🤝｜tech-support

[worn frigate]

it finds this

so i just click next?

[ornate elk]

Yes

[ornate elk]

Okay click next

[worn frigate]

its here now

[ornate elk]

Click ok

Then on Quarantine

[worn frigate]

should i check it before quarantine

the acer predator name

[ornate elk]

Not this one

Its just the acer updater

[worn frigate]

Acer predator is laptop nname

so i keep it uncheckd

and click qurantine

ok?

i am going for it

here is the log

Start: 03-06-2024

Duration: 00:00:01

OS: Windows 10 (Build 19045.4046)

Cleaned: 11

Failed: 0

***** [ Services ] *****

No malicious services cleaned.

***** [ Folders ] *****

Deleted C:\DRIVERTALENT
Deleted C:\Program Files (x86)\OSTotoSoft
Deleted C:\ProgramData\DRIVERTALENT

***** [ Files ] *****

No malicious files cleaned.

***** [ DLL ] *****

No malicious DLLs cleaned.

***** [ WMI ] *****

No malicious WMI cleaned.

***** [ Shortcuts ] *****

No malicious shortcuts cleaned.

***** [ Tasks ] *****

No malicious tasks cleaned.

***** [ Registry ] *****

Deleted HKCU\Software\OSTotoSoft
Deleted HKLM\Software\Classes\Installer\Features\A38C15B2D5649AE4C9CDE19DE50DA96C
Deleted HKLM\Software\Classes\Installer\Products\A38C15B2D5649AE4C9CDE19DE50DA96C
Deleted HKLM\Software\Microsoft\Windows\CurrentVersion\Installer\UserData\S-1-5-18\Products\A38C15B2D5649AE4C9CDE19DE50DA96C
Deleted HKLM\Software\OSTOTOSOFT\DRIVERTALENT
Deleted HKLM\Software\OSTotoSoft
Deleted HKLM\Software\Wow6432Node\OSTotoSoft
Deleted HKLM\Software\Wow6432Node\MICROSOFT\INTERNET EXPLORER\MAIN\FEATURECONTROL\FEATURE_BROWSER_EMULATION|DRIVERTALENT.EXE

***** [ Chromium (and derivatives) ] *****

No malicious Chromium entries cleaned.

***** [ Chromium URLs ] *****

No malicious Chromium URLs cleaned.

***** [ Firefox (and derivatives) ] *****

No malicious Firefox entries cleaned.

***** [ Firefox URLs ] *****

No malicious Firefox URLs cleaned.

***** [ Hosts File Entries ] *****

No malicious hosts file entries cleaned.

***** [ Preinstalled Software ] *****

No Preinstalled Software cleaned.

---

[+] Delete Tracing Keys
[+] Reset Winsock

---

AdwCleaner[S00].txt - [2341 octets] - [06/03/2024 14:20:07]

########## EOF - C:\AdwCleaner\Logs\AdwCleaner[C00].txt ##########

[manic snow]

it didn't recognise the file as malicious, so it could be a false alarm.

[inland zephyr]

ay I'm using Forge WebUI and it seems like it's unable to use XL models properly. Are there settings somewhere that I'm missing?

[worn frigate]

hi

i think it did not work?

[ornate elk]

You need to restart the PC

[worn frigate]

ok i am restrating

[dense obsidian]

via con dios

[ornate elk]

What's the issue?

Any error in cmd?

[inland zephyr]

No, it just produces terrible images

Like I did "an anime girl in a fantasy village" as just a basic prompt with nothing else enabled, no LoRA, no extensions in use

[worn frigate]

restarted

[inland zephyr]

This only happens with XL models

[worn frigate]

and its still there

i think i lost my pc

right.?

[ornate elk]

What are your settings for that?

[ornate elk]

Can you select Remove?

Then start action

[worn frigate]

i can select remove

then start actions

[inland zephyr]

768x512, 30 steps, 10 CFG, DPM++ SDE Karras

Nothing else in use

[worn frigate]

but it still does nothing

😭

[ornate elk]

Can you try 768x1024 ?

[worn frigate]

when i go to this folder, i see nothing like that there

[ornate elk]

Can you click "see details"

[worn frigate]

and the cleaner shows nothing too

[ornate elk]

Go into the Appdata folder and delete the Temp folder

[worn frigate]

ok

[ornate elk]

Its in C:/Users/YourUserName/Appdata/Local

[worn frigate]

ok

says its open in another program

i closed all programs

[ornate elk]

Then go into the temp folder

Select everything and delete

[worn frigate]

ok

[ornate elk]

And if something is in use press Skip

[worn frigate]

done

only adobepcbroker left

and one nodejs thing left

[ornate elk]

Good

Now you can run a windows defender scan again

[inland zephyr]

@ornate elk 1024 and higher seems to work better, thanks

[ornate elk]

Good. Thats because sdxl models are trained on a 1024x1024 resolution

[inland zephyr]

I did know that, but I didn't think it was to the exclusion of lower resolutions

[worn frigate]

its still therre 😭

[ornate elk]

See details

[worn frigate]

where to see details

oh

wait

this

my life is flashing before my eyes

[ornate elk]

Clear your trash bin

And then restart and scan again

[worn frigate]

ok

[dense obsidian]

I vote back up the work and factory reset

[inland zephyr]

@ornate elk ay before I go back to creating dumb shit

Does Hires fix have the same value for XL checkpoints as it does for earlier non-XL checkpoints?

Or is it something that's kind obsolete now

[worn frigate]

I am with my phones browser now. My dad is home and he does not let me use Pc. I had to shut down. And i will be able to do it next time i get pc. But i saw a thing when i shut down

[ornate elk]

Its still good to use it. Sdxl at higher resolution gives good quality but still can be improved with highres fix

[worn frigate]

It says Virtual box connection has a connected, still shut down?

And i clicked yes shut down

[inland zephyr]

Alright, thanks for the heads up

[worn frigate]

Is someone like seeing my screen? I see scammers can see screen of other person.

It also happend earlier when I restarted. It showed before restart

I had no applications On. All closed.

[ornate elk]

The best practice would be to backup your files and to clean install windows with an USB stick

[worn frigate]

I am shaking.

So i nust reset window?

Like is there a button?

To just reset it all.

[ornate elk]

There is a button in the Settings/Update & Security/Restore
The button is called Reset this PC

The securest way would be manually with an USB Stick

[worn frigate]

Do i need to download window 10 on my usb?

[ornate elk]

Yes, windows has a tool that will prepare the USB stick for you

[vocal burrow]

https://support.microsoft.com/en-us/windows/create-installation-media-for-windows-99a58364-8c02-206f-aa6f-40c3b507420d

[worn frigate]

Will my phone work?

Because i do not have usb

[vocal burrow]

I don't think so

phones connect using some special protocol (mtp)

[worn frigate]

Updatr i managed to open oc

And its its still there

ig its lock scren

what i do 😭

[vocal burrow]

I didn't start reading from the beginning cause your adventure is quite long. But LockApp.exe is a standard thing.

not necessarily a virus

[worn frigate]

hm

so whats best i can do

i do not have usb

I guess there is nothing i can do right.

I have 200+ GB of files and data.

Have no usb.

If i delete anything, it will wipe it all.

[lost chasm]

unplug the pc from the internet and back up all those files before you do anything else.

[worn frigate]

Where woulr i even backup

[lost chasm]

to a cheap external HD

[worn frigate]

I dont have that much storage. I have 45 gb on phone.

I will backup only importsnt stuff.

[vocal burrow]

your options are
1/ keep running adwcleaner + defender scan until it all comes clean
2/ backup the data and factory reset the pc. You might be able to get away without a usb disk if your hard drive has multiple partitions. I actually never used any "factory reset" option but maybe it will only reset c:
3/ backup the data (create a second partition to backup stuff). Get an usb key, create a windows live usb, reinstall windows,

[lost chasm]

a 1tb external HD is super cheap

[vocal burrow]

and uninstall virtualbox if you have it installed for some reasons and don't use it

[lost chasm]

these are dirt cheap now

if cost is an issue

[vocal burrow]

(depends of the country)

[ornate elk]

In the "Reset this PC" setting there is an option to keep "personal files"
So that could be an option to try before the USB way is needed

[vocal burrow]

still before doing any kind of reset, backup stuff

you never know

[lost chasm]

yes backup first

[vocal burrow]

and relax, your computer is not gonna self explode.

[inland zephyr]

This is why having multiple hard drives, with your OS and stuff on its own drive is important

[worn frigate]

Problem is i am a like kid. And my dad is super strict with pc. And i sneak in and use it everytime he is not home. So My dad 1 would not buy me A storage chip becaue he wants me to only study. I am really sorry but he is very strict. I can not let him know i still use pc. I can not just steal money from home to buy. I really have to manage to with 64 gb at max of phone. Even now i can not use pc because he is back home. I am on phones browser. I have After effects and Photo prohects but i need only 2 of them alon with keyboard shirtcut fikes. Other than that there is nothing on pc. And i can install stabke diffusion later by downloaing.

So i can 100% reset it all.

