

[karmic crown]

i found how

[viscid badge]

dont know if i can help, all my pc is in french

[viscid badge]

that it

[karmic crown]

yes

thats it

[ornate elk]

the issue is that your torch version is too old, because you have dreambooth installed

[karmic crown]

press enter there

[fair oxide]

What do I do next

Okay

[void cairn]

What do I do about that?

[karmic crown]

now my script should work

[viscid badge]

it it update now it will work

if u have no demon on u, it will work

[void cairn]

I will wait offcourse when the others are ready for my problem:)

[ornate elk]

delete the dreambooth folder from the extensions folder. then delete the venv folder and relaunch the webui-user.bat

[fair oxide]

You see, Chip’s demon is a very special and painful thing

[viscid badge]

im not a curse exorcist

srry

[fair oxide]

[void cairn]

thanks, I will look into that !

[viscid badge]

Microsoft.DesktopInstaller need to be close

[ornate elk]

some popup is open in the background maybe, wating for an Accept

[karmic crown]

please dm me

[fair oxide]

Will do

This infernal thing is resisting me

[karmic crown]

wait

i posted a link and now i cant see you

[fair oxide]

Wot

[ornate elk]

Let me know if it worked. You can check the version at the bottom of the webui in browser

[karmic crown]

i am pretty sure i am on your spam list now

please check

[fair oxide]

you're not

[viscid badge]

viking is blacklisted

[fair oxide]

I guess discord is siding with stable diffusion

[viscid badge]

x)

[karmic crown]

send me a message

[fair oxide]

I did bro

[karmic crown]

again

[viscid badge]

add eachother in friend

if u dont find eachother, create group

[void cairn]

well its back and this is the version, and btw I got an error code everytime I started it, that on is gone I think, so thanks for that ^^

[ornate elk]

Okay good the next thing you should do is edit the webui-user.bat
And at the line COMMANDLINE_ARGS=
you add: --xformers --no-half-vae
Then save and relaunch
Xformers will improve the performance

[void cairn]

okey , I will finish my work and look into it 🙂

[void cairn]

thanks alot, its there again and the version is :

[ornate elk]

Perfect no problem 🙂

[void cairn]

do you think I can run the SDXL checkpoints now, without having to install SDXL?

[ornate elk]

Yes that should work without problems

[void cairn]

its working without that error, thank you so much 🙂

[ornate elk]

No problem have fun!

[calm cairn]

Hi all, 3rd time here, I dedided I want to install XL and I really don't know where to start. Can anyone point me to a simple guide?

[karmic crown]

os?

graphic card?

[calm cairn]

ah I'm running Windows 10 and my card is a NVIDIA GeForce GTX 1660 Ti

( I expect im at the bottom end)

[karmic crown]

@ornate elk i really have no idea what else to try here

he tried to manually reinstall, install, add to path, run my script

but it is always like this

[kindred tinsel]

Did you reinstall python? I got that and added to path and it worked

There might be other installations already in path and it might be conflicting

[karmic crown]

other than 3.10 you mean?

hmm

[kindred tinsel]

Like 3.10.6 and 3.10.11 at the same time

Maybe a windows version from the store is there

But idk it’s so weird maybe venv folder messed up

[karmic crown]

this is my script that just clond a1111 in new folder

he removed the complete folder before running the script

so no venv

and see the path for python, it is not a venv one

actually

wait

[kindred tinsel]

There really shouldn’t be issues in there then

[karmic crown]

lets check the version first

[kindred tinsel]

If the venv is correct then that means it’s q python installation error. Pip should’ve been added with the thing can he access pip through normal cmd ?

[karmic crown]

see the path

[kindred tinsel]

It’s referencing an exe

The path is correct tho that is the location of install of python in every manual install

Maybe the fact that it’s falling directly python.exe is the error

[karmic crown]

python --version returns 3.10.11 🙈

[kindred tinsel]

Dude I use that version and I have never ran into this problem so weird

[karmic crown]

C:\Users\elogi\AppData\Local\Programs\Python\Python310\python.exe --version
also returns 3.10.11

pip --version
Not recognized as internal or external command

☠️

and this is version installed by winget that perfectly worked on my windows

winget installs the one from python.org

just donwload and run installer for you

how it is possible that python is in path, but pip is not?!

