🤝｜tech-support

[elder mauve]

no same error

removed pip cache and venv , reinstalling

[ornate elk]

Maybe you need a PC restart again too

[elder mauve]

same 😦

[ornate elk]

Okay then do this:
go into the C:\ZLUDA\ Folder.
There copy the cublas.dll and the cusparse.dll and paste them inside it again.
Rename the copys to cublas64_11.dll and cusparse64_11.dll
Copy these two files to stable-diffusion-webui\venv\Lib\site-packages\torch\lib, and overwrite if asked.
then relaunch the webui-user.bat

[elder mauve]

where do i paste cusparse.dll and cublas.dll ?

[ornate elk]

Inside the zluda folder

Then rename them

And then copy them again to the other folder

[shell crystal]

Which cloud service would you guys recommend for automatic1111? Thinking of using paperspace

[ornate elk]

services like: runpod.io, paperspace, vast.ai are worth to checkout and compare

[elder mauve]

WARNING: ZLUDA works best with SD.Next. Please consider migrating to SD.Next.
Using ZLUDA in H:\SD-Zluda\stable-diffusion-webui-directml\.zluda
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug  1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Version: v1.9.3-amd-5-g674c5cac
Commit hash: 674c5cacdac3392279653be6b3da0c8e62898693
Skipping onnxruntime installation.
You are up to date with the most recent release.
no module 'xformers'. Processing without...
no module 'xformers'. Processing without...
No module 'xformers'. Proceeding without it.
H:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\pytorch_lightning\utilities\distributed.py:258: LightningDeprecationWarning: `pytorch_lightning.utilities.distributed.rank_zero_only` has been deprecated in v1.8.1 and will be removed in v2.0.0. You can import it from `pytorch_lightning.utilities` instead.
  rank_zero_deprecation(
Launching Web UI with arguments: --use-zluda --medvram-sdxl --update-check --skip-ort
Press any key to continue . . .

[formal trout]

I just updated my BIOS and now I'm agetting this, help

[gleaming copper]

hello, when i double click the webui-user.bat I keep getting the runtime error skip torch cuda test

[formal trout]

I'ma trying this

https://www.reddit.com/r/StableDiffusion/comments/z6nkh0/torch_is_not_able_to_use_gpu/?rdt=65485&onetap_auto=true&one_tap=true

[gleaming copper]

ok thanks i'll take a look

[waxen cradle]

Hi guys what monitor recoomend me for 4070ti súper 16gb?

[solemn jay]

i love controlnet. in SD.15 i can do super amazing stuff. Trying to migrate to SDXL and i cant get it to work. it seems it just ignores it. i have tried multiple models and nothing

dont know what im doing wrong

[formal trout]

Mmm that's my experience with openpose in SD XL (I don't remember results in SD 1.5). Most of the time doesn't produce the same pose, but you have a coherent image loosely based on it so you should be thankful

[ornate elk]

What's your GPU?

[solemn jay]

in sd 1.5 you can do magic

[formal trout]

I'm figuring it must work much better. In SD XL I just stop using it as it takes resources and no matter what it doesn't produce the exact pose, if you force it with more strength it produces deformed things. It kind of work for a general "idea" like your example.

[solemn jay]

this is very disappointing. hope it gets either fixed or SD3 improves uppon both

[waxen cradle]

Help

[formal trout]

I reinstalled venv but now I'm getting @ornate elk any idea 🙏

[open fox]

Are you on Linux?

If you you might need disable secure boot mode in bios.

[formal trout]

yes

[formal trout]

Oh yes I just had enabled that

[open fox]

Disable it and retry. It should work again.

[formal trout]

yes that was it thanks! what a champ

Yes I updated my BIOS and tried to set the secure boot because League of Legends wasn't working on Windows and I though it was Vanguard security

[ornate elk]

You should reinstall rocm then restart the PC

[solemn atlas]

hi, since the new update i cant find my negative embeddings

[dusty dragon]

animate

[gleaming copper]

I tried twice and i'm still getting the same error, i also tried updating my nvidia drivers

[formal trout]

No idea, just found that, but my problem had a different cause

[gleaming copper]

ok thanks

[ornate elk]

What's your GPU?

[gleaming copper]

2 rtx 3090 ti nvlinked

[ornate elk]

Okay and whats inside your webui-user.bat?

[gleaming copper]

How do I find that out .

[ornate elk]

Right click it and then edit

[ornate elk]

Hey thats a good GPU for WQHD resolution.
Like the ASUS TUF Gaming VG27AQ3A

You can also go on FullHD 240hz if you want

[gleaming copper]

@echo offal@echo off

set PYTHON="C:\Users\Admin\AppData\Local\Programs\Python\Python310\python.exe"
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=
git pull
call webui.bat

[ornate elk]

At the line commandline_args=
Add: --xformers
Then save.
Then open up a cmd and type ans run
Pip cache purge
Then delete the venv folder and relaunch the webui-user.bat

[gleaming copper]

ok one min

[elder mauve]

so u said not to install the pro driver, how do i not install the pro driver?

when i click on the download button it gives me this pro file

[gleaming copper]

ans is not recongnized as an internal or external cmd

[native kernel]

When I'm using num_images_per_prompt in the code below, what is being changed from image to image? The seed is manually set to 0, so what is being alterered to create different images with same prompt?:

`from diffusers import StableDiffusionXLPipeline
import torch

pipe_base = StableDiffusionXLPipeline.from_single_file(
"./models/stable-diffusion/sd_xl_base_1.0.safetensors",
torch_dtype=torch.float16,
use_safetensors=True,
variant="fp16",
)

pipe_base.to("cuda")

seed_gen = torch.Generator(device='cuda').manual_seed(0)

images = pipe_base(
prompt=prompt_positive,
negative_prompt=prompt_negative,
guidance_scale=5,
num_images_per_prompt = 2,
generator=seed_gen,
num_inference_steps=50).images`

Is it just auto-incrementing the seed / or randomly selecting different ones?

[gleaming copper]

it worked thank you soooo much

[ornate elk]

From the guide link?

[elder mauve]

yes

[ornate elk]

Ah okay yes thats the right one. But make sure when installing it the Pro driver at the bottom is not selected

[maiden coral]

this thing works, although openpose editor isn't perfect it still reproduces the exact thing

especially if you set it to the highest strength

it's needed for openpose editor

DWPose estimator

[formal trout]

I use Automatic1111, but ty

[maiden coral]

that's pity

[gentle sundial]

Hi @ornate elk Thanks for fixing my recent issue , I have another one tho, when I use sdxl it take 15 min to render an image and some times I get cuda error, I have the midvram-- thing in my webui script tho , any idea what is the problem? also I tried simple propt without anything , it takes same time

and the result is very bad unless I use auto for Vae

[ornate elk]

Hey, what was your GPU again?

[gentle sundial]

3050 rtx ti laptop

[ornate elk]

with 4gb vram?

[gentle sundial]

display memo is 4gb and shared is 8gb

[ornate elk]

okay thats noot good for sdxl

what you can try is to activate th FP8 mode for sdxl in the settings

[gentle sundial]

enable for sdxl or just enable?

and cache fp16 for lora should I check that?

[ornate elk]

only for sdxl

[ornate elk]

no

[gentle sundial]

also animdiff takes 30 min for 1 8 frame gif 512 * 512 is that because of my gpu?

*8 frame

[ornate elk]

yes animatediff is really vram hungry

[gentle sundial]

shoot, any alternative for that? like comfyui?

[ornate elk]

comfyui doesnt change the vram usage of animatediff

you would need to use other extensions for getting a video

like Loopback Wavescript

[gentle sundial]

Oh where should I find that?

[ornate elk]

Here: https://rentry.co/sd-loopback-wave
it goes in stable-diffusion-webui/scripts
then its under scripts in img2img

with that you can make stuff like this:

[gentle sundial]

Oh so no movement ?

[ornate elk]

yea not much, its more like morphing

[gentle sundial]

so sdxl is fast now thank you! but I get weird images like these and still blurry when it finish rendering

[ornate elk]

sdxl is trained on a 1024x1024 resolution, so dont use 512x512 for it

[gentle sundial]

Oh so I'm using hires , isn't that the same?

[ornate elk]

nope

1024x1024 needs to be the native resolution for sdxl

for 1.5 its 512x512

[gentle sundial]

ok thanks very much , I'll try it

[gentle sundial]

I'm getting this error when I made it 1024 x1024
RuntimeError: Expected weight to be a vector of size equal to the number of channels in input, but got weight of shape [640] and input of shape [2, 1600, 128, 128]

[ornate elk]

can you show the settins you used?

[gentle sundial]

here

[ornate elk]

oh you used negative embeddings that are made for 1.5

[gentle sundial]

sorry it worked , animdiff was activated

[ornate elk]

oh okay

[unique bluff]

Is VAE finishing supposed to take a long time to finish?

[ornate elk]

A few seconds

Sdxl VAE takes longer

Installing and using tiled vae extension can fix it

[unique bluff]

Mine is taking more than 20 min in the VAE finishing stage

(The image currently has an awful quality as well)

Oh, I have vae on automatic if it matters

I’ve been able to generate couple of images in the past now it just won’t progress past this stage (I’ve tried using other images sizes as well)

idk what to do

(Now that I was finally getting hang of the prompts I need)

[ornate elk]

what does the terminal shows?