Barely 5 gb fies i need to savem

[dense obsidian]

after effects isn't cheap...hm

[worn frigate]

I did not buy jt

I just try to create animtion and i love drawing and animation

[vocal burrow]

I mean .... if he uses the pc (and have a functional brain) he's probably gonna notice any kind of reset. stuff like browser history, stored passwords are gonna go missing.

[worn frigate]

I do not have have passwords save but 1 lkgin to google only

I have like destructive pc? I have no history. I just use private tabs

I do not have photos except for projects which arr all downloaded not prsonak.

So um.

I wilk reset guess.

[dense obsidian]

best save as much stuff for you dad and replace after the reset

I should be working on my project...but this is riveting.

[vocal burrow]

If you are the little kid trying not to get caught then your files are probably not what you should be worried about at first. You probably should save your family's files.

And I can't guess from afar what your family is using the computer for.

[worn frigate]

My mom knows it and allows me to use pc. Just dad dont know. This pc is not family and thankfully no one uses it it does not have single picture of us or any file which is related to real life. I wanted to learn graphucs animation all that, its like hobby. So i onky have projects. For example simple cartoon s or images.

So i do not need those prijects now because i already done that. I do not need anything except keyboard shorts

And just 2 projects which i was working on

[vocal burrow]

ok then

[worn frigate]

Then my stupid brain seens Topaz ai.

And i go to getintpc

Follow their tutorial.

It was saying me to delete and i thought oh getinto pc is good.

I also have 3 othrr things which never caused problem

So i ran it.

It patched and topaz ai started working.

It was supposed to generate a key for topaz ai.

And it did. Althiugh i deleted the maine exe.

Patch. Exe

It installed dup2patcher in temp folders

And now I this is where i am.

Regretting.

[vocal burrow]

Relax, it's not the end of the world.
1/ it's not your family's computer so nobody is gonna notice
2/ very worse case you just wipe the whole thing and it will work again
3/ you can always try to keep running virus and adwcleaner scans
4/ backup + reset is usual stuff, especially when it comes to windows machine.

[worn frigate]

Thank you😭💚

[dense obsidian]

and you're still young, hopefully you learned a very valuable lesson

[vocal burrow]

yeah exactly what I was typing :p

don't run "GetMoreRam.exe"

unlimitedVBucks.bat

[worn frigate]

I will never ever run any exe never. And will not use pirated software ever again. Even after reset I am forced to get 3 application though because i dont have miney. But they are safe i have usrd them for 3 years.

Photoshop 2022, after effects 2021 and Davinci 18.4

[vocal burrow]

I won't discuss high sea stuff here. Sure some of them work, and also sure some of them will install shady stuff (some very "in your face", some will just run stuff in the background)

[worn frigate]

Hm. I try my best. Keeping firewall strict. But i think it can still slide

But i have no choice. Atleast for those 3.

I will never download anything else never.

I am so sorry😭 i dont know. I have just lost control and i want to say thank you for being here for me. Thank u. I feel better. Thanku. Sorry if i disturbing. But still thanks u.

[lost chasm]

Out of curiosity how come your dad won't let you learn professional software that people can make careers out of using, its not like you are pissing away time playing games of surfing for pornz (well maybe you are too, but lets focus on the productivity software instead lol)

[solar geyser]

It's a little early in the morning for me, so it may not be wise to get into this right now, but while it's fresh in my mind, I wanted to ask for some pointers on installing Stable Diffusion for someone brand new to it. I'm following this: https://community.amd.com/t5/ai/updated-how-to-running-optimized-automatic1111-stable-diffusion/ba-p/630252 and although I believe I've followed it correctly, I'm still getting a runtime error regarding no NVIDIA driver found on my system. Stable Diffusion webui launches, but can't generate anything and gives that error on the webui itself as well as the CMD window. To be clear, I am using an AMD build.

Am I in the right place?

(Excuse me for butting in btw.)

[dense obsidian]

there's a pinned walkthrough for amd install I used, very helpful

[ornate elk]

Hey, this guide is outdated.
What's your GPU model?

I've made 2 amd guides. You can find them in the Pinned Messages of this channel

Zluda is the recommended one, but requires a newer GPU

[worn frigate]

(I am writing for fabricated girl)

[solar geyser]

Radeon RX 6800 XT. I'll look at these guides then, thank you. I wasn't sure where to start so I just grabbed a guide 🙂 .

[ornate elk]

Then you should follow the Zluda guide 🙂 its pretty fast on AMD

[solar geyser]

A million thanks, if I have issues with it, I'll post here again.

[ornate elk]

Sure we can get it to work ^^ using AMD too

[dense obsidian]

now I am wondering if I should update to zluda XD

[ornate elk]

What was your GPU again?

[dense obsidian]

7900xtx

[ornate elk]

Ohhh

For real use zluda

I have the same card

From 4it/s to 16it/s

[dense obsidian]

[ornate elk]

With that you can do much more video stuff

Batch img2img will run in minutes

And image generation takes seconds even when upscaling

[dense obsidian]

oof, well, guess I'll update to that now then

[ornate elk]

It was experimental at start but I've tested it a lot and can say everything now works, even sdxl finally

[lost chasm]

3d software can be complicated, you probably need a specific task to accomplish as a mission to find the most useful path through it (I've been doing 3d animation/vfx stuff for idk 28 years)

[vocal burrow]

We're kinda drifting away from #🤝｜tech-support stuff here. But please remember that different countries have different traditions, different family structures, different religions, etc... So for instance comparing the life of an average afghan teenage girl to an american one doesn't make much sense. It's pretty easy to forget this when we're all talking through discord.
I won't get too much political in here, so I'll just say this here to whomever might need it. If you ever need to talk to someone else there are websites such as https://www.7cups.com/ that allows you to get in touch with """"random listeners"""". And while those "listeners" are not professional psychiatrists it can help you to get someone to vent to, to give you a different perspective on your situation, etc.

[worn frigate]

.
.
.
Sorry. So sorry

I thought i would try to write it for fabricated girl. I will stop writing. That was all. Sorry.

You may delete msg.

[lost chasm]

its all good, i asked, I wanted to hear the answer. If its not apropriate in this channel, that's for others to decide.

[worn frigate]

Thanks I will delete

[vocal burrow]

I will probably not. But I also think that there are more appropriate channels / website for this. Not only for this channel but also mostly for yourself.

[worn frigate]

Its ok : )

[vocal burrow]

(and please don't be sorry when you're explaining your situation to someone who asked for it)

So yeah, in the end.... little steps

[worn frigate]

I will get some food and try to relax. And reset next time i use pc.

All solved i guess.

[vocal burrow]

when it comes to the pc situation, you now have your attack plan

[worn frigate]

Yes thanks to you all. I will follow it.

[lost chasm]

well i wish you the best circumnavigating your malware issue and hope you get to put the stuff you've been learning to use professionally.

[solar geyser]

I've added this command in webui-user.sh and webui.bat so far and it still gives the same error when launching the webui. Just wondering what else I can do here.

[ornate elk]

Oh, you can go with the next steps

I need to edit that part again
Its fine for now

[trim raven]

hello, the amd user guide is up to date ?

[solar geyser]

I'm using it right now 😉

[ornate elk]

Yes what's your GPU?

[trim raven]

6950XT

[solar geyser]

Right, I'm at the end where you launch from the webui. I'm running webui-user.bat again since it's giving me this error.

[ornate elk]

So your at the launch after you installed zluda and hip ?

[solar geyser]

Correct

[ornate elk]

Then the zluda guide should work for you

[ornate elk]

Oh okay, then show me the paths you set

[solar geyser]

The path in environment settings in Windows?

[ornate elk]

Click on Path

And then edit

[ornate elk]

Yep they are missing

[solar geyser]

Strange, I remember putting them in, ok I'll fix that.

Apologies, it's 6am and I haven't slept yet 😉

[ornate elk]

No problem ^^

Why your still awake then haha xD

Tried to get SD running the whole night?

[solar geyser]

Yeah, using the wrong guides and youtube videos, as previously-mentioned, lol. I started late though, around 1am.

I was pretty upset with Midjourney and their staff so I decided to learn for myself.

[ornate elk]

Ahh damn, I know there are so many outdated and wrong guides our there. Or guides that are just for nvidia but they don't say it in the beginning xD

[ornate elk]

Thats a great idea. You will also have more freedom in generating images with more tools

And more models

[dense obsidian]

now I know why I didn't start with zluda, it said for advanced users, lol definitely not advanced when I saw it

[solar geyser]

My feelings as well. I like to do my own things, host my own stuff, etc. Just new to SD and generative software in general.

[ornate elk]

You dont need to add the path to bin
Instead type the
%Hip... From the guide.
And then the C:\Zluda

So you have two new entries in the end

[ornate elk]

Yea it was experimental and required more advanced steps to get it to work. And it still does for some GPUs. But the Directml dev made a few updated to make the install easier

[solar geyser]

Ok, let's try this, haha. What a morning:

Same error

😦

[ornate elk]

Are you on the latest amd driver ?

At least 24.1.1

[solar geyser]

I'll double-check that, but to be clear, I'm also getting this error hen launching webui-user.bat:

[ornate elk]

Ohh do you had a nvidia gpu before?