[kindred tinsel]

Well I’m not so sure about pip version since I have never run that in my pc but I’m sure it should at least detect pip for some install commands: how about you install with normal python installer and add to path through there

[karmic crown]

didnt work

he tried modify + add to path
uninstall
there was no option for uninstall?!
so he uninstall with winget uninstall --force

than installed with winget (basicly same thing as manual install)

[kindred tinsel]

No option to uninstall python?!

Now I’m no expert but that’s not how windows works LMAO

His copy got fucked? Is he on the latest version of windows ? Like on the latest 21h1 build ?

[karmic crown]

this is the last time i decided to help someone with windows

[calm cairn]

are you talking about me? i thought you were just busy

[karmic crown]

@fair oxide i am really sorry, but i have no idea how to help you

i tried every think that i could

[fair oxide]

[fair oxide]

I’m sorry for wasting your time

[karmic crown]

we even tried to put this whole path to webui-user

[fair oxide]

Computers have always give me trouble

[karmic crown]

we are talking about chip

i was with him on private

but i didnt managed to help him, since his pc is obviously cursed

[calm cairn]

ok cool, i was worried I did something to upset you, glad I didnt

[karmic crown]

i have no idea what else to try

[fair oxide]

@viscid badge it's jover

It's all over lads

[karmic crown]

#🤝｜tech-support message

click on link and follow the guid

[ornate elk]

Install the latest python 3.10.11 and then run pip cache purge
Then delete the venv

[fair oxide]

Slightly unrelated, but my main pc also gives me a lot of trouble
Currently it has locked me out (password not working), no matter what I try
I'm on my laptop

[karmic crown]

its not about venv

pip doesnt work

he reinstalled python a few times

he cloned a1111 in a new folder

we tried manul and winget methods

its always the same

python works, but pip does not

[ornate elk]

What's the python version?

[karmic crown]

3.10.11

@fair oxide one last thing

[calm cairn]

so I think I may have miscommunicated--
I already have SD installed and have been using for a while
I always use Automatic1111
I normally use 1.5 models, tho I did get a 2.0 working with some -options in the batch file that I use to run it
now I want to use SD XL, I have a Juggernaut XL model and can't use it without changing something

so I appreciate the link but that looks like it's for someone who has not installed Automatic1111 at all

I just want to know how to use the Juggernaut XL model--hope that is clearer, sorry about any confusion

[karmic crown]

download this installer (or find it on your pc)
https://www.python.org/ftp/python/3.10.11/python-3.10.11-amd64.exe

uninstall, and then install again, but for ALL users

[ornate elk]

What's your GPU?

[karmic crown]

same as mine

@fair oxide be sure to install like this

[karmic crown]

#🤝｜tech-support message

[calm cairn]

NVIDIA GeForce GTX 1660 Ti

[fair oxide]

Yeah I did it that way

[kindred tinsel]

But that version is too old, webui needs between 3.10.6 to 3.10.12 I think so choose smth like basic .6 or .11 to get the best out of both worlds

Maybe that is why it doesn’t work

[fair oxide]

Wtf do I do

[karmic crown]

i have this on 1660 super --xformers --no-half --medvram-sdxl

just add that to webui-user.bat and run that one instead of webui.bat

[kindred tinsel]

Let me try and see where the pip is because some ppl say your need to add the scripts folder to path

[karmic crown]

@kindred tinsel which version is too old?

3.10.11

[kindred tinsel]

3.10.5

[karmic crown]

that is a screenshot to see what to click

[kindred tinsel]

3.10.11 works fine for comfy and webui

[karmic crown]

i sent link for 3.10.11

please read what i wrote 🙂

#🤝｜tech-support message <- which version this was? 🙂

i cant waste time and find a screenshot of 3.10.11 just so show which things should be checked

[kindred tinsel]

@fair oxide

Notice how this doesnt say the file is located in the general python310 folder, go ahead and add this path to your PATH variables under system variables . I have found that pip is located in the Scripts folder
C:\Users\santi\AppData\Local\Programs\Python\Python310\Scripts try it out maybe it detects it from there

like forcing the path to look directly into the folder where the pip can be found in your installation, maybe then it will directly recognize it and use it to install the modules, or at least allow for checking the version

[fair oxide]

How do I get to that

[kindred tinsel]

go ahead and enter ```%appdata%````

[karmic crown]

Python310\Scripts ☠️ just kill me!