[wet flame]

just use windows snipping tool

who takes pics with their camera of their monitor anymore

doesn't mac have print screen xD

btw AI generators don't have any issue with AMD cpu do they?

its only the gpu right

amd cpu and gpu is much cheaper and performs similar in most apps

nvidia and their expensive price tags 😦

[ornate elk]

Nope, AMD CPUs work totally fine

[unique bluff]

Yes, but my computer is almost at 100% usage (I ASSUME) when sd is opened

If I try to screenshot it, there’s a high chance the computer gives up and dies (quite literally restarting itself)

[unique bluff]

19:53:13-936228 INFO Startup time: 2069.89 torch=225.93 onnx=6.30 olive=2.72 gradio=62.68 libraries=95.60 1dm=0.56 samplers=0.96 extensions=16.67 models=0.32 face-restore=70.15 upscalers=0.96 networks=0.23 ui-en=0.35 ui-txt2img=0.14 ui-img2img=0.11 ui-control=0.18 ui-extras=0.11 ui-models=0.07 ui-settings=1.56 ui-extensions=4.11 ui-defaults=0.13 launch=1.05 api=0.39 app-started=0.49 checkpoint=1578.05 
19:53:29-533774 INFO Base: class=StableDiffusionPipeline
Progress 14.74s/it |||||||||||||||||||| 100% 20/20 04:54 00:00 Base```

[zenith sail]

Any ideas?

[onyx solar]

You need to look at your console for details.

[zenith sail]

Update custom node 'UltimateSDUpscale'
Update: ['https://github.com/ssitu/ComfyUI_UltimateSDUpscale']
CUSTOM NODE PULL: None
Update was successful.```

[onyx solar]

That might be the update, but it didn't load. So restart and watch the console for when it's loading.

[zenith sail]

No errors either

[onyx solar]

Way too generic of a node name to know. Probably should just delete the node and replace it manually.

[limber pier]

yeah that's pretty bad lol

"Text"

if i ever release a node pack, i'm gonna have a node just called "Node"

[zenith sail]

I spent months on making that workflow...

I can't figure it out.

Thanks lads

[austere gull]

why doesn't control net work? first time using it and watched a tutorial but ion get it ( using webui-forge )

[ancient wadi]

my image generation seem to be slowing down over time, anyone know what might be causing this

well mabtye

how can ifix that?

wheres that?

in cmd?

sorry im new

thanks man

[elder mauve]

There is no option?

[ancient wadi]

no idea if i did this right

ok so would that be the webui python?

alright il give that a shot

thanks a ton

[fallen quest]

That's correct, yes. 🤣

[forest zealot]

https://github.com/Akegarasu/lora-scripts

dunno how to install dis

ran install.ps file but its not working

[civic remnant]

Hello all! Why does ControlNet not working? I press preview and nothing happens

And I'll appreciate if somebody helps me with videofromimage workflow) I installed "ComfyUI-I2VGenXL" custom nodes, but when I start videofromimage workflow, SD doesn't recognise these nodes and manager doesn't see them neither. What am I doing wrong?

[ornate elk]

Then its okay, for some it isnt there

[ornate elk]

Wrong openpose model

[austere gull]

idk i used it for open pose too and it didnt work, i just switched to dw since i read on reddit that it works

this is where i got it

https://huggingface.co/TencentARC/t2i-adapter-openpose-sdxl-1.0/tree/main

i just download the safesensor and put it in models controlnet no?

[ornate elk]

You need to download the correct openpose models:
If your on sdxl you find all supported here:
https://huggingface.co/lllyasviel/sd_control_collection/tree/main

[austere gull]

huh why do they have different names

it seems like this is the one i downloaded but they have the correct name instead

[ornate elk]

Because they are made from different research groups

[austere gull]

https://huggingface.co/lllyasviel/sd_control_collection/blob/main/t2i-adapter_xl_openpose.safetensors

[ornate elk]

Yes but thats a t2i adapter

[austere gull]

theres multiple t2i-adapter?? didn't know

does it not work

which ones in there actually work

for controlnet atleast

im just gonna install the largest file one and assume the big number makes it good

[ornate elk]

The 700mb file at the bottom

[austere gull]

ok epic ty

[civic remnant]

Why this error may appear?

[jaunty quarry]

I have a problem with stable diffusion:

and I can't find anything to solve the problem

[bleak plover]

does anyone here use webui Forge? Any advice on how to fix the current bug with pydantic https://github.com/lllyasviel/stable-diffusion-webui-forge/issues/674
GitHub

[ornate elk]

Hey, what's your GPU and whats inside your webui-user.bat?

[ornate elk]

Go into the stable-diffusion-webui folder
Click in the explorer bar (not searchbar)
There type cmd and hit enter.
In the cmd type and run these 3 commands after each other.

venv\Scripts\activate
pip install albumentations==1.4.3
pip install pydantic==1.10.15

[jaunty quarry]

the file is mostly empty:

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

call webui.bat

usualy I just start sd with webui.bat

[ornate elk]

Okay at the line commandline_args=
add: --xformers
Then save and relaunch webui-user.bat

With that it generates faster and uses less vram

[ornate elk]

Never start with only webui.bat

[jaunty quarry]

It has problems to start xformers:

Error running preload() for B:\stable-diffusion\stable-diffusion-webui\extensions\adetailer-23.10.1\preload.py
Traceback (most recent call last):
  File "B:\stable-diffusion\stable-diffusion-webui\modules\script_loading.py", line 28, in preload_extensions
    module.preload(parser)
  File "B:\stable-diffusion\stable-diffusion-webui\extensions\adetailer-23.10.1\preload.py", line 5, in preload
    parser.add_argument(
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\argparse.py", line 1448, in add_argument
    return self._add_action(action)
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\argparse.py", line 1814, in _add_action
    self._optionals._add_action(action)
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\argparse.py", line 1650, in _add_action
    action = super(_ArgumentGroup, self)._add_action(action)
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\argparse.py", line 1462, in _add_action
    self._check_conflict(action)
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\argparse.py", line 1599, in _check_conflict
    conflict_handler(action, confl_optionals)
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\argparse.py", line 1608, in _handle_conflict_error
    raise ArgumentError(action, message % conflict_string)
argparse.ArgumentError: argument --ad-no-huggingface: conflicting option string: --ad-no-huggingface```

    PyTorch 1.13.1+cu117 with CUDA 1107 (you have 1.12.1+cu113)
    Python  3.10.9 (you have 3.10.11)
  Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers)
  Memory-efficient attention, SwiGLU, sparse and more won't be available`

and when I try to reinstall torch:

Looking in indexes: https://pypi.org/simple, https://download.pytorch.org/whl/cu117
ERROR: Could not find a version that satisfies the requirement torch==1.13.1+cu114 (from versions: 1.11.0, 1.12.0, 1.12.1, 1.13.0, 1.13.0+cu117, 1.13.1, 1.13.1+cu117, 2.0.0, 2.0.0+cu117, 2.0.1, 2.0.1+cu117, 2.1.0, 2.1.1, 2.1.2, 2.2.0, 2.2.1, 2.2.2, 2.3.0)
ERROR: No matching distribution found for torch==1.13.1+cu114
Traceback (most recent call last):
  File "B:\stable-diffusion\stable-diffusion-webui\launch.py", line 360, in <module>
    prepare_environment()
  File "B:\stable-diffusion\stable-diffusion-webui\launch.py", line 269, in prepare_environment
    run(f'"{python}" -m {torch_command}', "Installing torch and torchvision", "Couldn't install torch", live=True)
  File "B:\stable-diffusion\stable-diffusion-webui\launch.py", line 89, in run
    raise RuntimeError(f"""{errdesc or 'Error running command'}.
RuntimeError: Couldn't install torch.
Command: "B:\stable-diffusion\stable-diffusion-webui\venv\Scripts\python.exe" -m pip install torch==1.13.1+cu114 torchvision==0.14.1+cu117 --extra-index-url https://download.pytorch.org/whl/cu117
Error code: 1

[mortal compass]

Hey @ornate elk, you don't happen to know of some good upscaling guides for controlnet? Everything I find is like a year old which is super out dated

[ornate elk]

You need to open up a cmd then run
Pip cache purge
Then delete the venv folder
And remove the --reinstall-torch command.
Only add --xformers
Then relaunch the webui-user.bat

[heavy perch]

@ornate elk any idea why I might be getting a grey video?

[jaunty quarry]

did not change that:

ERROR: No matching distribution found for torch==1.13.1+cu114

[turbid elk]

try reinstalling python the latest version 3.12 didnt work for me

[jaunty quarry]

I have like 5 different versions on python

[turbid elk]

so i installed 3.10.6 and it worked

[turbid elk]

was the same case with me uninstalled all of them

and installed a fresh one

took me almost 2 hours to crack this error lol also dont forget to add the python path while installing

[chilly knot]

Applied providers: ['CPUExecutionProvider'], with options: {'CPUExecutionProvider': {}}

its bad?

correct if show cuda?

[ornate elk]

What's your python version?

You need 3.10.11 64bit

[ornate elk]

Is that with mov2mov?

[ornate elk]

Yes, I can show you one later

[heavy perch]

yeah

[jaunty quarry]

it's still not able to download torch:

ERROR: No matching distribution found for torch==1.13.1+cu114```

[viral knot]

can you install another version? cause 1.13.1 has no match with cuda 11.4 as seen from the error

[ornate elk]

Can you show the complete cmd log?

