🤝｜tech-support

[errant thistle]

I just installed this too a few days ago >_<

[muted scarab]

idk about automatic

but the lora were hidden cause they weren't compatible with your SD i think

check on civit or whatever what vers you're meant to use with each checkpoint/lora

[vocal burrow]

by changing your gpu.

vram is soldered onto it

[muted scarab]

😱

[errant thistle]

It isnt "base model" that normally tells me what its compatible with is it?

[vocal burrow]

also loading sdxl models pretty much require at least 16gb of ram

[muted scarab]

might be gg

I have 16 but old pc

looks like it

[errant thistle]

I dont think changing a gpu should be an issue? If it isnt gonna overload it or anything i think price is your biggest concern.

[muted scarab]

ye but im lazy

[errant thistle]

Thought i was told base model had nothing to do with automatic version... Ill just have to look into it more.
Fair enough. Idk if you can manage on 6gb vram maybe. Could cause a strain though

[muted scarab]

i don'twant to break it more cause i use it for everything and i want to play games mainly

[vocal burrow]

ok so if it s a 1.5 based model using --medvram command line arg should be enough with 4gb of vram

[solemn moon]

pls guis for muh family, can we get a canny flux auto

[half island]

Not really an "issue", but how do i make a comfyui install i have completely contained, where even a freshly installed windows can just launch a bat file, and it'll load and launch all python related it needs within it's venv?

[sonic ice]

oh.. ok. just got directed here

[grim moon]

Heyyyy

I have this error when i load any model (I just downloaded sd). Does anyone know how to fix it? Thanks in advance.

[noble belfry]

Theres a nice zluda guide pinned to the channel, Teeles

#🤝｜tech-support message if its hard to find and youre on mobile or whatever

[stiff mountain]

out of curiosity, can I rename the lora and models with easier or more intuitive names for me or do I risk compromising their effectiveness?

[ornate elk]

What's your GPU ?

[muted scarab]

1050

[ornate elk]

That can work with SD

4gb vram is the minimum, but it also can work with 2gb

[muted scarab]

but SDXL

[ornate elk]

For that you need to enable FP8 mode in the Automatic1111 Settings

And increase your Windows pagefile

[muted scarab]

if i just rename it model.cptk or whatever it is will it replace regular SD and work

[ornate elk]

You don't need to rename anything

You can select the models in the dropdown on the top left

[muted scarab]

but it say SDXL but also a checkpoint and a lora

[ornate elk]

Just make sure they are .safetensor files and at least 2gb

[muted scarab]

on a starting point i want to use

[ornate elk]

Sdxl is like 1.5 a base model.
You can download any community made model that's based on sdxl

[muted scarab]

yeah but how will it know to use it

i thought the way it worked was it uses whatever is named model, which i think rn is SD 1.10.1 for me and then up to 1 checkpoint and any lora you select

[ornate elk]

Don get confused between model versions and webui versions.
SD webui (automatic 1111) is a tool to use models. Its version is 1.10.1

Sdxl and 1.5 are base models that can be loaded into the webui itself

[muted scarab]

so they are checkpoints?

[ornate elk]

Models = checkpoints

[muted scarab]

im loading up the thing again that i want to replicate

how can you select 2 checkpoints?

[ornate elk]

Loras go into models/lora folder

[muted scarab]

ik

[ornate elk]

You only can use one at a time

Select it in the dropdown

[muted scarab]

on civit they named another checkpoint other than 1.5 or sdxl but said both could be used to replicate what they did

[ornate elk]

Pony ?

Pony models are sdxl based

[muted scarab]

1 checkpoint, 1 lora

both pony

and the lora didn't show up in the folder, i brute forced it using the setting and then ran it and it froze my whole pc and i had to restart it

and they recommend sdxl or 1.5 as i mentioned before

BUT how can i make it show up in lora folder

ohhh, by using sdxl ig

ill try

[hasty orbit]

Are there any tools/scripts that would allow me to change out prompts automatically? For example, every hour Stable Diffusion will generate 10 images, and then the next hour it will generate 10 images with a different prompt.

[muted scarab]

ok guys, how do you select both sdxl checkpoint and refiner 🤔

[ornate elk]

It should show up when selecting an pony model

[timber oak]

@ornate elk hey bro sry for ping. You have helped me many times already. I kinda need help again. My friend is trying to install sdweb ui. RTX4070 and Win11. we got python 10.11, git and downloaded the sd webui linked. we also did the userbat editing so its all set up. whenever he launches the user.bat, it opens a window and instantly crashes.