[fair oxide]

Idk what I’m supposed to do

[karmic crown]

now i understand how it is posible that python works and pip does not

[kindred tinsel]

shot mb, run windows + r

that will pull up the run windows

then input the %appdata%

[karmic crown]

@fair oxide looks like there is hope for you, @kindred tinsel will help you escape your python hell

[fair oxide]

What do I do next

[kindred tinsel]

then go to local -> Programs -> Python -> Scripts.

Take the entire address that is in the top bar above the folders.

[karmic crown]

my brain still trying to process Python310\Scripts

[sonic ferry]

GTA V style LoRA, I see a trigger word but...is that it? What's stopping my other ones from being activated by trigger words? How do I know it's gonna activate, I thought I knew what I was doing but now that I found Expression Helper LoRa and it only shows trigger words instead of a lora:LoRa_Here:# now I'm confused.

[kindred tinsel]

In here you can will arrive to the scripts folder, IT WILL HAVE A BUNCH OF .PY FILES IN IT.

In here you can take the full address by clicking on the address bar above the files

click here and press ctrl + c

[fair oxide]

No idea how to go to local

[kindred tinsel]

show me where you are now, did you input the command?

[fair oxide]

Yes

[kindred tinsel]

okay send screenshot of folder

[karmic crown]

send me civitai link where you found it

[kindred tinsel]

if it gives you a bunch of folders is because the command sent you one folder deep that you dont need,

you probably are here

click on appdata

[fair oxide]

Yes

[calm cairn]

trying it now--i want to know, will this configuration be able to run XL but not 1.5 models? Or will it be able to run both?

[kindred tinsel]

Head to local

[fair oxide]

Done

[karmic crown]

both

[kindred tinsel]

Navigate to the folder called programs

[sonic ferry]

https://civitai.com/models/119307/gta-v-loading-screen-artstyle this and this one https://civitai.com/models/105857/expressions-helper-realistic

[kindred tinsel]

you should have this there

click on Python, then Python3.10

[fair oxide]

Okay

[kindred tinsel]

should look like this

[calm cairn]

then great, thank you for the help

[fair oxide]

Yes

[kindred tinsel]

okay no head over to the Scripts folder

[fair oxide]

Wat

[kindred tinsel]

there is a folder, inside of the python310 folder we just entered that says Scripts

[karmic crown]

but be mentally prepared, sdxl is sllllloooow

[kindred tinsel]

press inside of it

[fair oxide]

Done

There’s nothing inside

[kindred tinsel]

what do you mean there is nothign inside?

send screenshot lmao

[kindred tinsel]

MY MAN, I think you realy did butcher the python install

it should be populated with the PIP at least

[fair oxide]

Will you stand where Viking fell, are you match for my blasted pc

[kindred tinsel]

let me check on smth real quick

[fair oxide]

I gtg but I'll be back

[ornate elk]

Do you have an antivirus software installed?
Or are you from Asia?

[kindred tinsel]

okay three things we can do rn

1. Reinstall python which I assume you already did with Viking. If you were to do this you would have to unistall everything from windowsx, and do everything once more including the add python to PATH step

2. Repair python installation if you isntalled it thhrough the python .exe installer

3. Repair the install through python -m ensurepip

[karmic crown]

I guess you are confused about why there is no lora:filename:multiplier there? right?

[sonic ferry]

Yeah, I did see one example on the expression helper that had a 'trigger lora:last:#' as if that makes it SEARCH for the trigger in the folder of Lora or something?

[kindred tinsel]

I understood that you couldn't unisntall python for some reason right @fair oxide ?

[fair oxide]

Both

[kindred tinsel]

There is no way that the antivirus would stop the installation of python

:0

[karmic crown]

that is right in most cases, but in some cases it is enough to use lora:file:#
i am not sure if you can trigger lora just by trigger word in A1111, you can in comfy

try with just trigger word and see if it works or not

[fair oxide]

What do I do

[karmic crown]

if lora is used it will be mentioned in the genration info at the end

[sonic ferry]

Ah, I do use ComfyUI, I'll try and make something simple real quick and see if it shows an expression

[kindred tinsel]

Did you run the python -m ensurepip command or does that also not work?