[ornate elk]

You also need to delete the venv folder after changing python versions

[elder mauve]

but why doesnt it work 😦

[ancient rune]

Hi, for some request (like "une patate" for example), I recieve an empty response.
( litteraly {"artifacts":[]} )
When I try with another prompt, it works but this specific prompt and some other just never work.
(I use the API)

Why and how to fix it ?

Here is the response for "une patate"

[ornate elk]

What you can try first is to open up a cmd and type
SFC /scannow
After thats done restart the PC.
Then if it still doesn't work, move your ZLUDA folder into the C:\Programm Files folder.
Then update the zluda path in the system variables.
Sometimes windows 11 makes path problems

[slim basin]

Hi, can anybody help me to install SD?
Windows 10
AMD 7900 XTX
miniconda3 3.10.11 64
Trying since 3 days, did't getting to run

[ornate elk]

Hey, sure I'm on a 7900xtx too.
First uninstall miniconda.
Then follow my Zluda auto1111 install guide, its in the pinned messages of this channel

[slim basin]

i try it acn i DM you if i stuck ?

[ornate elk]

Sure

[naive dragon]

so i still have the thing going on with my hdd being used 100% so i might as well put it on my ssd lol and it still loses connection.
can i just copy paste it to my ssd or would that not work or is it justbetter overall to reinstall it over again on a new ssd.
(because im pretty sure its the fault of my hdd that cant handle the speeds ;/ but that rly that big of an issue bcsi cna put it on my ssd)

[ornate elk]

If you move it, then you can't update it anymore

[elder mauve]

what are control nets used for?

[naive dragon]

thank you for that answer ill install it over again, also how do i update it bcs it doesnt let me type in the beginning when it tell me there is a new version

[elder mauve]

this means good?

[ornate elk]

Yes now restart the PC

[ornate elk]

By adding git pull to an empty row in the webui-user.bat

[naive dragon]

i put it in here? (can i jsut keep it there so it automatically updates) and do i put it afterthe COMMANDLINE_ARGS?

[ornate elk]

It needs to go in the empty line

[pine smelt]

So i am facing a problem when i try to use image to vid models . (file attached: full cmd ). the problem is that after running this it takes 1.5 hours! and then error massage shows up and it stops . However i able to generate image but this error message also popup there { UserWarning: 1Torch was not compiled with flash attention.} but at least it gives me a output . please help and thanks for checking out

[chilly knot]

its try download model

[pine smelt]

expand the code and you will see the problems

[chilly knot]

Applied providers: ['CPUExecutionProvider'], with options: {'CPUExecutionProvider': {}}

what could the message be related to?

what is running on my cpu now

insightface/models/buffalo_l/w600k_r50.onnx recognition

its lke that

its normal or bad

[ornate elk]

@elder mauve did it worked?

[elder mauve]

no 😦

same error

[ornate elk]

can you delete the venv folder again, then relaunc the wbeui-user.bat
then check the taskmanager ram usage

@elder mauve oh wait there was a fix yesterday to the webui. (regarding zluda installation)
what you should try is to delete the stable-diffusion-webui-directml folder.
then do the steps 2 - 5 from my guide and relaunch

[elder mauve]

do i also delete zluda folder?

[ornate elk]

nope

[elder mauve]

and what about rocm?

[ornate elk]

nope

[elder mauve]

alr

[heavy perch]

@ornate elk

[ornate elk]

can you show more settings?

[ornate elk]

try with 30 fps

also whats your gpu? and whats inside your webui-user.bat?

[elder mauve]

same

[ornate elk]

Can you show me the zluda folder and the path again?

[ornate elk]

Any third party antivirus Programm running?

[elder mauve]

nah

none

even windows defender is off

[ornate elk]

Can you show me the full screenshot?

Of the cmd

[ornate elk]

okay open up a cmd and type
where python

[elder mauve]

C:\Users\sixsi\AppData\Local\Programs\Python\Python310\python.exe
C:\Users\sixsi\AppData\Local\Programs\Python\Python312\python.exe
C:\Users\sixsi\AppData\Local\Microsoft\WindowsApps\python.exe

[ornate elk]

do you need python 3.12 for anything?

[elder mauve]

not really

[ornate elk]

okay, then uninstall (not deleting) everything from python

[elder mauve]

everything?

[ornate elk]

yes python 3.10.6 and python 3.12 and the python launcher

[elder mauve]

and then i reinstall?

[ornate elk]

then download python 3.10.11 64bit (link is in my guide)
install it and check "add python to path" at the first window
then delete the venv folder

[cosmic furnace]

76 gpu temp while hotspot 91 is okay or nah ? hmgm

[ornate elk]

76gp temp is not good, what are you doing?

[cosmic furnace]

playing at ultras xD

[ornate elk]

what game?

[cosmic furnace]

and gpu usage 100% kinda

bellwright) ue5 & bad optimisation -_-

[ornate elk]

on 4k ?

with rtx?

[ornate elk]

?

[cosmic furnace]

nope)

and its in silent mode =\

lil bit higher than shud probab ?

or fine?

others seems to have lower hm

[ornate elk]

its too high

@cosmic furnace when im playing Cyberpunk 2077 on WQHD with Max settings (RTX off)
im getting like 65 degree max and Hotspot is on 72

you never should be anywhere near over 80 with hotspot

[cosmic furnace]

ill look into it ... mb cuz silent ? 🤨

[ornate elk]

yes

dont do that when playing games

also adjust your Fan curve

[cosmic furnace]

i see some ppl having that hotspot

[ornate elk]

here is my Fan curve

[cosmic furnace]

is it smthing that shud be disturbing or can be ignored?)

why gap between gpu temp and hotspot is so big? about 13-15 degrees

[ornate elk]

if you want to have your gpu for a long time then you should do something about it ^^
first thing is to cap the FPS in the game to not let your GPU run on 99%

then set the gpu to normal mode, and then adjust the fan curve

90 hotspot is very bad on the long run

[ornate elk]

@mortal compass
Here are my settings for Upscaling images in img2img with controlnet: (getting much more details)

[round ledge]

hey coming back to stable diffusion for the first time in a while, I just fresh installed everything but its only recognizing 2 out of my 5 loras, is there any reason for this?

[ornate elk]

hey, loras only show up when loadin a compatbiel model

sdxl loras work only with sdxl models
same goes for 1.5

[round ledge]

I get what you're saying but how would I get these loras to work now

I tried multiple checkpoints and nothing works

[ornate elk]

change the checkpoint to an 1.5 based model, then reload the loras

[round ledge]

do you have an example of a good 1.5 based model, I thought abyssorange was tbh

[ornate elk]

abyssorange is 1.5 yes

divineanimemix is a good 1.5 one

[round ledge]

ty

I'll give it a go

so just for future reference them loras will only work on 1.5 checkpoints

I noticed this pony checkpoint is really popular now

but none of my loras work there

[ornate elk]

the 2 loras of your screenshot are pony and sdxl based

the others idk

but mostly 1.5

[barren spoke]

hello! I have an application which creates an image using stable diffusion XL pipeline_text2image; then iterates over it using pipeline_image2image.

Things had been working fine for weeks but as of last night I'm now getting an error when using pipeline_image2image, it seems like it is not accepting the previous input which was a PIL image and am now getting this error:

cannot reshape tensor of 0 elements into shape [0, -1, 1, 512] because the unspecified dimension size -1 can be any value and is ambiguous

anybody else have this issue?

[upper linden]

All of a Sudden my Zula Amd Stable Diffusion isn't working

[ornate elk]

It seems it tried to use your amd CPUs gpu instead of your gpu

You need to deactivate it in the device manager
Also you should update your webui-user with git pull

[sage vessel]

Im new to ai art, can anyone help me how do i get into the webui using automatic1111 and google colab
I've paid monthly subscription for the colab but i still can't find a beginner friendly ipynb file, if you could help me i'll be so grateful

I keep getting error no module pytorch lightning

[upper linden]

how would i do tthat?

this?

[sage vessel]

I have the same gpu series, and i still don't know how to use my gpu instead of cpu 😂😂😂

[naive dragon]

i have installed it on my ssd because i tough it might have been my hdd but i still get the same error maybe i can use some help 😅 (hope i got some good cmd screenshots)

[sage vessel]

That's why my only option is colab probably

[maiden coral]

I think @ornate elk had a solution to increase filepage capacity on Windows and force models into f16 mode instead of f32 and ifyou have low VRAM and RAM previously.

I had the exact same "errored out" error but didn't know it was because of my low notebook specs

[vocal burrow]

Models load in fp16 (or something similar) by default and as for pagefile https://www.thewindowsclub.com/increase-page-file-size-virtual-memory-windows but that's probably not the problem here and we'd need to see the full log (copy paste it into a .txt and drop said .txt in here)

[maiden coral]

hmm... ok, it's just that he found this solution for this and I had the same error

[vocal burrow]

full log ?

[maiden coral]

it just happened suddenly, no warning prior, no settings changed, but it has to do with SDXL models

[jovial wadi]

Anyone know how to make the faces better?

Honestly not that good face wise lol

[nova wren]

anyone knows why when using hires fix with tiled diffusion and region prompt control, hires does not follow the regional prompts?

[keen geyser]

I am trying to install a style but it says I need PonyDiffusionXL. How do I use two models at once?

[young bridge]

hello

how to install forgeUi properly>

update.bat then run.bat?

[ornate elk]

Then you don't need Colab.
You just need to follow my install guide, its in the pinned messages of this channel

[ornate elk]

Right click on it and select disable

[ornate elk]

Thats maybe a RAM error, how much do you have? And you should recheck the windows pagefile setting. Also what was your GPU again?