I appreciate any help

🫂

aaand i found the error😭😭

bro forgot call webui.bat

😭

[ornate elk]

Hey, He has to use the webui-user.bat

[timber oak]

yes he did

but he forgot to paste it inside

so he only had git pull

[ornate elk]

Ahh well xD

[timber oak]

ye

its installing now ^^

[timber oak]

bro are you familiar with comfy ui?

i am thinking of trying it out

am using automatic 1111 rn

[ornate elk]

I have a comfyui install guide on my github yes.
But for questions about the usage we have #🧣｜comfy-ui

[timber oak]

kk

thx

🫂

[ornate elk]

Np

[azure magnet]

Someone can help me to install FaceId of ipadapter?
I need help

[crimson grove]

Since the last time ive used SD Months ago, Ive installed a Second GPU so i could have more monitors. Now, when i use Upscaler im getting a long error. i don't know which gpu its trying to use, hopefully its going to GPU 0 which is my original better GPU, But now im getting this below.

Without the highres fix upscaler it works fine

[foggy drum]

Has anyone tried using the Stability API? We tried to connect it to unity and we kept getting this error message. We tried tweaking it a few times but I still get the same response.
request Response: {"success":false,"message":"Malformed FormData request. No initial boundary string (or you have a truncated message)."}

[balmy matrix]

So doggone confused. Yesterday I generated images in automatic1111 in 45-50 seconds. For most of today after I set up comfyui I got that down to 20 seconds. And then a few hours ago, whenever I started to generate images, after no settings changes, it started to take two minutes or longer. I booted up automatic1111 to see what it does, and it also takes two minutes. What could have possibly happened? I don't have any games running or the like. :<

[ruby yoke]

suggest checking your sytem logs for memory issues and make sure you're not running anything in the background. also make sure that youre not trying to generate with the CPU instead of the GPU

[balmy matrix]

"An error was detected on device \Device\Harddisk4\DR4 during a paging operation," a warning from the System log in Event Viewer.

I have 125 GB free on that hard drive , 132 GB of total virtual memory, 92 GB of available virtual memory, but only 3.84 GB of available Physical Memory.

[ruby yoke]

it tried to write and coulding. how much paging space do you have?

open a ticket in the #1010934719455707218 channel

tough. the mods ignore people like you

our trolls won't though

if you want them to ban you, just open a ticket

[balmy matrix]

100 GB of page file space

[ruby yoke]

sigh @elder turtle

why do you want discord to ban you?

yeah, discord, cause that's who i reported you to

[balmy matrix]

Surely 100 GB of page file space is enough, but idk.

[ruby yoke]

i would expect that it was

i'd start looking for disk errors

[timber oak]

cool

another idiot

[ruby yoke]

see, no one's going to ban you

we're goign to make you lose that bet

[balmy matrix]

Yeah, I also noticed my computer will freeze for a second during generation once or twice. That did not occur at all before.

Somethin' is wrong

[timber oak]

damn

[ruby yoke]

pull the system logs up and start looking for disk errors - also check and make sure you're not overheating

[balmy matrix]

All I got were 3 of these: An error was detected on device \Device\Harddisk4\DR4 during a paging operation. Everything else says Volume is healthy, no action needed.

[ruby yoke]

and not what the error was. i hate windows error logging

you might consider getting a copy of diskkeeper and running it

[balmy matrix]

what is that

I didn't think SSD drives don't really have serious issues anymore. And ya definitely don't wanna defrag it

chkdsk reported 0 bad sectors.

finally got everything to work today and now it's all ogre :<

[ruby yoke]

everything mechanical falls apart.

i'd be trying to use a different drive. i can think of a number of things that could cause write issues that arne't the drive itself - starting with the port you have it plugged into, cables, etc

[balmy matrix]

alright, I'll try moving everything over to a different drive

[ruby yoke]

it's paging. you'll need to move your paging space

is THAT on the ssd?

[balmy matrix]

paging space is split across my 3 main hard drives, but I could remedy that

[ruby yoke]

do that

[balmy matrix]

pick a hard drive and do it all there?

[ruby yoke]

put it all one drive, preferably the system drive

[balmy matrix]

Most of it is already on the system drive

[ruby yoke]

put the rest of it on the system drive

and 1. do a shut down, leave off for 10 seconds, then boot up FIRST

then fix your paging space

and then do the same shut down again

not a restart

[balmy matrix]

alright, boss, will do that.

moment of truth, let's see what happens

[ruby yoke]

i'ts not happy with your version of cuda

[delicate urchin]

what is this error about. My cuda gpu works fine in all other besides reactor

[ruby yoke]

it specifically says the version isn't what it wants - just follow up on all the things it says - right version, in the path, and that your GPU can use it

[balmy matrix]

now... nothing happens. I press the generate button, and... nothing transpires within auto1111 or the cmd prompt for webui-user.bat

[delicate urchin]

all looks very complicated. It wants me to install 1.14 from this site https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#cuda-11x but the site has no download for that version and I am afriad changing cuda like this will mess up other nodes/programs

[leaden tiger]

this is a long shot but do you have python Python 3.10

[balmy matrix]

I have both versions. I restarted my PC and it started up.