[calm cairn]

launch.py:error: unrecognized arguments: -sdxl

[kindred tinsel]

that should install pip automatically for you, in my case it threw a messagae since I already had it

[karmic crown]

screenshot please

you probably have some unwanted space

[kindred tinsel]

I propose that we create different threads for the ppl who help out here so the main chat is not full :), it could be a good thing to have

[karmic crown]

how can you make a thread here? i do not see that option

[kindred tinsel]

no no we can't it would be a mod thing

[calm cairn]

if you want to branch off, just let me know where
until then, here are screenshots of:
-my original batch file (before I came here)
-the new batch file (modified with your help)
-the error message in context

[kindred tinsel]

aint there a missing - and a space in between vram and sdxl?

[karmic crown]

revert yours, you have older torch

it is good enough for sdxl on that gpu

[kindred tinsel]

@fair oxide

did you click there when installing from installer? I mean did you try installing from original python.org files or how did you installed python?

[karmic crown]

medvram = use medvram everywhere
medvram-sdxl = use medvram only for sdxl
so mine just make 1.5 faster

[kindred tinsel]

but it says unkown command .-.

[fair oxide]

I’ll try it when I get access to my laptop

[kindred tinsel]

:c okay

[karmic crown]

latest install was this
winget install -e --id Python.Python.3.10

it download installer from python.org and run it

you can try

[kindred tinsel]

what is that :0, run it from the python.org install

[karmic crown]

winget is magic

[kindred tinsel]

even stable diffusion github says, install from python.org

[karmic crown]

that cs1o and i both use 🙂

[calm cairn]

maybe we miscommunicated again, let me retry:
If you want this conversation to branch off into another location on the server, let me know
What does "revert" mean with regard to torch? I'm not sure I know how to do that

[kindred tinsel]

Literally download -> pip -> add to path -> next -> done

[karmic crown]

but he does not say others to use that 🙂

[karmic crown]

use your params that you had, mine does not work for you

@calm cairn to use mine, you need to upgrade pytorch

[calm cairn]

oh ok

[karmic crown]

winget install -e --id Python.Python.3.10
next
done
🙂

[calm cairn]

so I am not going to be able to use XL until I upgrade pytorch? Or I am, but not optimally?

[kindred tinsel]

no respect the developers of python.org they know best

jk idk

never used a winget thing, to be completely honest I dont work with python all that much to care about it, I use c++

[sonic ferry]

I put 'surprised' and got a sort of surprised face, trying it again. As I was waiting for my gen I was looking at other examples with someone using 'surprised' in every shot but it seemed more like it was 60/40 and gen I'm waiting on now is looking like it's just a stare now too, 50/50 not a good deal if I'ma be waiting an hour on my CPU gen. LOL

[karmic crown]

winget is microsoft reponse to package magers we have on linux

[kindred tinsel]

ew linux

[karmic crown]

instead of manually finding, downloading and runing install
winget thad that for you
but it is the same exe file from python.org

[kindred tinsel]

then it should've installed pip through the pip install advanced option, since you cant manage that inside of it there is no guarantee that that happened

or not idk :)

[shadow pagoda]

yes

[kindred tinsel]

In any case which mod should I talk to to ask for those threads for the helpers so each person can get their own separate help

[karmic crown]

@kindred tinsel just for you, installed on another pc

notice the link

it download the right fill and install it as it should, but something is wrong on his pc

he tried with manual uninstall that didnt work, he only had modify

he had to use winget uninstall --force to uninstall his old python

but as i said, i will not react to windows related messages anymore

i will be like a lora with triggerword MAC

[kindred tinsel]

LMAO

okay

[karmic crown]

this is too frustrating, i tried literally everything with manual install first, and than hoped that script will work but it didnt
same script worked on two windows machines without any issue

[kindred tinsel]

It should work in any machine

[fair oxide]

With or without spacing

[karmic crown]

python -m ensurepip

like this

[kindred tinsel]

lol it’s the windows defender

[fair oxide]

Am I doomed

[kindred tinsel]

I wouldn’t know for sure as I don’t have any experience on that regard

[fair oxide]

I want to create art

[kindred tinsel]

If it’s proving to difficult maybe use a different provider like collab

[dark owl]

It says the file contains a virus, you got the right files?