[sage vessel]

i specifically have rx 6500 xt gpu, is it compatible?

[young bridge]

TypeError: 'NoneType' object is not iterable i am getting this error

and when I copy generation data from civitai jugg model, it is generating a low ress image

[ornate elk]

Ohh thats the only card thats not supported rn :/
It may could work with some adjustments but I can't guarantee it

[sage vessel]

there no forum that discuss about the same problem anywhere on the internet but i've found a japanese dude that can actually work with it

https://www.youtube.com/watch?v=ZXpgOM0m3DU&t=56s

unfortunately he didn't give a clue on how to do it in the video

or am i just stupid to do it myself

[ornate elk]

There are two ways for AMD to get it running on windows.
The directml way = very slow and vram limited.
The zluda way = faster but needs rocm (which your card is not officially supported) but with workarounds could work

[viral knot]

why is amd struggling with ai stuff, im so happy to be with nvidia lol

[young bridge]

How can I fix my errors?

[viral knot]

you can fix errors by switching to comfy 🙂

[terse lichen]

yes this is what i tried to do i think or direct install the code into the comfyui repository

[open fox]

So you were in the custom node directory and used git clone to get all the files from the custom node repository.

And then you run python install.py?

Or „pip install -e .[cuda]“ ?

Double post. You posted the same 1 min. ago in #🏞｜general-with-images

[dry swallow]

Ok

[terse lichen]

more like i installed it directly and put it in the clone repository and when i am updating the manager it is saying inference core failed

[terse lichen]

yes

also where should i add the animatediff v3 like folder

[open fox]

Did any errors appeared during the installation?

[terse lichen]

no errors popped up only after updating it said inference core failed

[open fox]

I guess you used animate diff evolved node. Then let me recite their GitHub:

Place models in one of these locations (you can rename models if you wish):
ComfyUI/custom_nodes/ComfyUI-AnimateDiff-Evolved/models
ComfyUI/models/animatediff_models

[open fox]

And during the start of confyui you do not get any error report (code etc. ) just the import failed

[terse lichen]

yes

[open fox]

You are using windows and the portable version of comfyui right?

And python 3.10 or which version.

[terse lichen]

yes

[terse lichen]

i am using python 3.12 version

[open fox]

I found this:

if you want install this with python3.12 for now [2024.4.10]. just try:

1. download mediapipe-0.10.11-cp311-cp311-win_amd64.whl from pypi site

2. rename it to mediapipe-0.10.11-cp312-cp312-win_amd64.whl

3. pip install

4. edit custom_nodes\ComfyUI-Inference-Core-Nodes\pyproject.toml line 9: requires-python = ">=3.10,<=3.13"

5. run python install.py
   enjoy...

https://github.com/LykosAI/ComfyUI-Inference-Core-Nodes/issues/12#issuecomment-2051199606

[terse lichen]

thanks will try this

https://civitai.com/models/372584/ipivs-morph-img2vid-animatediff-lcm-hyper-sd

[terse lichen]

i have this model downloaded and do i need to add it to the animatediff model folder or checkpoints

@open fox so a lot of these nodes depends on proper installation right

[open fox]

This is a comfyui workflow and it uses some nodes. These nodes needs dependencies. Python packages, models,… on the page of this workflow you find informations about which models you need and where to put them.

[vast apex]

does someone know how i can inpaint using comfyui/stability matrix?

[terse lichen]

cant seem to understand what all nodes are needed got the ip adapter animatediff v3 but now not sure where to place the model i downloaded

[open fox]

If you copy a workflow from another person you are confronted with the nodes that are used to get the result. Best case only public available nodes are used and you can simply install them with the comfyui manager „install missing“. But still some nodes will require manual steps to work. Google the node, visit the according GitHub and read the requirements. One node after another. Just read and learn.

[terse lichen]

i am trying to figure this out but not sure whats going wrong

[vast apex]

does anyone know about the differences from comfyui and stability matrix?

[ornate elk]

Yes, comfyui is a webui that uses stable-diffusion to generate images.
Stability Matrix is an installer for webuis like comfyui or Auto1111

[vast apex]

Error starting the server: 'Namespace' object has no attribute 'tls_keyfile'

anyone know what that means :,)? cant generate or find webui link

[sinful needle]

Questionnnnn: Is it possible to use animatediff for SDXL with Automatic1111 or is it only possible with ComfyUI?

[ornate elk]

Yes it works with sdxl too in Auto1111

You need the animatediff sdxl bets model

[sinful needle]

I used the SDXL model and it only redirects to a still image. https://civitai.com/models/108836/animatediff-motion-modules

I used that model and placed it in the Animatediff/models folder.

AnimateDiff works for 1.5 models for me, just not this.

[ornate elk]

What does the cmd shows when you try to use it?

Full log pls

[sinful needle]

Full from boot to end?

Reads that there is no motion module for SDXL, yet I've redownloaded that model various times.

[ornate elk]

Ah well, it gets blocked by the webui because its a .ckpt
You have to convert it to an .safetensor file.
An can tell you how

[sinful needle]

ah

[ornate elk]

Here is a quick explanation:
#🤝｜tech-support message

[ornate elk]

Also what was your GPU again?

I see --no-half --precision full in your log.

[stark mulch]

can i get some help

everytime i use one and it has a error it deletes one of my loras and only lets me use it on another model how do i fix it to where i can use it on that model again

nvm i figured it out had to do with sdls

why does that happen?

[ornate elk]

Sdls?

What error do you get in the cmd?

[naive dragon]

i have 16gb of ram and rx 6700 xt gpu

[naive dragon]

i see now that my ram speed got readjusted back to the default speed and not its full speed(idk why it got reset maybe thats what causing it?)

[naive dragon]

i put it back to its speed now

[sinful needle]

givin it a spin! thanks lets see

[sinful needle]

3090 RTX. im not sure why i require it but A1111 shits itself without nohalf. i should be fine without it, but no

[ornate elk]

You should remove both as these slow down your generating time tremendously! And increase your vram usage
Better add --no-half-vae and your fine

[sinful needle]

okiedokie

i tried the SDXL animatediff and it didnt error out but god damn did it slug my PC. regular animatediff doesnt slow me at all. didnt make it past 0%

[ornate elk]

Yes because you had --no-half --precision full in there

[sinful needle]

alrighty. removing

[slow saffron]

The classic and safe way to earn up to €900 a day online only requires that you get a Smartphone or Laptop, for more information about the offer write me a direct message now

[sinful needle]

some progress but now: NansException: A tensor with all NaNs was produced in VAE. Use --disable-nan-check commandline argument to disable this check.

[ornate elk]

Please no scam like advertising here. Please remove your message

[ornate elk]

Have you added --no-half-vae ?

[sinful needle]

yea

[ornate elk]

Okay, whats the model? Lora's and embeddings?

They all need to be for the same base SD version

[stark mulch]

Idk but how come some sdl 1 and sdxl don’t work together for Lora’s?

[sinful needle]

Model: https://civitai.com/models/143043/starlight-xl-animated?modelVersionId=182077
Lora: https://civitai.com/models/267667/16-bit-pixel-backgrounds-v2
Embeddings: none

[ornate elk]

Okay looks right.
Can you give me the cmd log again?

No wait

What VAE did you used?

[sinful needle]

none associated. none were part of it either.

[ornate elk]

They are trained differently and on an different resolution too

[ornate elk]

Then try to use the sdxl fp16 VAE fix vae

[sinful needle]

where to get it?

[ornate elk]

Here:
https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/tree/main
Download the sdxl.vae.safetensor

[sinful needle]

im guessing use the one second to the bottom of the list?

[ornate elk]

Also if you haven't already do this:
go into Settings, User Interface, Quicksettings, there add sd_vae
Then hit apply and reload ui.
Then you have a VAE dropdown menu

[sinful needle]

ur no damn coder

ur a wizard

get that patreon started dude u need support

[ornate elk]

I made a Kofi account a few days ago ^^

[sinful needle]

ill have to make an account

[ornate elk]

Idk if others need an account for it

[sinful needle]

drop the link ill tip a little for now

[ornate elk]

The link is now in my Discord bio

Thank you! Appreciate it 🙂

Thank you! Will check when I'm at home

[river forge]

Hello. I would like some help getting auto1111 + zluda running with my 7900xtx.
It currently fails at this stage:
File "D:\SD-Zluda\stable-diffusion-webui-directml\venv\Lib\site-packages\torchvision_meta_registrations.py", line 4, in <module>
import torch._custom_ops
ModuleNotFoundError: No module named 'torch._custom_ops'
I thought it my be related to using Python 3.11 vs 3.10?

[ornate elk]

Hey, yes python 3.11 is not supported

Uninstall 3.11, install 3.10.11 64bit
Then delete the venv folder

[river forge]

Thanks, mate.

[ornate elk]

Np, I have a 7900xtx too, works perfectly with Auto1111 on Zluda

[timid notch]

hi i wanna generate a longer video, which node should i use right now? can some give me a hint

the context options node is not affect anything~~

i find it

[fathom shuttle]

yo i just installed everything correctly (i think), but when i wanna generate is gives a queue or something. im a newb

[ornate elk]

Hey, what does the cmd shows?

[fathom shuttle]

what do you mean?

[ornate elk]

Did you installed Automatic1111 webui?