[solar geyser]

Nope, not on this machine.

[dense obsidian]

and did I unzip in the right spot?

[ornate elk]

It needs to be in a separate folder

Not together with the actual webui

[dense obsidian]

ohh, like this?

[ornate elk]

Okay, as you may followed some outdated stuff before.
Open up a cmd.
And then type
Pip cache purge

Hit enter.

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

[solar geyser]

Doing now.

[haughty elbow]

help help. please.
i had installed stable-diffusion-webui 5 hours, but it is still failed.

macbook pro. Intel UHD Graphics 630 1536 MB, python 3.10.6

[ornate elk]

Hey for mac we recommend using this auto install script. That will setup everything correctly:
#🤝｜tech-support message

[dense obsidian]

does it matter where I put the --skip-torch-cuda-test in my commandline args?

[vocal burrow]

it shouldn't

[ornate elk]

Dont do that

[solar geyser]

Is it normal to fail a patch between torch and zluda?

[dense obsidian]

lol too late XD, well, it's installing requirements at the moment. I'll wait till it errors out again

[haughty elbow]

@ornate elk
i had try,
1 restar computer
2 delete venv
3 search the solution in https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues?page=2&q=Segmentation+fault
4 i have try webui-macos-env.sh, change python_cmd="python3",

but not take effect

[ornate elk]

How does your Zluda folder looks like?

[dense obsidian]

alright, it error'd out

[ornate elk]

Then remove that --skip-torch line and delete the venv folder again

No I mean the zluda files folder

[solar geyser]

If you're referring to the folder where I unzipped the Zulda download:

[ornate elk]

Yes and this should be in your system environment path

But I guess you linked the webui path

[dense obsidian]

ooo my zluda is up and running

[solar geyser]

Gotcha, that's the issue.

[ornate elk]

Quick info.
The first image generation will take 15-40 minutes

Thats a one time thing

[dense obsidian]

ahh, well now it's time to download all my extensions again, move my lora's

[ornate elk]

You don't need to move them if you also want to use the other webui

But you can

[dense obsidian]

I was thinking about adding the other line to the command arg, but, I don't think I need the other one if it runs slower

[ornate elk]

True

You will be shocked how fast it is ^^

[dense obsidian]

I'm down to be electrocuted for a good cause 🌩️

[solar geyser]

Alright, When I launch, I get some other errors, but I don't want to keep you up all night/day. I'm going to get some sleep in a few and then will follow-up with you. Thank you so much for your help so far.

[karmic crown]

@ornate elk@vocal burrow if someone complains that deforum does not work for them, this is the situation

https://github.com/deforum-art/sd-webui-deforum/issues/948#issuecomment-1980665335

[ornate elk]

Feel free to post the errors here anytime or @ me.
Get some sleep, we can figure it out later ^^

[karmic crown]

i literally beg them to merge that already 🙂

[autumn void]

Hi, I've been having seems like memory problems at Focus ai launch.

is that normal?

I'm running 16GB - I turned off XMP in the bios because it was just crashing my computer with that function on

[dense obsidian]

I downloaded deforum since that is what someone I'm following on yt suggests, haven't even used it yet XD

[karmic crown]

if it is not working, read what i posted, if it still not working, show me the error and ill help you fix it when i find time

[dense obsidian]

no worries, it's on the backburner at the moment.

[karmic crown]

i need to get back to work, talk to you people later

[sand panther]

i created an image on stable diffusion and it happen like this. It create an image but i can't see on stable diffusion. I must open the folder to see it. How i could solve this bug.

[sterile garden]

Anyone seen mud/clay results like this before? It seems to happen after changing models a lot. It gets to where a particular model always generates a result like this. After restarting the SD instance, it's back to normal for a while.

Happens with different models. But it seems to only happen with XL models

[ornate elk]

Whitelisted the webui in any browser adblocker

[distant yarrow]

Many times i have tried to get images of bangladeshi items like shahid minar (Language movement memorial), liberation monument, Kash flower etc, but can't get those images. I have checked that SD was trained with these subjects but not generating image of these. Why? How to get those item's image?
Thank you.

[upbeat crane]

really thought i had worked out all my kinks with Forge yesterday but now i see that i have zero controlnets. i can click to select the preprocessor then the models drop down box to the right is just straight empty, anyone have any ideas or seen this? not much online about it other than a few bug reports

saw one person say that creating a link in his webui-user.bat for GFPGAN solved his issue

which made no sense to me but i tried anyway, guess i can do a symlink but id really like to have these integrated CNs workin, yerkk

[dense obsidian]

ahh here's a new error

ModuleNotFoundError: No module named 'tqdm.auto'

[ornate elk]

Make sure your controlnet models are in models/controlnet and not in Extensions/sd-webui-controlnet/models

[upbeat crane]

oh hmm cool

[ornate elk]

Would need a full error log

[upbeat crane]

so i should move them there then?

that is an issue with TemporalNet overwriting the tqdm version i think

[ornate elk]

Thats after you installed extensions right?

[dense obsidian]

yeah

[upbeat crane]

or it just straight didnt get installed right did you just install temporalnet?

[dense obsidian]

the one thing I realized I made a mistake on was I think I installed something that needed controlnet before I installed controlnet.

[ornate elk]

Rename the Extensions folder and relaunch the webui-user.bat

Or show which extensions you installed

[upbeat crane]

https://github.com/CiaraStrawberry/TemporalKit/issues/81

[dense obsidian]

ohhhh

[upbeat crane]

poke through there and you see the comment with the most likes and huggy faces

i dealt with it a couple times and deleting tqdm and the folder, then going to your python install copying and bringing it over

[dense obsidian]

hmm, deleted temporalkit, still got the same error. Hmm

[upbeat crane]

answer 100% is on this page

you can do a pip install tqdm --ignore-installed from cmd

[ornate elk]

You need to delete the venv after

[dense obsidian]

I think I deleted the venv, but, doesnt hurt to delete it again

[upbeat crane]

Delete tqdm and "tqdm-4.11.2.dist-info" folders from stablediffusionfolder/venv/Lib
run pip install tqdm --ignore-installed

Go to: C:\Users\USERNAME\AppData\Local\Programs\Python\Python310\Lib\site-packages
copy tqdm and tqdm-4.65.0.dist-info folders in to stablediffusionfolder/venv/Lib

thats what fixed it for me

[dense obsidian]

just got to that part, yeah, looks like I'll need to get into python

[upbeat crane]

there is something in temporalnet that like over-rides the tqdm version causing a mismatch or a tqdm.auto not found issue

so @ornate elk were you sayin that i need to move my controlnets away from the extensions folder?

and put them them into models/controlnet folder (which is currently empty)

cool hope they stay workin in a1111

much thanks

[ornate elk]

Yes they should work in both

Np

[upbeat crane]

righteous

[dense obsidian]

hmm, I don't see a tqdm or tqdm-4.65.0.dist folder. I see this though

[upbeat crane]

yeah they aren't in the root, one sec

lib\site-packages i think

if you dont see both the file and the folder in there then you probably need to reinsteall tqdm with the pip i posted above

[dense obsidian]

got it I think

[upbeat crane]

yeah just delete the real old 4.1.11.2 or whatever from vevn/lib if its still there

grab the freshie out of the python site-packages then bring it back to your venv/lib

[worn frigate]

@vocal burrow @lost chasm @dense obsidian @ornate elk @manic snow Sorry for the ping guys thank you so much. The issue is finally fixed. And laptop is fully safe now.
What was happening is the malware .dll files were deleted right when the malwarebyte detected and removed them completely. But the windows defender did not Know this. So it had a scan history in program data. And everytime It was saying 1 issue that was from history. So when I clicked To take action and remove it. It could not remove it because it was already gone the first time we deleted. So there was no malware present. Even I tried to reveal every hidden file, deleted whole folders, and nothing was there because it was already gone. so by deleted the scan history and restarting fresh. Results come clear. Fully safe and no threats anywhere. Also malwarebytes says 0 threats.
Fortunately did not had to go through reset pc process and all files did not vanish. But i did create backup for files on my phone. And I swear, never ever neverrr I am going to run an unverfied publisher's exe. Neverrr. Oh god ny breath still stucks when I imagine it. And We know this was least that happened. This could have been way worse. It was not much of virus, it could destroy main system or encryt my files or anything... Thank you all soo much. I thank you wholeheartly. And you almost did as much as saving my life. It really was very precious for me and almost my whole future was depending on it since we could not afford another pc and i would had to wait until I grow up and work. I am very grateful to you and appreciate your help. Thanks💚💚🙏

[vocal burrow]

if it's so important, it's worth backing that stuff regardless of viruses

[worn frigate]

https://tenor.com/view/tears-of-joy-happy-tearful-tears-of-happiness-gif-2045793645984542743

[upbeat crane]

very nice

[dense obsidian]

nice to hear there was a happy ending

[worn frigate]

Yes, i did backed it up now. I will not forget this lesson.