[fair oxide]

Yes

[dark owl]

Try right click run power shell as admin first?

Your PC is blocking the install because it thinks it’s unwanted

[fair oxide]

What's the most important part for stable diffusion gpu or cpu

[dark owl]

Gpu and it has to be nvidia or you’ll be a sad boy

[fair oxide]

Well that's why I bought an NVIDIA GeForce RTX 4090

[dark owl]

Second important hardware is RAM

[fair oxide]

Ok

[karmic crown]

@fair oxide do you have mcafee by any chance?

it is know for things like that

[fair oxide]

Yes

But it’s expired

[wind sun]

Helloooo

im baaack xd

[karmic crown]

please uninstall that, restart the computer, and try again with
python -m ensurepip

[wind sun]

so todays problem is coming from when i swap faces or generate images my computer is freezing the fu out of it and my C drive is getting red lighted until the proccess is over and i want to fix all that, idk if i will have to buy some new parts or idk but i have no idea

[karmic crown]

@ornate elk if i remember correctly all the guys with pip problems who actually added python to path was those with mcafee, right? at least since i came here

i guess you can add a note at a bigning of your guide, that mcafee couse all kind of errors and problems, and that python will probably not working as long as it is install, which leads to not worging a1111

[wind sun]

and everything is installed on my E drive (that is Ai) on the D i have games and on C its only my windows

[wind sun]

and im not with nvim and ssd

[karmic crown]

you musth have at least 15 GB free at C

even if your a1111 is on other drive

[wind sun]

whats the deal between a1111 and my C drive

[karmic crown]

its windows

page file (swap) needs space

[wind sun]

i mean ik but when im doing something from 30GB free on my C it goes sometimes to 1GB free

out of nowhere

[karmic crown]

when windows need more ram, it writed to c drive

so you need to have enough space

the less ram you have swap is bigger

https://www.techtarget.com/searchwindowsserver/definition/swap-file-swap-space-or-pagefile#:~:text=A swap file is a,up memory for other programs.

you can find more info here

[wind sun]

okay i will try to clear my C but can i also ask what will get better if i get more powerfull gpu

[karmic crown]

if you have like 16GB ram you still need swap file

[wind sun]

yup i have 16

[karmic crown]

read that when you have time

linux have a reserved space for that on a separate partition

but for windows it is usally c

it can be moved to other drive, but do not do that if your other drive is slower, for example from ssd to hdd

you whole windows will beacame slower in that case

use c drive wisely

[ornate elk]

Then you need to use a VPN for the installation of python and SD.
And uninstall MCaffee

[wind sun]

well my E disk is with nvme

[ornate elk]

Yea its mostly because if antivirus software. But he also maybe needs to use a VPN too. Most people fron Asia have install problems when not using one.

[wind sun]

so how can i see whats my motherboard

cuz my boxes are in my old apartment

[wind sun]

i mean is there other way from the bios

[reef patio]

I don't get why I am still recieving this RunTimeError

[wind sun]

@ornate elk can i ask you if 64ram is enough for sd without lagging my pc

[burnt badger]

ram seems to only affect size of image to be produced. I am running 32gb and doing fine

So this is the error I get when running control net. does this mean I have a bad control net base python file?

maybe im not using SDXL like i thought i was?

[ornate elk]

Yes 32 would be enough for everything on your PC

[ornate elk]

No the error says your using a controlnet version that isn't compatible with your model version

Sdxl and 1.5 are not compatible

[ornate elk]

What's your GPU?

[reef patio]

Terrible

Thought I could just override though

Suppose not

[unique trench]

A giant skull

[livid jetty]

So how do loras work?

[livid jetty]

BINGO THANK YOU

[hazy agate]

Greetings everyone.
Need some help, please.
Using revAnimated, having this problematic result every time.

[ornate elk]

What do you mean?

The commands you use are for CPU
Do you have a GPU?

[ornate elk]

What's the issue?
The quality?