[fathom shuttle]

no

was gonna do that after

[sinful needle]

everyone make sure to support CS1o on his kofi since hes taking so much time out of the day to help

[ornate elk]

What did you setup now?

[fathom shuttle]

ye i feel bad for him having to deal with dumb me xd

i did this

and copied the url

idk just followed a tutorial

[ornate elk]

What's your GPU?

[fathom shuttle]

1050 ti

4gb

._.

[ornate elk]

Okay then close the cmd.
Go into the stable-diffusion-webui folder.
There edit the webui-user.bat,
at the line commandline_args=
you add: --xformers --medvram --no-half-vae
Then save and relaunch the webui-user.bat

[fathom shuttle]

like this?

alr its running

thx boss

im completely new with this all but automatic1111 and stable diffusion are 2 different things no?

[ornate elk]

Np, when you get new models from Civitai.com, go for the 2gb ones. They work good on your GPU

[ornate elk]

Automatic1111 is the webui you installed right now.
Stable diffusion is a process to generate images.
Models (checkpoint's) are needed for image generation.
There are community models.
These are based on 1.5 or sdxl (the two base stable diffusion model versions)

[fathom shuttle]

ah so i already have automatic1111

cool

do i keep the sampling method on DPM++ 2M?

[ornate elk]

Its a good and fast sampler. You can test others too.
For example Euler a is a good one too.
DPM++ sde also but slower

[vernal robin]

Hi all, I’m getting a laptop to do sd work remotely from my employer, and I’ve been using PC to do my SD work at my previous job for about a year, is there any reason to choose mac vs PC? I’m comfortable with both OS but just havent used Mac for this yet, would appreciate any answers 🙏🏼

[river forge]

Running SD locally??

[open fox]

Depending are you paid by hour or by image 🙂
Most notebooks with a good nvidia card will be faster then a Mac.

[ornate elk]

If you generate locally on your PC, then you want a dedicated Nvidia GPU with much VRAM.
If you only connect to your works PC then you can use any Laptop

[terse lichen]

@open fox can you pls tell whats causing this error

[heady pine]

Is there a way to remove the box that is formed from enhancing the face?

[ornate elk]

The box? Is it darker or brighter than the image?

Make sure your using a VAE file

[heady pine]

Sorry for the not showing a picture first, this thing, I don't usually use VAE files, but will doing so remove this box?

[ornate elk]

yes it should remove it

[heady pine]

Alright, thank you. I appreciate your help.

[ornate elk]

no problem

[young bridge]

ValueError: images do not match

hello I am getting this error

[fathom shuttle]

I think they don’t match

[young bridge]

what do you mean

[fathom shuttle]

idk was just playing

[sinful needle]

so... now that AnimateDiff is animating. How do i control it better so that it doesn't animate lightposts lmaooooo?

[terse lichen]

i am in need of your help bro @sinful needle

[sinful needle]

Oh and I don't use comfy ui. lol

[young bridge]

What is LDSR upscale and why is it slow

[terse lichen]

how to fix this error @maiden coral

[terse lichen]

i see thanks for the reply

[maiden coral]

Is it an error while loading ComfyUI?

What is your python version? Are you using AMD or Nvidia GPU? Have you gone through the steps on github on the installation of ComfyUI? What are you specs? (your GPU with VRAM, amount of RAM, CPU)

also the workflow what you're trying to do

Also, are you using multiple GPUs?

[terse lichen]

its an error after prompt is qued and i am using a quadro p4000 and ram is 32 gb and cpu is intel xeon

https://civitai.com/models/372584?modelVersionId=469548

[maiden coral]

Can you show me the workflow? just a screenshot is fine, I build my own and with the help of the internet

[terse lichen]

i am using this workflow

[maiden coral]

I can't help you with img to vid, sorry.

Try a tutorial on youtube which goes step by step. You might be missing a node or something.

[terse lichen]

this is the workflow @maiden coral

[maiden coral]

Have you installed all the requirements?

You can't install this one through manager.

[terse lichen]

yes i have installed all the requirements

[maiden coral]

probably tensors aer allocated to GPU and CPU and not on one device

[terse lichen]

i am getting this error

[fathom shuttle]

quick question do you always launch sd through files or is there another 'better' way

[maiden coral]

Have you tried other SD models?

[terse lichen]

this model only accepts sd 1.5 models

[maiden coral]

I know but other variations of SD model

there are models based on SD 1.5

they count as SD models too, 2nd category are SDXL models which were trained in higher resolution 1024x1024 (SDXL) instead of 512x512 (SD1.5)

[terse lichen]

i have majority sdxl models

[maiden coral]

When did you update Comfy?

or when was the last time you update comfy

I heard there are a lot of compatibility issues with AnimateDiff

[fathom shuttle]

i downloaded a model but after refreshing it doesnt show as an option, any tips?

[open fox]

As far as I can read/tell from the error log is that most comfyui nodes use your GPU but at least one is using the CPU instead. For the sampling it is needed that all tensors are on the same device.

[open fox]

Is the name of the model correct (.pth, .ckpt or .safetensor? Is the file readable for the user who executes your UI?)

[fathom shuttle]

they said u have to save as in your model folder

[open fox]

The last one looks like a link the first one like a zip folder?

[maiden coral]

you have to restart the server through comfyui manager or close comfyUI completely and start it up again. Probably for models refresh should be enough, but try it.

[fathom shuttle]

the second is the one i already had

[maiden coral]

it must be in model/checkpoints folder

[fathom shuttle]

how do i do that

[open fox]

He is in this folder. But both actual files are not a safetensor but instead a directory and a link.

Please show size and file ending of the first and last file

Don’t need to restart. A refresh should work.

[maiden coral]

ckpt I see

[open fox]

As mentioned no model there. The link does not help. Can you show the content of the folder.

[open fox]

It is the old model

[fathom shuttle]

and a lot more

[open fox]

He had it already. The new ones are the ones with the actual timestamp.

[fathom shuttle]

thats what is in the folder

[open fox]

Not the model.
This is the link to the model

https://civitai.com/api/download/models/425083?type=Model&format=SafeTensor&size=full&fp=fp16

[knotty cosmos]

28 mega pixel tomato

[young bridge]

4xNMKD-Siax_200k

I cant find that model

[knotty cosmos]

4 mega pixel tomato

[open fox]

It should download a nearly 4gyte file. Rename to revAnimatedV2.safetensor and it should work

@fathom shuttle

[fathom shuttle]

ye i gotta log in first

[open fox]

Wrong channel mate.

[open fox]

Yes some model owners require you to log in to be able to download it.

[fathom shuttle]

ii cant log in or something

every possible username iis taken

i dont get it

nvm got it

[fathom shuttle]

i assume i put it in the model file?

[open fox]

Good as soon as you downloaded the large file check for the file extension (should be .safetensor) and then put it In the directory /models/checkpoints/

[fathom shuttle]

i dont have checkpoints?

[open fox]

Which UI?

Do you use.

[fathom shuttle]

should be in here no?

[open fox]

The directory you shown on the screenshot earlier with the text file put models here.

[fathom shuttle]

do i put it in the txt ?

with*

[open fox]

No just in the same directory.

[fathom shuttle]

ye like that right

how do you all launch it

since i do it through the explorer and the cmd

[amber cloud]

guys, if a pc has dual 8GB gpu… would that make faster for image generation and deforum processing ? or the 2nd gpu has no effect on the SD ?

[open fox]

Depends, in general PyTorch and other libraries can work with two GPU. If they are Linked and worked as one NVLink they would appear as a 16gbyte unit and that would help a lot as memory is very useful for higher resolution, animations etc.

If not you could a least both parallel so generating a batch of images would be faster.

[amber cloud]

what should i check to know if both has NVLink and will work as one ? i know SLI does not exist anymore

[open fox]

What GPU do you have?

[young bridge]

which upscaler is the best one to use

there is lanczon lsdr

and others

[amber cloud]

i already have 3060.. thinking of getting my brother’s 3060 and install both on my pc

[open fox]

Not a technical issue but a question of taste. I use different ones depending on the subject (anime vs. photorealistic). Sometimes when I am lazy even just upscaly or freescaler….

[young bridge]

photorealistic

how about in extras tab, which one to use with codeformer

[amber cloud]

3060 is installed, would another 3060 help ?

[open fox]

From classic machine learning you would gain like 55% in performance. So you will lose some due to the dual setup which seems not be able to use SLI. So memory limitations (8gbyte) would still be a thing.

[open fox]

For photo realistic I use RealESRGAN_x4Plus

[young bridge]

i used that but it makes the image so smooth

loosen the details

how can I use ADetailer with codeformer

[open fox]

Sorry not using adetailer. Using a control net Tile after upscale.

[young bridge]

ok

[fair oxide]

Have simple question, whats the shortcut to disable/enable groups in comfyui?

[ornate elk]

thats not a model, you can delete the whole thing, models are single files (larger than 2gb)

[fathom shuttle]

ye i always ffound it a bit odd

[warped canyon]

How do I rollback to controlnet 1.1.440 on a hosted service using jupyterlab? I feel like the new controlnet ip adapter face id v2 is giving me a slightly different face output to the one ive been using all this time with controlnet 1.1.440.? How do I rollback?

[mental sentinel]

Hello, does anyone know why I get this error (I don't know much about python)

[cursive sparrow]

Hey all for those of you using A1111 when going for ultra realism for "people" anyone figured out the best combinations of settings? I find this combination is working pretty decent but has anyone figured out the holy grail of settings? Edit: Besides the main checkpoint model.