[dense obsidian]

and my zluda is running, we are all winning

[worn frigate]

Yes

[upbeat crane]

juice boxes for the whole team

[ornate elk]

Glad you got it fixed 🙂

[dense obsidian]

wth, now I'm running across that torch not able to use gpu nonsense again XD, I just did the pip cache purge

[ornate elk]

You may need to copy the two zluda files onto the right folder again

Torch/lib

[dense obsidian]

ohhh, that sounds right, since I deleted the venv folder, the new copies would have been gone

hm, files replaced, still torch not able to use the gpu

[ornate elk]

Hmm strange, but it did worked before you installed extensions

[dense obsidian]

I think that was when I had the skip cuda core in the bat file

I deleted that after you said not to do it, but that was while I was installing the extensions

[ornate elk]

Can you show your Path settings in the environment variables?

[ornate elk]

There is the hip path missing

[dense obsidian]

ohhhhh, I did not put that in

[ornate elk]

Also ia Strange that you only have 3 entrys there

Normaly windows placed more there

[dense obsidian]

oh there's more, I just wanted to show what I had put in there, everything else is below that

[ornate elk]

Ah you just put it to the top

[dense obsidian]

yup, now to see about putting this %HIP_PATH%bin in

[ornate elk]

That should fix it

[dense obsidian]

spacing out on the adding the path part, do you mean add the C:\Ai\SD-Zluda\zluda to the user variables?

like this?

[upbeat crane]

hmmm so my controlnets still not populating the dropdown in forge after moving them to the models\ControlNet folder in A1111

i know i could just copy them to my forge install but im tryin not to waste the 50gb if i dont have to

[ornate elk]

no

[sand panther]

i tried but didn't work, i guess i need a code to start

[ornate elk]

a code?, whats your gpu and whats inside oyur webui-user.bat

[dense obsidian]

ohhh, just put in the %HIP_PATH%bin, gotcha

same XD

ooo. I put in the wrong zluda folder, apparently there are 2. Got the right one this time.

so far so good

[sand panther]

@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--no-download-sd-model --xformers --no-half-vae --api

call webui.bat
thats all.
I had heard youtube. If i add a code, it could fix. but i dont remember this code. I use pinokio as well. this problem happened when i update it.

[ornate elk]

okay, there shouldnt be a code to fix such issue,
please first rty out a different browser.
if that doesn help, delete the venv folder and relaunch the webui-user.bat

[worn frigate]

Oh this exact same issue had happened with me. In my case, Cmd did not have internet permission. So i had disable firewall. Then in control panel i could see i downloaded necessary files. And it started working.

See ur cmd.

[sand panther]

i tried but it doesn't work

[ornate elk]

@sand panther good advice from alain, check your cmd

when starting the webui and when generating an image

[dense obsidian]

how long did you mention my first generation on zluda would take?

[ornate elk]

around 15-40 minutes

[ornate elk]

and you have the path in the wbeui-user.bat in forge? because i had to add it myself to get it to work

[dense obsidian]

ahhh, time to leave comp on while I get to sleep then :D. Hoping "dog in a hat" is beautiful once I wake up

this is what it should look like in the cmd window though right?

[ornate elk]

your generate button should be gray and say interrupt

[dense obsidian]

yup, all is good then

off to generate output dream.

Have a good one everyone, thanks for all your help @ornate elk !

[worn frigate]

Good day!

[sand panther]

i can download anything. i can access images which i have created as well but i can't see on browsers. no adblock, no access problem. i tried them

[ornate elk]

np, now i have to wait for your reaction on the speed, damn xD

[ornate elk]

did you checked the cmd for any errors?

[sand panther]

there is no error

[worn frigate]

Also I heard about juggernaut lighting. And tried it. So it said 4 steps good image.

But my 4 step image is terrible because I think I am supposed to be using SDXL Diffusion?

I am using SD 1.5 (1.7).

Also if i installed controlnet sdxl models will I be able to use with juggernaut on SD 1.5.

I am asking because I can run SDXL models on 1.5 but it looks like they are meant to be for SDXL Diffusion.

[worn frigate]

Its hard to say.

I advice to Click generate then send a screenshot of what Control panel says ok?

[upbeat crane]

why not use juggernautXL

[ornate elk]

dont get confused between model versions and webui version
1.5 and SDXL are boath Model Versions.
The Webui (the programm to use the models) is on Version 1.7 and the latest update was to 1.8.0

you can use sdxl and 1.5 with the webui on 1.7

[worn frigate]

So there is nothing like SDXL diffusion?

[ornate elk]

nope

[worn frigate]

whaaaa...

[ornate elk]

with auto1111 you can use every model

[worn frigate]

For a month I was in illusion thinking there are Two auto1111.

One is SD 1.5. and other is SD XL.

Sdxl works on 1024x atleast.

So i thought that the SD i have is smaller one 512 one. So i can not use controlnet of SDXL one on SDXL models.

Yes I can us SDXL models. Thought there is new thing.

[ornate elk]

why is the image so cropped?

[worn frigate]

No, this is not command panel.
Its just screenshot of browser. Send the command panel/cmd.

[upbeat crane]

i guess my only two options on this forge controlnet issue is to either duplicate the files over into Forge or create symlinks at this point @ornate elk any other ideas as to why controlnet dropdown still blank after copying my controlnets over to the models\controlnet folder in A1111?

cuz we both def know my webui-user.bat is all set correctly now

and everything else is populating but that

i feel so blessed to be having all these one-off issues nobody else is lol

prob all would have been sorted by the dang win-installer lol

[worn frigate]

Maybe surf web for someone who' s having same issue. You may stumble upon solution. @upbeat crane

[upbeat crane]

i always do that but there isnt anything on this

tried the only solution i found but zonked

[worn frigate]

Whats issue in short.

I will try to ask ai.

[sand panther]

it is not cropped. it is created like this 😆

[upbeat crane]

the dropdown box to the right of my controlnet preprocessors, labeled MODELS, where all the control nets should be is empty

[worn frigate]

And what are you using

[upbeat crane]

webUI forge

[worn frigate]

Ok it says there is not much infomation on this issue on web. And it told simple things like checking compatibility of extensions. Or removing extra ones.

Its huggingchat. You can try by yourself too. Turn on Search web when asking.
Ita free and good.

[upbeat crane]

yepp my luck lol, i just reinstalled yesterday and have disabled all extensions

well actually i have deforum extension enabled still, i'll at least give it a shot at killing that real quick

[worn frigate]

Its like searching whole web for info you want in one command.

[upbeat crane]

ahhh cool no i dont think i've been on huggingchat

[worn frigate]

Disable and re-enable certain extensions to fix the issue. Specifically, some users reported success after disabling the following extensions: After, Tension, and Baof.
Delete specific files such as "v1-5-pruned-emaonly.safetensors" and relaunch the application.
Uninstall certain extensions like Marigold DepthControlNet and TiledVAE as they might conflict with other features.
Remove unnecessary command line flags and allow Forge to handle model loading automatically.

[upbeat crane]

i saw a github thread where illyasviel even saw it but really didnt have anything to say about it

so i figured i was kinda screwed for a bit, since he wrote forge AND the controlnets haha

[worn frigate]

Ye sometimes things dont work.

[upbeat crane]

yeah im so stoked about forge too, like a carrot on a stick for me to get in and do some deforum testing

all good things in all good time, i suppose

[unique urchin]

ok so this is probably gonna be very vague, but i dont know what else to ask to be able to get help. I'm using TurboVisionXL and Tifa Lockhart - SDXL. im running a 4070 but for some reason its taking about 8 mins to generate an image. i've tried different sizes. and there's not much difference in generation speed. initially started at 512 x 768. and if i use upscaler 1.5 then it wont even generate and i get a Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. error.

I tried using SDXL in the past and its beyond slow for me. I'm really not sure what im doing wrong here lol. how are people generating in just a few seconds?

[upbeat crane]

i feel like if i had a 4070 i would be a tad wary of having automatic nvidia drivers enabled

[ornate elk]

can you show me the webui-user.bat again?

[upbeat crane]

yeah

[ornate elk]

hey, whats in your webui-user.bat?

[unique urchin]

``@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS= --no-half

git pull
call webui.bat``

i removed the medvram, as generation would take even longer

[upbeat crane]

the webui-user.bat never lies

[worn frigate]

I tried searching and found this for you.

https://github.com/lllyasviel/stable-diffusion-webui-forge/issues/96

I am not smart enough to understand or guide you. But i hope it resolves by doing what others tried. Check it.

It is same error which you are getting in control panel.

[upbeat crane]

i "think" @ornate elk might have just solved it already lol

dudes a beast

[worn frigate]

And funny amazing thing is he solves half of problem by just asking asking for userwebui.bat XD

For me too webui was wrong lol.

[upbeat crane]

i moved all of my control nets out of A1111 extensions\stable-control net\models and put them into models\controlnet, right?

[worn frigate]

This or vice versa.
One which works for you i guess.

[upbeat crane]

not realizing that i would then be able to use the --controlnet-dir %A1111 linkage now

yeah cs1o's "whats in your userwebui.bat line is like his "have you tried turning it off and on again" routine haha

[worn frigate]

Btw whats forge?