[hazy agate]

Honestly, don't know how to describe.
Its just messy, not good.

[burnt badger]

So basically the checkpoint i am using determines which control net model I should use?

[ornate elk]

Try to use Euler a or DPM++ 2m SDE Karras

[ornate elk]

Yes

[hazy agate]

Yeah, i tried, but this problem persists only one revanimated model.

[ornate elk]

Ah okay then try to use hires fix to upscale and smoothing

[hazy agate]

Tried hires, esrgan 4, same result.

[proper pike]

Anyone know why my image won't come out more cartoony? This is img2img an anime in the aesthetic of attack on titan, man with a mustache Negative prompt: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, horn, realistic photo, extra eyes, huge eyes, 2girl, amputation, disconnected limbs Steps: 45, Sampler: DPM++ 2M SDE Karras, CFG scale: 6, Seed: 2244835904, Size: 1673x696, Model hash: 2f41de6769, Model: AoTStyle, VAE hash: 63aeecb90f, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Denoising strength: 1, ControlNet 0: "Module: none, Model: control_v11f1e_sd15_tile [a371b31b], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 1: "Module: none, Model: diff_control_sd15_temporalnet_fp16 [adc6bd97], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 2: "Module: softedge_pidinet, Model: control_v11p_sd15_softedge [a8575a2a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", Pad conds: True, Version: v1.6.0-298-g5e80d9ee

this is txt2img

If I have 2 controlnets enabled but I want to do an animation and have both controlnets working together do I put the directory into the Input Directory of both or just one? Will both controlnets work even if I only put in one "Input directory" under the Batch tab?

[topaz canyon]

Hi. How do I contact billing support for Dream Studio?

[ornate elk]

Can you show me?

[topaz canyon]

My Discord account got hacked 27th oct, and I had to make a new one. And I need someone from billing to transfer my account over to this one, if possible.

Had to do the same with Midjourney, Suno

[ornate elk]

Hey, please make a ticket in https://discord.com/channels/1002292111942635562/1010934719455707218

[ornate elk]

You need to input an image in boath

[topaz canyon]

Thanks, done.

[proper pike]

this will make them both work on a single image right?

[wind sun]

anddd im changing everything

4070, ryzen 9, 64gb ram, motherboard

[ornate elk]

Yes

[ornate elk]

Thats a good setup

[topaz canyon]

@ornate elk I don't expect an answer until maybe wednesday though. 😄

on my support ticket

no stress

[tight olive]

Any ideas on how to speed up generation time on an AMD Radeon RX 6800XT on Ubuntu? I'm running 8 steps. The LCM Lora slowed things down. Currently getting 40-50 seconds without Lora using: Steps: 8, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1403182942, Size: 512x512, Model hash: cc6cb27103, Model: v1-5-pruned-emaonly, Denoising strength: 0.6, Version: v1.7.0

[hallow latch]

What can I do with this error message?

[sonic ferry]

Should I be getting rid of 2 of 3 of these Detail Tweaker/Enhancer/Sliders? Should they all work the same way anyways?

[ornate elk]

You can try to add --opt-sub-quad-attention --opt-split-attention --upcast-sampling --no-half-vae to your webui-user.sh at the line COMMANDLINE_ARGS=
(The line needs to be uncommented)

[calm cairn]

ok I need someone who can solve this--might be hard or really easy, i have no idea

I have been using SD for months, and just started downloading 2.0 and XL models

How do I use them? It just...won't load

[sonic ferry]

I started downloading GIGABYTES of checkpoints (20GB unusable because of problem described ahead) , but I'm on a small office PC using CPU mode. SDXL will not work here, SDXL Turbo will, SD 1.5 works, SD 2.0 I am not sure about.

[calm cairn]

so we have the same problem? you load a new model and it just stays on the existing model?

I know I got a 2.0 to work once, but I had to add some configuration thingies and when I didn't load it with those, it stopped working--and I don't remember what they were

[sonic ferry]

Well, mine would just exit, no errors or nothing, just 'Press Pause to continue...' (as in exit).

https://tenor.com/view/peace-out-deuces-gif-11520099

SDXL Turbo has been working 98% of the time but has closed on me probably twice, I think because I wasn't using clip skip.