[frail bay]

can someone help me. i want my photo to me more colorful but all mine are just coming out kinda greyish tint

[smoky jay]

hello, I installed the roop face swap extension for A1111 but a cant see the the roop on the interface and cant use it, someone please knows about that issue ?

[unique bluff]

Okay, I generated an image but the pose is incorrect, is there any way to fix it post generation?

[mental sentinel]

I was able to fix that but now I don't know what to do

[smoky jay]

hellow does anyone know why this error loading roop swaper?

[viral knot]

you need to install onnxruntime, preferably the gpu version onnxruntime-gpu

[formal trout]

Having trouble installing Automatic1111 in Ubuntu 24.04

[viral knot]

yea python above 3.10 is a problem, try to stick with 3.10

[formal trout]

I installed Rocm and pytorch nightly faor Rocm 6.0

[formal trout]

How can I change that? I have no idea

I tried to "solve it" but I messed it last time

Ended up reinstalling ubuntu from scratch

I guess I have to created the venv with python 3.10 but how

[viral knot]

it's been a long time since i used a1111, did you use the precompiled version or what?

[formal trout]

I follow mostly this

[viral knot]

but that is strange... it even says there to use python 3.10, so how did you end up with a higher version?

[formal trout]

I think those command didnt' work

And created the venv with the current python version

[viral knot]

oh boy

[viral knot]

you would have to remove your current python installation completely i think and install a python 3.10 version then redo this (assuming this is not a huge deal for you python wise, or projects wise)

[formal trout]

I tried that, run some command for wiping out python but it messed up everything then

[viral knot]

im not expert when it comes to linux based stuff, i did a little bit, but was a long time ago, mostly windows these days, hmm doesnt linux have an easy way to install python 3.10 ?

[formal trout]

I'm going to try to find it, ty

[viral knot]

yea do some search and then redo the installation of a1111

[sinful needle]

i got animatediff to work, but for some reason it wants to animate literally everything in the scene all the time. ways to fix?

[limber pier]

if you eat enough acid it'll look like all those objects are at rest

[viral knot]

sadly the tech is still not at a level to do that, maybe one day

then again im no expert at animation stuff, so maybe there is a way :3

i mean i can think of one way of doing it actually, but it would be a very complex workflow

[ornate elk]

You need to use a VAE file, for color fix

[ornate elk]

Roop is abandoned, you need to delete the extension. Then install either Reactor or Roop-Unleashed

[ornate elk]

Your third party antivirus Programm broke the Python installation.
You need to:
Delete the venv folder.
Uninstall Python.
Disable your Antivirus Programm
Install Python and check "add Python to path"
Launch the webui-user.bat
Enable the AV again.

[terse lichen]

how to identify which tensor is using my cpu ?

[terse lichen]

what does this mean now

[north sky]

I'm looking for an assistant to help with data science tasks. Prior knowledge of data science is not essential. Applicants must have the right to work in their country of residence. Preference will be given to candidates from the US, Canada, Australia, Argentina, or Mexico. Please DM for more details.

[fair oxide]

guys can stable diffusion convert videos into anime?

also which gpu recommend for stable diffusion i need to convert 30 minute 1080p@60 fps video into animation

any1 who has tried converting videos plz report ur waiting time in minutes

[terse lichen]

@old sky can you pls help me figure this out if free and thanks

[fair oxide]

i was thinking 4080 vs 7900xtx

[ornate elk]

That will take a long time, as you need to convert every frame.
But you can also reduce the video to 30fps and later Interpolate the frames back in.

[ornate elk]

For Stable Diffusion right now a Nvidia GPU is better supported, but it works with AMD too.
I'm running a 7900xtx. Great card.
But if you only want to do Ai stuff then get the 4080 or 4090, or a 3090.
The more VRAM the better for Ai tasks.

[open fox]

Only chance is to disable parts of your workflow and see what is working and what is not working. As mentioned copying a complex workflow with different nodes might require configuration of these nodes and Manuel steps. So to identify the one that is running on your cpu deactivate parts and replace them by a standard node and see if it works. If the error still remains activate the path again and chose the next one.

As obviously one node that you use is not happy to have the choice between GPU and CPU.

@terse lichen

[terse lichen]

then what does this error mean @open fox

this is the workflow i am using

[open fox]

It states that all three parameters have to be the same datatype (query, key and value). In your case the query seems to be a dtype half while the key and the value are from datatype float. So either key and value need to be also half or the query must be a float value.

[terse lichen]

this is the workflow since i am hella confused

[open fox]

Which node turns red when you hit the queue button ?

[terse lichen]

the ksampler turns purple always

[fathom shuttle]

your gpu is worth my entire pc 😭

[open fox]

So back to trial and error as written. I guess the model is fine. So you still got positive negative condition as a source of error.
Switch the positive to a simple clip prompt node and see if it runs (result will not be as wanted). If it runs you know that the input for latent and negative is not the problem. If it does not work additional change the negative to a simple prompt and see if it works.

[terse lichen]

this is the workflow but what should i change in this can you mark it of sorts pls @open fox

[fair oxide]

how many minutes?

[fair oxide]

did u try converting ur video into animation with 7900xtx

sdo thats why i m confused 7900xtx has more vram 4080 has less vram but more compute units

[ornate elk]

I can't say how long it takes. Would need to try

[ornate elk]

AMD Cards are good if they are on Linux where they can use Rocm.
On windows you would need the cuda emulation tool Zluda to get a good performance.
A 4080 would be much faster than a 7900xtx on windows in ai

[timid notch]

what is this problem?

why failed?

i update everything and restart it but still this error ~

[fair oxide]

if we compare both in different systems then both equal or 1 faster than other?

[ornate elk]

The 4080 would still be faster

[fair oxide]

ooh

[fair oxide]

should i wait for 50 series for SD?

AI perf?

or 4080 will do?

[humble cobalt]

Hello, does anyone know how to fix image with "Glitch" effects that appears a the generation ? Im using sdxl vae with PonyV6XL

[ornate elk]

Depends on you, newer cards always perform better.

[ornate elk]

Hey can you show an example?

[humble cobalt]

here, like its not really glitch but its like oil paint idk

even with hi-res

[ornate elk]

Make sure you have clip skip 2
And sampler Euler a or DPM++ 2m

[humble cobalt]

how do i put clip skip 2 ?

[ornate elk]

Do you use auto1111?

[humble cobalt]

yes

[ornate elk]

Then go into Settings, user interface, Quicksettings, there add:
sd_vae and Clip_stop_at_last_layer
Then hit apply and reload ui

[humble cobalt]

doesnt seem to work

[humble cobalt]

its still messing at the end of the generation

[ornate elk]

thne try an other vae

this one: (sdxl.vae.safetensors
)
https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/tree/main

[humble cobalt]

isn't this the same ?

[terse lichen]

can i send you a visual studio script of the error i am facing and how to fix it of sorts

[ornate elk]

nope

[ornate elk]

i cant help with comfyui workflows sry

[terse lichen]

ok but just wondering is the code the same for sd webui and comfyui ?

[ornate elk]

depends on which code

a lot differs between them

but the initial stable diffusion is the same

[terse lichen]

the animatediff i am trying to use is of sd1.5 version mainly

[ornate elk]

ok

in comfyui you have more settings for animatediff as in auto1111

[terse lichen]

this i did not know about

[ornate elk]

whats your issue in comfyui?

[terse lichen]

its mainly 2 nodes not being compatible with the other in gpu and cpu

[ornate elk]

hmm

[terse lichen]

its mainly this @ornate elk

[azure magnet]

Hi, i need an help how i get this

I only have this

[ornate elk]

you need this extension:
https://github.com/w-e-w/sd-webui-hires-fix-tweaks

[azure magnet]

Wait what? is it a bug that the extension fix?

[ornate elk]

nope, the extension gives you these additional settings

these are not in auto1111 by default

[azure magnet]

I haven't this extension but i have it, why?

[ornate elk]

i dont understand this question xD

the extension adds additional features to the hires fix

[azure magnet]

Ok, I'll explain well. I have this. But i haven't install the extension, why?

[late tree]

Some extensions are installed by default

[azure magnet]

It show only when i do png info

[ornate elk]

oh, then check your extensions tab, maybe its installed?

[azure magnet]

This is my extension

[ornate elk]

very strange then, are oyu using auto1111 or Forge ?

[azure magnet]

auto1111

[ornate elk]

whats your webui version?

check it at the bottom

[azure magnet]

1.9.3

[ornate elk]

it does only show up wehn you load an image from png info?

[azure magnet]

https://tenor.com/view/jeffm-minnmax-shrug-minnmax-show-jeff-marchiafava-gif-18027116

Yes and then i can use

[ornate elk]

that means the creator had the extension installed, that i linked

[azure magnet]

ok, so i have some others questions, i'll add this extension https://github.com/w-e-w/sd-webui-hires-fix-tweaks, i had posted my extension, do you have other to suggest me?

[ornate elk]

maybe tiled diffusion & tiled vae extension

helpful when generating large images

[azure magnet]

With new version of automatic1111 i have sampling method and schedule type. Can you link me a guide that simply that explain how it work, and how i should use?

And last question what is downcast alphas cumprod and what it does

[ornate elk]

its the same as in the old webui version, just leave the scheduler on automatic

[azure magnet]

And the last one question?

[ornate elk]

Yes?