I guess use webui? Auto1111

[upbeat crane]

it is a webUI based on the gui of A111 but cram packed full of useful stuff and fully memory optimized

runs super fast, its nice af.....im in love

[worn frigate]

How much fast it is compared to normal?
110%? 125?

[upbeat crane]

i work with deforum and xl models and sdxl was getting pretty bad to work with

[ornate elk]

you need --xformers --no-half-vae
remove --no-half or change it to --no-half-vae

[upbeat crane]

so forge has kinda been a breath of fresh air for me

[ornate elk]

the lower the vram the better forge performes against auto1111

[worn frigate]

Ty.
Any idea what I am missing when using Juggenaut lighting?
My 4 step genration is not good. Barely any image.
While they say 4 steps can produce stunning image.

[upbeat crane]

yeah i heard that like 4060s and up, the memory and performance increase is negligable

[worn frigate]

It said for me it would be 10% difference. Not worth it.

[upbeat crane]

what card you got?

[worn frigate]

3070

[unique urchin]

i'll give that a try thanks, Should i add the --medvram-sdxl in there too? i see people saying to add it with cards that have 8GB - 16GB vram?

[upbeat crane]

i have a 3060 12gb and the performance is through the roof better than a1111

its not necessarily juist about a speed increase, its about how well it is memory optimized

you should 100% give it a spin tho

[worn frigate]

You do not need to add medvram. Its for gpu around 8 gb ram.
More than 12 ram gpus do not need medvram. It will make it slower.

[sand panther]

thanks your advise. i am gonna look at. i recorded my problem. you can watch

[ornate elk]

what happens if you use 512x512 as resolution?

[sand panther]

same

[worn frigate]

I don't mind because i will be using the photoshop Ui. Not browser. I am using a plugin. And use it inside Photoshop while working.

[ornate elk]

if a sdxl 1024x1024, 30 steps, euler a, image takes more than 30 seconds, you can add it

[upbeat crane]

oh thats cool

[ornate elk]

that could be caused due the wrapper of pinocio

[upbeat crane]

big photoshop guy but finally made the switch over to affinity because im a broke bastard

[worn frigate]

Use one piece version. ||If safe|| and if u like.

[upbeat crane]

ummmm yeahh my problem is fixed

yay thanks buddy

[worn frigate]

Cheeers yay.

How

[upbeat crane]

i had to move my controlnets out of the extensions folder and put them into the normally empty models/controlnet folder in A1111, then create a link in my webui-user.bat "--controlnet-dir %A1111_HOME%/models/ControlNet ^" so that Forge could share the controlnets from my A1111 install. all thanks to the reverend cS1o

[worn frigate]

wow cool

Im happy for you.

[upbeat crane]

thats very kind of you, im stoked

Illyasviel is a stone cold baller

[sand panther]

i can't choose any model now but i could do it 1 hour ago.

[upbeat crane]

closed and reopened your command window and restarted it, right? sorry had to ask

[worn frigate]

I think the reason you are not seeing output image is because of Deforum it said in Error.
And maybe you closed stable diffusion so connection timed out.

Note that I am just a regular person and do not know much.

[upbeat crane]

what browser are you using morpheus, just curious

[sand panther]

i don't know what mean is Deforum. And stable diffusion is not closed. i can't create any image anymore, i can't choose model. something is happen but i didn't anythink. Maybe i must wait until 1 day. Maybe it will fix in time 😆

[upbeat crane]

weird cuz i dont even see the deforum extension on your user interface

*deforum tab

[worn frigate]

See

Try to read it.

[upbeat crane]

ohhh hes using pinnokio

deforum is my main interest as well

i would start shifting away from pinnokio honestly

[ornate elk]

A normal automatic1111 install would work better
Here is my install guide: #🤝｜tech-support message

[upbeat crane]

deforum is a super powerful tool but it actually doesnt require all that much for local install

yeah for sure @sand panther drop pinokio and do an Automatic1111 install

then add the deforum extension there

here is an invite to the deforum discord if you aren't already there

oops blocked dm'd

was just flipping through some history and i wouldn't try to push deforum through pinokio

but the good news is, A1111 is super simple and you got a ton of help whenever you need it

just ask

[worn frigate]

Ye

[upbeat crane]

thats like a niche within a niche so support wouldnt be as forthcoming, via direct and online

[worn frigate]

We can't help much unles they try to help themselves.

[ornate elk]

trying AnimatedDiff rn

its using 20gb vram

[upbeat crane]

in a1111?

[worn frigate]

Oh hey wait

I was curious is there free alternative to runway ai?

[upbeat crane]

kinda funny cuz i am actually starting to play around with SVD right now for the first time

[worn frigate]

To add motion to image. Like if I have a picture simple park photo cartoon style. And i wnt leaves to be moving. Like motion.

[upbeat crane]

text to video on the free?

[worn frigate]

Stable video is limited to low fps.

[upbeat crane]

there are but like limited to 10 seconds i think or less

[worn frigate]

Image to Video.

Just motion.

[worn frigate]

4 seconds.

[ornate elk]

yea that could be done with animateddiff

[upbeat crane]

pika

[worn frigate]

Animate diff is what?
Animated stable diffusion?

[ornate elk]

here is a quick test of it:

[worn frigate]

Its discord based ig. Ye will try surely.
But local install thing would be best.

[ornate elk]

its an extension for auto1111

[worn frigate]

Yes like thiss.

But i hav rtx 3070 only.

[upbeat crane]

like an advanced version of SVD image2video, kinda like pika but really good

[worn frigate]

16 gb ram.

Will it work.

[upbeat crane]

deforum is an absolute blast, i am in love with it

[sand panther]

raise ValueError(
ValueError: An event handler (refresh_models) didn't receive enough input values (needed: 1, got: 0).

how i solve this problem

[craggy coral]

hello guys, i have amd gpu and i decide to download stable broadcast and I have this error.....how can I solve that?

[ornate elk]

i think it could work, here are some examples (if you scroll down)

https://github.com/guoyww/AnimateDiff

[ornate elk]

whats stable boradcast ?

and hey

@craggy coral you should follow one of my two AMD install guides. They are in the Pinned Message of this channel.
Here is the link to the zluda one for Newer gpus (6000 and above):
#🤝｜tech-support message

[craggy coral]

okok thank you

[upbeat crane]

will animateddiff work for what btw

[ornate elk]

for animating pictures

[upbeat crane]

hell yeah, the way to go

[ornate elk]

wow, this test turned out great:

[upbeat crane]

have you played around with SVD yet?

nice, looks great

REALLY dig her...

eyes

[ornate elk]

true xD

they even turn blue lol

[upbeat crane]

maybe give pika a quick shot too, i know they have free renders

but not sure how long, 3 seconds, etc

[ornate elk]

ah still nice to test

[upbeat crane]

fun

i follow a ton of ai video content on youtube and i know ppl have a much better time

using animateddiff with comfyui, i mean if they go beyond just playin around

you can just download the prebuilt workflows for it so its not as crazy as it used to be

you should def start a tik tok dance channel with that girl soon

like by christmas i'm thinkin

i see where this is all going

[ornate gorge]

FFFFF too much

[upbeat crane]

civitai is barely chugging along right now lol

[ornate elk]

not for my AMD 7900XTX with 24gb 😛

but its still crazy, that it takes so much

[ornate gorge]

im really thinking of sending back to amazon 4080 super i got and getting 4090 soon, might hold a bit with that still

[worn frigate]

Thanks. I will try it.
Hopefully it works.
What does it require comfyui?
Whats like render time and max seconds.

[ornate elk]

i dont think your that limited with an 4080, but more vram is always good, we also dont know what SD3 will need

[ornate elk]

can work in auto1111 and comfyui
it can take a lot of time

[ornate gorge]

sd3 8b model fits onto 4090 they said on website

but there might be some optimalisation

later on

[ornate elk]

that would be great

the community will handle it haha

[worn frigate]

Sure, will try right now.

[upbeat crane]

https://media.giphy.com/media/v1.Y2lkPTc5MGI3NjExZmw0MXltZmFrNXJyZHkzeDE2MTY3dDhkbmQxOGE5azdvYnFmN2kzdCZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/3og0IOO2IhZ9XtGow0/giphy.gif

every pika video

[worn frigate]

I am thinking like These AI are using a fraction of power Sora seems to offer.

It astonishes me.

[upbeat crane]

soras fuckin nuts huh

[ornate gorge]

wait untill we get hands on sora, advertising and hands on experience are not always the same

[upbeat crane]

one minute

we will finally be at the age where like any 16 yo kid can make a blockbuster movie

its going to be crazy cool

[worn frigate]

Its gonna be paid though...

Not for someone broke like me

I will wait for its alternative

Which i don't know how long will take.

[upbeat crane]

ai video generation is moving a light speed rn and its a blast watching unfold

[worn frigate]

I tried it.
It does exactly what I was asking for but kinda 4 seconds.

Also i love locally installed Ais. Maybe animatediff would be better choice in my case.