[ornate elk]

What's in your webui-user.bat ?

For sdxl models on 6gb GPUs you need:
--xformers --medvram --no-half

For 2.1 models you need an additional .yaml file that goes in the same folder as the model. It also need to have the same name.

[calm cairn]

ohh i remember doing the .yaml thing

"--xformers --medvram --no-half" right now

so if I add "--no-half-vae" it will work?

[ornate elk]

Nope, was a typo, --no-half is needed for gtx16xx cards

Sdxl models are 6gb in size.
You can try to switch to --lowvram to load them.
I would recommend not using them with your GPU. Its not worth the low performance you get with them.

--medvram is much faster

[calm cairn]

do I need to update anything? Python, torch, Automatic1111 etc

its been months without any update, maybe that would help

[ornate elk]

Yea if you never updated the webui before then that could help a bit

[ornate elk]

Check your versions at the bottom of the webui in browser

[calm cairn]

ok will check, one minute

python: 3.10.6  •  torch: 1.13.1+cu117  •  xformers: 0.0.16rc425  •  gradio: 3.16.2  •  commit: ea9bd9fc  •  checkpoint: 748ff6eab2

[ornate elk]

Oh yea thats old stuff

You need to run
Git pull in a cmd inside the stable-diffusion-webui folder to update.

Or you add
Git pull
In an empty line (before webui.bat) in the webui-user.bat

Then delete the venv folder and relaunch the webui-user.bat

I'm off now. If you get any problems I can help tomorrow again.

[calm cairn]

thanks will try

[ashen geyser]

Can someone help with this error when running .bat file on setup

[opaque spindle]

Does anyone know how to install background removal extension,
https://github.com/AUTOMATIC1111/stable-diffusion-webui-rembg in this tutorial,
https://stable-diffusion-art.com/remove-background/#Installing_the_background_removal_extension

The issue is after the installation and restart UI, the remove menu is not showing at all.

[rotund lava]

the remove bg menu is at extras tab > then scroll at bottom where it says remove background

[calm cairn]

that's all it took! it's slow but it works 🙂 ty for help

[spiral ermine]

Greetings and sanitations! Hope I'm asking in the right spot, I have a Radeon 5700XT 8gb and I run Windows, this seems to be a critically incompatible combination for all flavors and forks of SDXL.

Question 1; Is there something I've missed, or some cutting edge development that would allow me to run SDXL with my current GPU on Windows?

Question 2; Which sub $250 Nvidia GPU would be best for SDSL? Looking at 3060 12gb ($250), 2060 12gb ($200), or 2070 (sub $200).
Would like to spend as little as possible since I'd be buying an Nvidia card that's similar or slower in performance compared to the Radeon I already have

[rocky dragon]

Hi, a question- what is the 'correct' way to move a11111 gui to a new drive, along with my models and loras?

Can I just copy+paste or do I need to change any extra variables (path, locations) as well?

[opaque spindle]

I can't find the remove bg menu even after installing the extensions and restart UI. Any ideas how I can make it appears? Thanks.

[rotund lava]

extras tab not text2img tab

[opaque spindle]

This is extras tab 😦 but still not there

[rotund lava]

you are using a111 version 1.7.0.1 thats probably why,some extensions need to be updated to work with that version

[burnt badger]

anyone have a good tutorial on how to fix hands or change faces on already produced images?

[opaque spindle]

never mind. it works now after restart webui.sh Thanks very much.

[tight olive]

https://youtu.be/No1_sq-i_5U?si=21WhIwyRcElEq-pb

This shows changing the face

[fierce nimbus]

Can someone help me with automatic1111. The "Lora" shortcut, to the right of "Checkpoints" is missing. It was suggested that I should delete the "venv" folder, but I am using the new install which doesn't create a venv folder.

[burnt badger]

thank you

[fair oxide]

@karmic crown @kindred tinsel I was recommended to simply use
https://github.com/lllyasviel/Fooocus

[fair oxide]

But I will try this one last time when I get the chance

[fair oxide]

Wtf now there's a Rav endpoint protection

Okay I uninstalled mcafee, mcafee webadvisor, rav endpoint protection and rav endpoint vpn

Finally

What next @karmic crown