[azure magnet]

What is downcast alphas cumprod and what it does?

[lucid sundial]

i really need help, i've tried to google an answer to this and got no helpful answers. I installed stability matrix and everything has been working fine, but whenever I try to use controlnet it throws up error 500 and says the server got itself in trouble. What is going on here? How do I troubleshoot this?

[terse lichen]

thanks for all the help @open fox i finally managed to fix the error

[open fox]

So you found the node which was using the cpu instead of GPU

[open fox]

Or you just switched to another input node to achieve the results

[terse lichen]

it was mainly 2 wrong nodes i used and added those and it worked

[ornate elk]

hey what does the cmd shows?

[lucid sundial]

how do i check that? there's nothing in the error message that says more.

[ornate elk]

in the black cmd window

[humble cobalt]

so i tried to use this vae, seemed to work sometimes, but now it just say connection errorred out after 2 generations and it showed me this

[lucid sundial]

this is all i got

[ornate elk]

can you translate?, i see a memory error?

[azure magnet]

So for better result should i change something or no?

[ornate elk]

whats your gpu?

[lucid sundial]

nvidia 1050ti

[ornate elk]

no you can leave it off, it was just disabled in the latest versions, cause its not needed

[ornate elk]

okay, i wouldnt recommend stability matrix, when you dont even see a cmd window with an error log.
Better install Automatic1111 webui the right way using my guide from the Pinned Message of this channel

[humble cobalt]

WerFault.exe - Application error
The application failed to start correctly
(Oxc000012d). Click OK to close the application.

The instruction to 0x00007FFBD8B95474 uses the address
memory 0x00000000000000000. The state of the memory cannot be read.
Click OK to complete the program.
Click Cancel to debug the program

[ornate elk]

okay, whats your GPU and whats in your webui-user.bat?
and also how much RAM do you have?

[humble cobalt]

rtx 3060 Ti 8GB VRAM 16Go RAM and i have xformers medvram-sdxl and no half vae

[ornate elk]

okay thats good.
but you may need to increase your windows pagefile

[humble cobalt]

How do i do that ?

[ornate elk]

https://www.thewindowsclub.com/increase-page-file-size-virtual-memory-windows

there you need to make sure the Pagefile is only enabled for the C:/ drive and NOT for any other drive

then set Min to 16000 and Max to 24000

then restart the PC

[humble cobalt]

Thanks it seems to work for now

I didnt know 16 Gb RAM wasnt enough

[ornate elk]

yea SDXL needs lots of resources xD

[mortal compass]

Hey @ornate elk, I finally got to try your settings, and I have some follow up questions if you don't mind (it was about upscaling).

First off, this is how my settings in img2img looked when I started out.

I noticed at first, the outcome was not what I expected

I changed a lot of settings, trying to see what would make the outcome better

[ornate elk]

where did you get the sdxl tile model?

[mortal compass]

I'll look it up

[mortal compass]

https://huggingface.co/bdsqlsz/qinglong_controlnet-lllite/tree/main

Anyway I got it to look pretty decent by lowering denoising to 0,05

[ornate elk]

yea lowering the denois can help much

[mortal compass]

I don't understand how you can crank that up to 6 lol

[ornate elk]

i used 1.5 tile model not sdxl

[mortal compass]

That much of a difference! Damn

[ornate elk]

yep, sdxl controlnet models are very different

i will do some tests with the tile sdxl

[mortal compass]

Looks decent, but noting like the images I've seen here. It also introduced artifacts to the image at top left

[mortal compass]

Cool! It's very new apparently

[viral knot]

hey, i didnt even know about this page, thanks 🙂

[mortal compass]

Anytime:) Hit me up if you find some dope setting configurations!

[smoky jay]

thanks so much @viral knot I'll try it

[limpid cove]

do i need to just set up webui to set up stable diffision 3 or are they completely different things?

[terse lichen]

and its again back to error land sigh

[fathom shuttle]

what is .controlnet? i saw something about that im missing it and need it or something

[mortal compass]

Controlnet are tools that you can use in stable diffusion. Controlnet itself is not only one tool or process, but several, with many ways to use it. For example Depth can analyze images and create layers to it that determines the depth of the image. Tiles is a way to manipulate or upscale images through painting in tiles, and so on. You don't need controlnet for stable diffusion to work, but it is an amazing asset.

[craggy scaffold]

Anyone know why Dreamstudio is no longer saving images/prompt history?

pretty frustrating to not be able to see what I entered or have access to images i created. nothing shows after april 29

[warped canyon]

how do i install this verion of controlnet? https://github.com/Mikubill/sd-webui-controlnet/tree/eb451a007f7040288e865f96e9ee0842aa6ef91c

a specific commit

[mental sentinel]

I followed all the steps and I still get an error

[open fox]

Your missing some requirements. Obvious torch and torchvison (both python packages)

You could try to install manually

[mental sentinel]

ok

[ornate elk]

Your python is still broken

What's your antivirus?

[mental sentinel]

It's RAV Although I had to uninstall it because it kept interfering

[ornate elk]

Okay make sure you restarted the PC after that.
Then uninstall python 3.10.7
And install python 3.10.11 64bit
Then delete the venv folder
Then relaunch the webui-user.bat

[mental sentinel]

oh, i didnt restart

[fair oxide]

plz tell me how long it takes to convert 1 video at 1080p@60fps which is of 1 minute into anime with rtx 4080

[open fox]

The reason why people not answering is because there are many factors which determine the time which it would need.
I would guess with a 4080 you get at least 30 it/s with 1024x1024 images. So with 30 steps and some controlnets, context, etc. I would guess 2-3 seconds per image. 60 * 60 * 2,5 could be the total time if the process work. If the model needs a large context etc. you might run out of memory and the swapping Beginns. Then it takes longer. But only a estimate.

[fair oxide]

so how many minutes according to u for 1 minute vudeo into animation

[open fox]

Do the math. 60 frames per seconds, 60 seconds per minute and 2-3 seconds per generation.

[fair oxide]

how if i reduce it to 30 fps

less time

i suck at maths

can u just tell me

thx

[open fox]

Sorry I am not a fan of spoon feeding. But with 30 fps it would take only half as long.
But if you need to generate the missing frames as you maybe need the 60fps it would take additional time. Maybe it is faster creating „in-between“ frames then creating anime images per stable diffusion.

[fair oxide]

tell me ill calculate then

so 60 sec video 30fps per second

how do i calculate this?

so 120 minutes?

no wait 120 seconds?

180 sec right?

@open fox

60x30

or 200 sec if i add in 2-3 secs?

[open fox]

30 frames per second (fps)
60 seconds per minute
2-3 seconds per image generation.
Go for it I believe in you.

[fair oxide]

but video is only 1 minute long so 60 secs

10800 sec am i correct?

@open fox

so 180 minute?

am i correct?

@open fox

[open fox]

Yes

[fair oxide]

omg so 3 hrs

for just stupid 1 minute video lol

[open fox]

3 Hours for 10.800 images.

Not quite bad.

[fair oxide]

ur not trolling are u or sarcasm ?am i really correct?

[open fox]

No as mentioned the 2-3 second is a guess but creating a image2image process to make a real image to anime should not take much longer

[fair oxide]

so image to image is much faster than image video to video?

[open fox]

No faster ways would be special networks for style transfer video like some online tools offering. As long as you need to generate each frame it will take time.

[fair oxide]

can u give eg?

which tools

how much time it takes to generate 1 image og 1080p resolution?

image to image

[open fox]

2-3 seconds as mentioned

[fair oxide]

what thats fast as f

[open fox]

Just google for video style transfer.

[fair oxide]

too fast

[fair oxide]

in sd?

[open fox]

Yes

[fair oxide]

is it faster than SD?

i mean video to animation conversion?

ooh its like deepfake AI

like deepface labs

do u think 5080 will reduce AI timing from video to video conversion into 1 hour or 30 minutes?

@open fox

should i buy 4080 or wait next year

[open fox]

You should think about that if sd is the way you achieve your goals. Pretty sure video style transfer models will do the job better. Just google them take your time to read thru the pages.
In general some are writing better to wait. Even if it is only helping that the prices will fall.

[fair oxide]

is it free like SD or pay monthly like AI tools?

@open fox

[open fox]

I repeat myself. Google it. There are free ones, payed ones, … the search term is video style transfer.

[fair oxide]

okays thx then can i do away with 4080 or 4070ti super with video style editor if thry are not demanding as Sd?

on gpu?

coz j alredy waited 1year for fd in 2023 this year supposed to buy pc i dont think i can wait 1 more bro

year

video style transger i mean

transfer

@open fox

[open fox]

https://github.com/westgarthb/style-transfer-video-processor
This is for example 3 years old but I guess there are newer repositories available. But you can see how they work.

[fathom shuttle]

Thx boss

[serene notch]

hey quick question, i have never done this before but i want to try and train some loras, i have an i7-6700, rtx 2070super 8Go, and 32gb of ram, is this good enough for that?

[ornate elk]

yes that will work

[hard kettle]

Helo, what is the best upscaler for comfy? I am thinking in 4x for video upscale. nothing overkill, like 8x or something. I am striving for sharp results.

[mental sentinel]

it worked

ty

[ornate elk]

perfect, np 🙂

[craggy scaffold]

anyone know why dreamstudio no longer saves prompts or images generated in its history?

@echo hill could you mention to devs? submitted a ticket but no response

[echo hill]

I'm not really sure about that, but if I find anything out, I'll let you know! In the meantime, I would check back and see (depending on when you sent in the ticket,) or send another e-mail about it!