[upbeat crane]

download the 54b parameter unrestricted minstral LLM and run it local next Alain

it will help you develop underground party drugs in your basement

you will get rich

[worn frigate]

How long does it take usually thats question though.
@ornate elk how long did your last renders took with your gpu and ram

[upbeat crane]

https://media.giphy.com/media/v1.Y2lkPTc5MGI3NjExcTloMXB3NTgwcmowZGVoZWNiMTJibjYyOXFiOWtpNjlyOW90dTVvZiZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/iDh1vAFaEWwjS/giphy.gif

[worn frigate]

I think I am using mistral 8x7 ai.
Its on huggingchat.

[upbeat crane]

it is hahaaa

[ornate elk]

on my GTX1080 a few months ago it took like an hour or more. (maybe less idk)
now on my AMD gpu with 24gb vram it takes 2 minutes

[upbeat crane]

me and my nephew installed it on his server a couple weeks ago, its insane

[worn frigate]

For a rtx 3070 how long would it roughly take

According to your idea.

[ornate elk]

i think maybe 30-45 mins max

[worn frigate]

30-45 mins for 1 video?

[upbeat crane]

oh i meant the dl is on huggingface but yeah probably

[ornate elk]

yes

[worn frigate]

[ornate elk]

but maybe im completely wrong on that and it could take you like 10 minutes

[upbeat crane]

how long would waht take, again?

[ornate elk]

AnimatedDiff 24 frames

[upbeat crane]

to learn how to cook meth

ohhh yeah sorry

[worn frigate]

Its long time XD but i cant complain

I will use Pika now🤣

[ornate elk]

if you try it out let me know the time ^^

[worn frigate]

Ok I will definately try and see.

[upbeat crane]

it takes me about 2 hours to render one min using deforum at 1280x720@30fps on a 3060, then i upscale with topaz local

[worn frigate]

Then check If its worth for me seeing its time aspect. Or i will have to use online ais

[upbeat crane]

totally worth it lol

[worn frigate]

Bro topaz gives me nightmares

[upbeat crane]

whys that

[worn frigate]

Ah......

Lets just.. forget

[upbeat crane]

bro my dad owns topaz

[worn frigate]

It is a crazy story .

[upbeat crane]

jk, now i gotta hear this

[worn frigate]

I would dm you i think.

Ok?

[upbeat crane]

seems fair

[queen plank]

There's not enough memory when it comes to the DWpose node. I've got 16gb of ram. Here's the error: Error occurred when executing DWPreprocessor: [enforce fail at alloc_cpu.cpp:80] data. DefaultCPUAllocator: not enough memory: you tried to allocate 9953280000 bytes. File "D:\new_ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 152, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

[ornate gorge]

ram or vram?

[queen plank]

RAM

[ornate elk]

then you need to make your pagefile larger

[queen plank]

it was using 15 gigs of it then tried to use 9 extra gigs

What do you suggest I change it to?

[ornate elk]

min 16000 and max to 24000
and then make sure you have 15-20gb free space on C

[queen plank]

alr, thanks

I'll try that and update you guys here

Alright, it's running. should be about 3 minutes till I see if the fix worked

Error occurred when executing LoadImagesFromDirectory:

[enforce fail at alloc_cpu.cpp:80] data. DefaultCPUAllocator: not enough memory: you tried to allocate 23788339200 bytes.

[fair oxide]

hello, is my AMDRadeon RX 6800 XT good enough for the desctop installations? becasue it always fails befor finishing

[upbeat crane]

16gb of vram you should be good yeah

open the pins here in this channe for some AMD related installation assist

@ornate elk wrote em i think but some good stuff to get you started

[ornate elk]

yes, use the zluda guide, it has the best performance for that gpu

[upbeat crane]

going to name my next chihuahua zluda

[magic hollow]

so automatic1111 was going good the first few days, but then it started crashing my pc like how comfy did. i think the problem is my old amd card keeps overheating and crashed but im not sure if it can be fixed or f there is an extension that might be able to help? so idk if stable diffusion can be used at the moment.

[hazy seal]

looking to upgrade my 3080 to something with more VRAM. AMD seems to be the best bang for the buck but i've read they're no good for Stable Diffusion because they don't have CUDA support. Thoughts?

[fair oxide]

i dont understand point 3 under getting zluda to run.

should i make a folder and insert the unpacked zluda folder in it? or the files from the unpacked zluda

[upbeat crane]

you have to install GIT, create a folder

then once git is on your computer, you use the GIT CLONE command to pull the contents into the folder

[ornate elk]

the files from it

[upbeat crane]

https://git-scm.com/book/en/v2/Getting-Started-Installing-Git

[fair oxide]

okay thanks

[upbeat crane]

if you need a little basic info youtube is a great place, just search, install git. its simple

[ornate elk]

he meant step 3 of zluda

not of the preparation steps

[upbeat crane]

oh my bad

[ornate elk]

whats your gpu?

[magic hollow]

rx 480 amd and yes i did the guide aswell.

[upbeat crane]

think he has a AMDRadeon RX 6800 XT

gunna go crawl back into my dumpster

[ornate elk]

okay, thats an old card true,
there is this extension you can try
https://github.com/w-e-w/stable-diffusion-webui-GPU-temperature-protection

if its really heat related

[worn frigate]

@ornate elk whats lighest iso i can get for vmware workstation?

Which i can use to check if sonething is shady.

[magic hollow]

im hoping thats the problem, its the only problem i can think of that might be happening , also how old is this extension?

[upbeat crane]

tinyos (tiny11)

you are in a never ending spiral tho if you torrent an os so that you can install vm so you can test sofware haha

actually that looks good alain

https://www.minitool.com/news/tiny11-download-iso.html

then pop that into your vm

looks adorable

[worn frigate]

I mean i can download it

TinyOS.

Bur what tells its safe?

[ornate gorge]

and what does it saves you, a bit of ram?

[upbeat crane]

it does look free i would go for it

i had forgotten about tinyos

[worn frigate]

Like why would we trust this website

[upbeat crane]

you gotta trust somebody, somewhere, sometime bruh lolol

[ornate elk]

6 months

[magic hollow]

thats a bit old but it should still work?

[worn frigate]

No bro, No after today.

Yesterday.. Yes i would have trustrd.
But today no.
I am in nightmare.

I will only use official OS if i can find somewhere.

[upbeat crane]

weren't you the villian in that story tho

hehe jk

[ornate elk]

i think so, but make also sure your pc case is dust clean and maybe turn up the case fans

[worn frigate]

Lol.

[magic hollow]

you have a point, will do.

[ornate elk]

why do you need 1.7 ?
but yes its possible

okay, ive done it too 😄

what you need to do is to clone the repo into a folder. but make sure its not in the same as the current webui.
You go into the folder then click in the File Explorer bar (not searchbar) and type cmd then press enter.
Then you copy and paste this command: git checkout 77c0e5f2e7f6e1f81f3a7477484a03f521465f7a
then edit your webui-user.bat like in my guide and run it.

then its 1.7.0

i didnt knew so i made it "every user proofed" ^^

nope, you would need to find the commit :/

[magic hollow]

umm so i got an error when relaunching the ui...
[Error GPU temperature protection] OpenHardwareMonitor Couldn't find temperature sensor for Radeon (TM) RX 480 Graphics

[ornate gorge]

fucked up temp sensor on gpu seems like

[ornate elk]

okay, then it wont work with your gpu :/

[magic hollow]

is there way to fix it or am i in trouble with my card.

[magic hollow]

well dang it

[ornate elk]

oh wait

[magic hollow]

whats weird is it runs text generation fine.

[ornate elk]

there is the openhardwaremonitor support

you would need that tool

to make it work

[magic hollow]

oh that makes more sense now.

do i just install it or is there a certain process?

[ornate elk]

i would say just installing and running it one time should be enough

[fair oxide]

i did mess soemthing up. i cant create anything. one thig is "fatal: No names found, cannot describe anything."

[ornate elk]

are you done with the guide?

what settings did you tried?

[fair oxide]

i think so

[ornate elk]

full error log helps too

[fair oxide]

venv "C:\SD-Zluda\stable-diffusion-webui-directml\venv\Scripts\Python.exe"
WARNING: ZLUDA works best with SD.Next. Please consider migrating to SD.Next.
fatal: No names found, cannot describe anything.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Version: 1.8.0-RC
Commit hash: 25a3b6cbeea8a07afd5e4594afc2f1c79f41ac1a
WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for:
PyTorch 2.1.2+cu121 with CUDA 1201 (you have 2.2.0+cu118)
Python 3.10.11 (you have 3.10.6)
Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers)
Memory-efficient attention, SwiGLU, sparse and more won't be available.
Set XFORMERS_MORE_DETAILS=1 for more details
A matching Triton is not available, some optimizations will not be enabled.
Error caught was: No module named 'triton'
C:\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(
You are up to date with the most recent release.
Launching Web UI with arguments: --xformers --autolaunch --use-zluda --medvram-sdxl --update-all-extensions --update-check --skip-ort
C:\SD-Zluda\stable-diffusion-webui-directml\venv\lib\site-packages\diffusers\utils\outputs.py:63: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
torch.utils._pytree._register_pytree_node(
Loading weights [e3edb8a26f] from C:\SD-Zluda\stable-diffusion-webui-directml\models\Stable-diffusion\ghostmix_v20Bakedvae.safetensors
Creating model from config: C:\SD-Zluda\stable-diffusion-webui-directml\configs\v1-inference.yaml
Running on local URL: http://127.0.0.1:7860

To create a public link, set share=True in launch().
Startup time: 4.9s (prepare environment: 8.5s, initialize shared: 0.6s, other imports: 0.7s, load scripts: 0.6s, create ui: 0.3s, gradio launch: 0.4s).
Applying attention optimization: xformers... done.

ohhh i see i need xformes

[ornate elk]

no

xformers is for nvidia

ohhh you added xformers in the webui-user.bat

thats why it wont work

remove that xD

and then you need to delete the venv folder

and then relaunch

[fair oxide]

ahhhhh yea sry

[ornate elk]

my guide said nothing about xformers xD

[fair oxide]

another guide dit it and i thought it was also here sry

do i need to redo point 4?

from the zluda part

[ornate elk]

yes

[trim raven]

What determines how long it takes for the first prompt to be generated?

[ornate elk]

idk, because it barely uses cpu and gpu and just blows the fans until its done

[trim raven]

wow I went from 1.6it to 6it👍

As improvements continue, there will be no point in switching to Nvidia.

[ornate elk]

nice! what is your gpu?

[upbeat crane]

nvidias market cap is now bigger than canada at 2.2 trillion dollars and AMD is at 340 billion

https://media.giphy.com/media/v1.Y2lkPTc5MGI3NjExcTB3amY2M3lkZWpvdDFlZjk0dzU2MGRrMTN2emk5NnBiZmdncDdueSZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/83QtfwKWdmSEo/giphy.gif

[trim raven]

6950 XT

[upbeat crane]

just thought i would add that fun fact

[ornate elk]

ah nice, enough vram and speed now for everything, even sdxl should work good

[vocal burrow]

step up your waifu game canada !

I mean ...

[cosmic furnace]

Y ive been told to try to... on sd next server 🙂 ... few days ago maybe🤔

Wazzup new & good with zluda ?

[ornate elk]

Yes zluda on Directml works perfectly

[river flame]

I tested
--xformers
set CUDA_MODULE_LOADING=LAZY
set NUMEXPR_MAX_THREADS=24
set PYTORCH_CUDA_ALLOC_CONF=garbage_collection_threshold:0.9,max_split_size_mb:512
in A1111, it was slower, went back to just --xformers, rebooted and now it generated twice slower then before testing... any ideas?

[old parrot]

Is there any risk for using --share to run SD on my network? Or should I run it through Docker to be safe?

Do not want hacked, but probably minimal worry

[karmic crown]

if your shared instance is not password protected, anyone can use it, and if that happen when you want to use it too, you might need to wait for your turn on your own computer 🙈

[old parrot]

Right, more worried about any way to infiltrate my PC

[vocal burrow]

probably want to use --listen instead if you want to be safe and keep it to your LAN

[karmic crown]

and there are some other ways to make it accesable from remove even with --listen

like some remote desktop app (rdp, vnc...)

[river flame]

nobody has any advice?((

[karmic crown]

i am on Mac, sorry

i am using Windows only when I really need to compile some of my legacy projects (windows applications)

[river flame]

can I reinstall just the system folder in A1111? wont it ruin anything? I dont know what else to do

[karmic crown]

you can try to remove venv folder, a1111 will create new on next start

it might help, try

[river flame]

should it be it in webui folder? I searched for venv and it found nothing. I have a 1-click installed A1111

[karmic crown]

in that case i cant help, sorry

[upbeat crane]

if you have a good amount of of models, loras, controlnets, you def dont want to wipe those

just because you woould have to redownload, prob easiest to just create a new stable-diffusion-webui folder

install to that, then you can move your contents back in place if you reinstall

and name back to the original folder name, just an idea if you are going to start over

[river flame]

thanks

[karmic crown]

better rename the original folder to something like a1111-bkp

than do a fresh install

copy/paste (or move files from old to new)

[upbeat crane]

yeah that works, i wouldnt delete anything at all until after you get a new ui up and running

[karmic crown]

it is easier like this, since you can follow the exact steps you did when you installed previous time

without need to worry about the name of new install

come to the dark side (Mac), i made script for easy installation 🙂

[river flame]

I wanted to try stability matrix, but the download link is 404

[river flame]

ok I installed a new A1111 and now back to normal generation speed

[upbeat crane]

right ON, right on

there for a while when i would reinstall i would get everything set back up

get all my extensions back installed and things back to where i liked

then i would create a 7zip backup file of the folder, save it

then move all my checkpoints, models, loras, etc back into place

kinda quit doing it because it really just doesnt take all that long

and everytime you reset, you learn new things

[quick forge]

well, I spent a bunch of time re-padding my 5700XT, and the mem temps are outstanding but for some reason the GPU keeps overheating instantly and shutting down.
I can't seem to figure out why it's overheating with a new thermal pad, and/or enough thermal paste to fill the gap of a thermal pad.

[cosmic furnace]

better than sd next?)

[ornate elk]

Better ui for sure xD

Performance is the same or better

But only if you dont use the cache mode of sdnext

[old parrot]

Better than SD Forge?

[ornate elk]

Zluda ? For sure!

[old parrot]

Even for Nvidia GPU? I'll test it out

[ornate elk]

No

Zluda is for AMD

[old parrot]

Gotcha I figured

Thanks!

[old parrot]

Do you know how I can get a link that doesn't change when using --share arg?

[karmic crown]

you can use --listen with reverse proxy, or port forwarding and static ip address

[old parrot]

I have both args there with 6666 port forwarded, the link digits change every time

[karmic crown]

that will be always random as far as i know

[old parrot]

Is this the proper order of args? Or does it matter

set COMMANDLINE_ARGS= --share --gradio-auth deez:nuts --listen --port 6969

[karmic crown]

yes, but you need to add other paras there too, like --xformers, or whatever you need for fastest a1111 performance

those are just for sharing

[old parrot]

I'm using SD Forge, when I have those args it tells me to remove them on boot

[karmic crown]

oh, ok then

in that case its probably fine

[old parrot]

Appreciate you! Hopefully there's a way to create a static link

Time to Google

[karmic crown]

i like forge, but it works slower on my mac 😦

a1111 works the fastest for me 🙈

[cosmic furnace]

@ornate elk what that cache mode do? 🤔

[old parrot]

A1 takes 4 minutes for me on 512x512, interesting stuff

1660 super though

[karmic crown]

i have 2it/s with a1111, and 1.5it/s with forge and comfy 🙈

[old parrot]

What GPU? I'm looking to upgrade next week

[karmic crown]

i have no GPU

i have macbook pro with M3 PRO 🙂

[old parrot]

Nice! Sounds epic

[karmic crown]

my mac is slow compared to MCurtos Mac, that thing is crazy fast M2 Ultra 192GB RAM 🙂

[ornate elk]

You already tried it

The additional Performance boost

But with that some stuff won't work on SD nexr

[cosmic furnace]

Will amd be better now with ray tracing on zluda? 🙂

[ornate elk]

With a 7900xtx its not even that bad

[cosmic furnace]

y kinda almost but on nvidia still better?)

[ornate elk]

Sure

[cosmic furnace]

and the problem is in soft optimization i guess

[ornate elk]

I dont think zluda helps much there

As Raytracing is based on rt cores

[cosmic furnace]

and if some dude said he had blender render running better on zluda than it makes sense can happen with RT?))

[ornate elk]

No

Raytracing is based on RT Cores and this is hardware that can't be emulated with zluda

[cosmic furnace]

😦

is ray tracing an nvidia thing ? o.0

i mean they made it?) Huan?)

Only 2-3 games of all of them out there. Have significant better fps with RT now. IT means smthing

Like they got shitty optimisation or got paid by nvidia 😄

Espec Cyberpunk with latest updates...

[ornate elk]

Yes Raytracing is an nvidia thing

So it works the best on their hardware

[karmic crown]

can someone on windows open CMD and type cd in any directory, and tell me if it print just the folder name, or full path

[solar geyser]

Can confirm this too.

[karmic crown]

it should be equal as pwd on macs and linux, but i do not have windows at the moment

❯ pwd
/Users/viking/stable-diffusion-webui

and pip in venv on windows in here?
venv\scripts\pip

its hard to help someone with windows, when you do not have it yourslef 🙂

[ornate elk]

Ahh, what was your GPU again?

Ah okay, and did you tried a 512x512, 30 step, Euler a, image without loras and all. And less than 75 tokens?

Ah okay

I think the vram management alone is worth using zluda xD and the speed is the nice addition

[cosmic furnace]

@ornate elk Uv got no issues using zluda whatsoever? Espec recently?)

Often zluda updates stopped 😦

[ornate elk]

It would work maybe with tiled vae enabled

[ornate elk]

No issues so far 🙂

Even animated diff worked

[cosmic furnace]

So i guess its just me huh ? 😒

[ornate elk]

Do you have any issues?