[ornate elk]

so i found out, that its not possible to use the sdxl tile resampler for the nice upscale you can do with the 1.5 tile controlnet

[sinful needle]

hey what did u tell me previously to add to the command line args to be able to checkpoint merge the SDXL motion model .ckpt?

something about nopickle

im making an instruction manual for dummies like me in case i forgor stuff

[ornate elk]

hey, it was --disable-safe-unpickle
but make sure you remove it after merging, as its a security risk to have it active

[sinful needle]

yep

thanks

[ornate elk]

np

[sinful needle]

my instructions are literally go-to-link, download something, place in specific folder with pathname, or edit file and specific line. most tutorials leave out steps because they assume us normies know a lot about it and its infuriating

[ornate elk]

true, thats why i made my guides as simple as possible and keeping every step

[chilly knot]

facefusion inswapper_128.onnx failed:Protobuf error

when start make video

[chilly knot]

🙁

i redownload inswapper_128.onnx and now new error with opencv

[brittle lantern]

Guys I have 8gb of vram should I use SDXL models or SD 1.5?

[maiden coral]

you can use both

you only cannot make complex workflows if you have ComfyUI

or probably can, I have half yours, so it's ok

[ornate elk]

you can use both at a decent speed.
if you have an nvidia gpu and using auto1111 webui, you need to edit the webui-user.bat
at the line Commandline_ARGS= you need to add: --xformers --medvram-sdxl --no-half-vae
for the best performance

[reef shoal]

Hello. what requeriments i need for run SD? i only want make small res (768x768) images (my graphic card: Nvidia GeForce GTX 750, 2gb)

[azure ivy]

hey guys , sd webui mov2mov or others that are in the master branch column does not show up where they should show up

how can i fix this ?

[woven locust]

Hello everyone, please suggest is this PC setup is ideal for Stable Diffusion 1.0 SDXL -
Component
Option 1: RTX 3090
Option 2: RTX 4090
Processor
Single Intel Xeon Gold or Platinum processor
Single Intel Xeon Gold or Platinum processor
(e.g., 16 or more cores)
Memory (RAM)
128 GB ECC DDR4 RAM
128 GB ECC DDR4 RAM
Storage
2 TB, SSD-based storage
2 TB, SSD-based storage
GPU
NVIDIA RTX 3090 (24 GB VRAM at least)
NVIDIA RTX 4090 (24 GB VRAM at least)

[brittle lantern]

AMD GPU on Linux with ROCm i have

[rich garden]

anyone know what this is and how to fix it? i tried running "webui-user.bat" just now after not using it for a couple of weeks and got this error

also should i be using "webui-user.bat" or "webui.bat"? i kinda forgot 😅

[oak sun]

Hi ..anyone know how to use old versions of torch with stabel diffusion..like force it to use certain older torch as I'm unable to update my gpu driver due to some reason and every time I update it it disconnects and i need to uninstall the driver and downgrade to get it to work .it used to work for me 7 months ago and I stopped then and now I'm getting this error saying torch unable to use gpu and adding skip torch cuda test but whenever I do that I get an error that my driver is old..any chance I can find a solution for that ..the gpu is gtx 1070 on my laptop and it stops working with latest nvidia drivers for some reason but as I said with this driver it was working 7 months ago but I'm finding it hard to even use older commit

[open fox]

Ideal is hard to define. I saw some pretty nice Benchmark results from a nvidia h100 with 80gbyte of Memory. If it is only for stable Diffusion you will not Need the cpu and the 128 gbyte ecc ram a lot. So for price-performance optimization you could go way chaeper with a non xeon/threadripper plattform.

[open fox]

In General the requirements.txt or the install scripts determine which torch version is needed. You might deinstall and force the install of an older package with the pip command. But I am pretty sure if the rest of the application is current it might use function of torch which were not implemented in older versions. So you could try different torch version from the past and hope for the best. If this is really the problem.

[rich garden]

@open fox why'd you skip me?

[brittle lantern]

some1 else came first lol

[rich garden]

yeah chin_key was before me but i was before diamond

[brittle lantern]

skill issue i assume

[rich garden]

do u have a solution to my problem?

[brittle lantern]

i do not otherwise i would've answered

[oak sun]

How do I force installation of a certain commit from the past along with its torch and requirements

[open fox]

It is written that “git” is a command that you did not installed. So you are missing support for git or in your path environment the path to git is not set. So install git and retry.

[rich garden]

idk how git somehow left my laptop when i've used this software fine before but ok i'll reinstall thx

actually nvm i uninstalled it not knowing what it was LMFAO

[open fox]

The python package manager: pip install -I torch==<VERSION>

[oak sun]

No not the version of torch

I know that

[open fox]

I use comfy so I can’t tell if auto1111 got releases that you can pull.

[oak sun]

Did comfy also update torch in the last 7 months

It used to work for me but auto1111 is a lot simpler

[open fox]

Pretty sure. 7 month is ages in terms of ai 😉

[oak sun]

Yeah but I remember torch not being updated that regularly in 2022 and 2023 even though there were newer versions available

[open fox]

You could check:

https://github.com/AUTOMATIC1111/stable-diffusion-webui/releases

And git clone an older release.

[oak sun]

How do u clone an older release ..that was my original question..

How do u clone an older commit

[brittle lantern]

guys using auto 1111 on windows with AMD GPU is using my 5600 cpu and not the gpu how do i fix this?

[oak sun]

It isn't downgrading the requirements no matter what commit head I use

[open fox]

git clone …stable.git -b v1.8.0
Basically two minus and branch followed by the name of the version.

[open fox]

Well to cite @ornate elk who is using an amd card as well. Try the guide which is here in the channel. Zuleda or so is what seems to help.

[oak sun]

I used v1.7.0 it seems to be collecting cu118 which should work I think

That's for downgrading

AFAIK

[open fox]

Good luck. And hopefully you get the option to upgrade your gpu drivers some day 🙂

[oak sun]

I will build a pc soon but I wanted to fiddle a little with Ai

It's working now

I don't think I will miss a lot

I ran sdxl on that card and there was no huge improvement over it in this last year AFAIK

[open fox]

It is more for some of the newer schedulers, or stable video, … more about addons that might not work. IPAdapters. … simple text2image will work

[oak sun]

That's all I want

Thank you

[woven locust]

Nvidia H100 is showing to be very costly, will RTX 3090 or 4090 give somewhat similar outcome?

[brittle lantern]

done it lets find out if it works

[open fox]

see by yourself benchmark of different systems and cards.

https://vladmandic.github.io/sd-extension-system-info/pages/benchmark.html

[brittle lantern]

now i have this issue "AttributeError: 'NoneType' object has no attribute 'lowvram'"

[open fox]

Did you edited the webgui.bat as mentioned in the pinned message. At least I do not see passing the lowvram parameter in the guide.

[terse lichen]

just curious but are there major difference in automatic 1111 and comfyui and which do you think is more simpler

[open fox]

Simpler is automatic 1111 but as a more technical guy I prefer the node based logic of comfyui. And there is a reason for both tools so it is only depending on the favors of the user and their focus.

[terse lichen]

i keep on getting that error of cuda and gpu for some reason and i am using the right tensors and all but it still happens

[open fox]

Reduce the complexity of the workflow. Get rid of parts, try if it works you get closer to the nodes that creating you trouble.

[past birch]

does anybody knows how to fix that problem the path is the exact path and I get this problem on every pc I try it on
please help

[ornate elk]

This GPU is not supported and won't work with stable-diffusion

[ornate elk]

Hey, did it worked?

What's your GPU?

[ornate elk]

Can you show the video path in deforum?

[past birch]

@ornate elk

[ornate elk]

Try changing the \ in the path to /

[past birch]

@ornate elk still the same

[ornate elk]

Can you move the video out of OneDrive?

[past birch]

@ornate elk still, do you get this problem to?

[ornate elk]

What's with the settings file path?
Is it correct?

[naive dragon]

alright im getting this still 😭 (its on my ssd now)

[naive dragon]

this is still right if SD is on E right?

[naive dragon]

i mean if you need the whole thing its super long so i dont know how to do the whole thing ;/

[past birch]

what do you mean settings
is it in deforum or is it in my pc?

[ornate elk]

Deforum

[ornate elk]

There were some changes made for zluda.
You need to recheck my guide.
You may need other rocm files.
Also make sure your zluda files are the latest (from 4 days ago)

[azure ivy]

do you know why the ones that are listed as master in branch column doesnt show up in a1111?

[naive dragon]

i redownloaded it 2 or 3 days ago so the files should be fine but ill recheck your guide thank you 😅

[naive dragon]

do i just have to download the default package again that i already have bcs i have a rx 6700 xt and not 6600 or did nothing change for me?

[ornate elk]

You mean mov2mov ?
It needs a special fix
I can link you the fix

[ornate elk]

You need to download the optimised files

Idk which ones you had

[naive dragon]

sry i might be blind but optimised thats only for those specific gpu's right

[ornate elk]

Oh wait. I forgot to add something!

[naive dragon]

😅 its fine lol

[naive dragon]

atleast your helping me haha

[ornate elk]

No the guide is updated

[naive dragon]

do i remove the old stuff in the libarary (i renamed the other library to old library but i still have the other stuff downloaded from not specific gpu)

[chilly knot]

for unlock cuda i need uninstall onnxruntime?