[karmic crown]

https://github.com/viking1304/a1111-setup/releases/tag/0.0.6

[cosmic furnace]

img2img not working for starters... thic deep cache mode starting to fill up to point where it says no more generation cuz cache full =\

overall less issues)

lately

[karmic crown]

i am gonna get crazy with deforum and controlnet

[cosmic furnace]

linux rocm 6.0.2 when will windows version catch up 🫠

[ornate elk]

Yea but thats SD.Next. I'm using directml with zluda

And for me everything works

[cosmic furnace]

btw even with control net it doesnt seems to understand 3 characters very close to each other

if the one between are badly seen

it seem to generate only 2 of 3

was hoping cntrlnet more accurate...or im doing smthing wrong hmgm

y mb shud try directml)

[random pagoda]

i wanna use high res like 2048x2560 for gens any way to push up the sliders?

[ornate elk]

You can edit the ui-config.json

[ornate elk]

You should definitly try it

[cosmic furnace]

but control net still not full there y?)

[ornate elk]

I think its fully there

IP-Adapter works

I only need to test depth

[vast crest]

Hi everyone, sorry to bother you, but I would like to know how to increase the size of the preview while creating the images in sequence

[solar geyser]

@ornate elk So I'm back. Sorry to be a bother, but I wanted to continue our efforts.

I'm getting a runtime error when I launch webui.bat. It says "invalid argument to memory_allocated". I'm assuming this is in one of the files somewhere, I just didn't want to crack open Python and muddle around if someone knows specifically what to do.

[ornate elk]

Hey, when do you get that error?

[harsh tendon]

Question: I am contemplating a laptop to run SDXL and similar, I would like to "future proof" it by at least being prepared for the next gen of it (3), is the requirements public yet? because I haven't found any--

[solar geyser]

When I try to generate

[ornate elk]

Can you show me the Settings you used?

Resolution etc

[solar geyser]

Sure standby

[solar geyser]

I can screenshare if that's easier and show you what I'm doing, but here's a screenshot.

At this point, I'm just testing to see if my installation works, not too concerned about quality or anything else yet. So I let everything stay default. If this is just a setting problem, then please excuse my ignorance.

[ornate elk]

You need to recheck your environment path settings and that you copied the right files into torch/lib

[solar geyser]

Took the two files from the Zluda folder again, renamed as before and copied them over. It's possible that, somewhere in the mess I either didn't move the files or I was remembering a previous installation. Because now, it's taking its time while generating. No errors yet. Will update you when it finishes. I recall you saying it could take a few minutes the first time.

[ornate elk]

Ah okay. Yes let it load and dont interrupt it if the generate button is Gray rn.
It can take 15-40 minutes

[solar geyser]

Will do. It currently just says "Applying attention optimization: Doggettx... done"

[ornate elk]

If the generate button is Gray everything is good

[solar geyser]

❤️

If this works, I"ll be back to let you know and thank you properly lol.

[ornate elk]

Sure pls let me know! As its late here I'm going off for now

[solar geyser]

Have a great night, thank you so much for helping despite the time difference.

[ornate elk]

No problem 🙂 gn

[quick forge]

would anyone know why ForgeUI keeps "cloning" itself into a stable-diffusion-webui folder within it's own folder?

[white stirrup]

generation speed drop despite exact same prompt and settings

wait what

UI restart seemed to have fixed it

no it didnt

lyCORIS is bottlenecking it

it seems to be the extention itself

[river eagle]

OK, FINALLY pulled the trigger and upgrading my gpu from the sad (but good old workhorse) RTX 2070 Super 8gb to a brand spanking new ASUS TUF RTX 4070 Ti Super 16gb! It should arrive in a couple days. What follows is how I had to tweak the webui-user.bat to get SA to run on the 2070 without freezing and constant crashing (at least not as much). Any suggestions for how it should read after I install the 4070?...

git pull
@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--medvram --no-half-vae --xformers --autolaunch

call webui.bat

[crisp prawn]

trying to figure out why i always get the import failed error on load: 0.0 seconds: D:\Ai\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_IPAdapter_plus
0.0 seconds (IMPORT FAILED): D:\Ai\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Stable-Video-Diffusion

[vale estuary]

I need to add the 4x-ultrasharp.pth file to the esrgan in the models file, but there is no such file, I don't understand what I need to install, can you help me if there is a guide video or what should I do?

[old parrot]

Create a new folder named ESRGAN within the Models folder and add it there

[minor aspen]

How to add AnimateDiff to forge?

[old parrot]

https://github.com/continue-revolution/sd-forge-animatediff

[minor aspen]

I saw that, but the tutorial is listed as TODO

[vale estuary]

That's what I did, I hope it works without any problems. I've been searching for this on the web for 4 hours. Finally, I found this solution. It's a bit primitive, but I hope it works.

[old parrot]

It works I did the same the other day

[ornate elk]

You need to remove --medvram for the 4070

[hoary plaza]

quick question how would i go about adding different models?

like waifu diffusion etc

[sterile garden]

I randomly get results like this from random XL models. It's fine again for a while after restarting the instance. It seems to happen after changing models a lot. It happens with different models, but once a particular model makes this result, it will always make results like this for that particular model until i restart. Any ideas what it could be? I've tried enabling/disabling different optimizations, changing samplers, cfg, etc, nothing seems to help.

[ornate elk]

You can download models and loras on Civitai.com
Its the largest model database.
Models (2gb or bigger) go into models/stable-diffusion
Loras go into models/lora

[hoary plaza]

cs1 do you have a guide on comfyui?

[ornate elk]

@hoary plaza

[hoary plaza]

gotcha thanks thanks

ohhhh thats what you meant by models for civitai

i thought models meant like

people dancing

[ornate elk]

Haha xD

Models = Checkpoints

In terms of SD

[ornate elk]

There is a link to comfyui in the Pinned Messages.
Installation is as simple as downloading it as zip and start

[hoary plaza]

gotcha gotcha thanks!

[ornate elk]

Hey, thats an sdxl model bug that can be fixed with this extension:
https://github.com/hako-mikan/sd-webui-prevent-artifact

[hoary plaza]

shouldnt i get stable diffusion first before i download comfy ui?

man this going to take me a bit to read all this haha

[dense obsidian]

From my experience (about 2 weeks into stable diffusion), stable diffusion is a good starting point if you are just getting into a.i. art generation. There's a lot to take in and the info you gain from stable diffusion will help explain a lot of what you'll need to know for how to work comfyui.

[hoary plaza]

okok i see thanks thanks

is there a way for my system to have both a1111 and comfy ui?

and have it somewhat organized

[karmic crown]

you can have both, with all models inside a1111 folder structure, and just use them from comfy

[karmic crown]

install a1111 add models, loras...
install comfy and find file extra_model_paths.yaml.example
rename that file to extra_model_paths.yaml
find line 7 and change base_path: path/to/stable-diffusion-webui/ to path where your a1111 is
do not forget to add / at the end

you can customize it further, but that is usually not required if you a1111 models are in default paths

[hoary plaza]

oh i see thanks thanks

[ornate elk]

You need to download that file. Then move into models/esrgan if your using Automatic1111

Try delete the venv and relaunch

[ornate elk]

Nvidia is still the recommended GPU for Ai stuff because of cuda.
But AMD works also with their cuda equivalent called Rocm.
Rocm with Stable Diffusion only works on Linux for now. But AMD is working on the windows support.
On windows AMD GPUs can work with Directml which is slow.

But now there is the Cuda Emulation tool called Zluda which was released a few weeks ago.
It emulates cuda to work on AMD Cards on Windows.
With that we now have significant performance boost on newer AMD GPUs.
With zluda my 7900XTX is now a beast in image generation

[little echo]

anyone know how to get z123 working in forge?

[meager berry]

Hi everyone! Noob here... My outpainting misteriously stopped overnight (from the logs it seems an hard crash). Anyway, is there a way to restore the work done?

[river eagle]

Keep the rest? Anything to add to COMMAND? Any new lines maybe? I want to start training it and make art based on my own work. I need A1111 to be the best it possibly can be.

[ornate elk]

Nothing to add.
With --xformers --no-half-vae your ready

[worn frigate]

@ornate elk

can i dm you once

I can not write here, and have a short thing to ask.

Its okay if you say no.

[ornate elk]

Sure

[paper crag]

anyone here dealing with this error?
'RuntimeError: The GPU device instance has been suspended. Use GetDeviceRemovedReason to determine the appropriate action.'

[ornate elk]

Where do you get this, webui, GPU,

[ornate elk]

Whats your GPU?, what does the cmd shows?
What did you tried

[paper crag]

this is during generation before it crashes and it gives me that error

[ornate elk]

1.5 models are not really intended as refiner or better not necessary
Try without it.

Also I see your on a 6800.
I would recommend you to use the Zluda directml webui because it will work much better in performance, vram management, and is able to get higher resolution done and upscaling easily

[ornate elk]

And also dont use batch size in directml, only batch count

That will reduce the vram usage alot

Less likely to crash then

[sterile garden]

Hey, thanks for the reply! I'll give it a try

[paper crag]

so i should open the webui.bat and add '--Zluda' in it?

[ornate elk]

Its a bit more than that needed for it to run. But its totally worth it.
You'll find my complete guide in the Pinned Messages of this channel.
Or direct link here:
#🤝｜tech-support message