Images generate in 30 seconds now.

[leaden tiger]

nice

welcome to the party

[balmy matrix]

Guess I needed to do a full boot down/boot up, then shift my page file.

It's been non-stop troubleshooting for the past few days lmao

[leaden tiger]

it's okay, i tap my head twice and jump on one feet everytime i generate a new batch

we all have our rituals

[balmy matrix]

The Machine Spirit must be appeased, and constantly.

[leaden tiger]

PRAISE BE!

[balmy matrix]

I do have incense.

[ornate elk]

Are you on AMD ?

Or Nvidia

[delicate urchin]

nvidia

[ornate elk]

Then make sure your GPU driver is updated to the latest version

[delicate urchin]

ok I will go on nvidia driver website and get the latest one

lets see if it workds

[delicate urchin]

same error

did a install of new 560.94 and restart

[ornate elk]

Can you try update comfyui and all nodes ?

[grim moon]

Hey, how to change lora intensity?

[lost crescent]

In which folder should I place Hyper-FLUX.1-dev-8steps-lora?

[lost crescent]

... strength model ...

[grim moon]

That doesnt tell me anything

[lost crescent]

When you open a LoRA Loader - you should see "strength model"

0.1 is LOW Strength

And 1.0 is HIGH strength

[rugged latch]

peanut

getting this nonsense when tryna use the --share

i dunno why

ive never had dis issue til now

hm

seems to be a problem with my shitty ISP

so nvm ig

[past mural]

Help pls

[rugged latch]

well

according to the error

it seems there is no error

ur good to go :3

XD

jk

[past mural]

?

idk what should i do

i cant generate any img

[rugged latch]

i dunno

what did u change

if anything before the error

try to retrace ur steps

those are the worst errors

the kind thats either blank or says unknown error

pft

[past mural]

I didn't do anything, just downloaded the tutorial and uploaded the model((

[rugged latch]

is every thig installed

pytorch, python, are u using the right python

[past mural]

3.10.6 ?

i used it

[rugged latch]

hmm

[past mural]

Should i download pytorch?

[rugged latch]

it should have installed automatically

when ever u first launch stable

[past mural]

i think

[rugged latch]

youll see downloads going on in the termnial

[past mural]

i should reinstall it again

[rugged latch]

ya try that

[fathom rose]

is there any way I can include multiple models folder using command line args?

for example this

[past mural]

not work

[lunar jolt]

how i solve this?

[karmic crown]

no, because that parameter expect exactly one value

[amber cloud]

Hello everyone, i had Python 3.12 installed on my machine. un-installed and i installed the 3.10.6 to run Automatic 1111. when i run the webui-user.bat i get this message (No Python at '"C:\Users\ali\AppData\Local\Programs\Python\Python312\python.exe'). even though, i checked (add to path) when installed the 3.10.6. Any ideas please !?

[fathom rose]

There is no way I can add another directory?

[ornate elk]

Change your DNS Server

[ornate elk]

Hey, delete the venv folder and relaunch the webui-user.bat

[ornate elk]

You need to make SymLinks

[amber cloud]

Thanks!

[lunar jolt]

can somebody helpme??

[queen marsh]

Well since my computer didn't work before with stable diffusion
And I'm now planning to buy a new laptop I wanted to ask if the laptop I'm gonna buy will be able to run it, laptop stats:
Dell G3 15 3579 (8th Gen) – Intel Core i7-8750H Gaming Laptop Proc (6-Core, 9MB Cache, up to 3.9GHz W/TB) / 16GB DDR4 SDRAM 2666MHz / 256GB SSD / 1TB HDD / 15.6-Inch FHD (1920 X 1080) IPS / 4GB GDDR5 Dedicated Graphics NVIDIA GeForce GTX 1050Ti

Tryna mainly run comfy ui

And teach the ai how to copy artstyles

[fair oxide]

Heya guys, I used to use stable diffusion on GTX GPU but now I have an AMD Radeon GPU, I cannot get it to work...is there an AI like stable diffusion (or even a form of SD) that I can continue to use?

[ruby yoke]

Read through the amd guides pinned at the top of this channel

[ornate elk]

First link of the pinned messages, You'll find my guides. Feel free to ask when you have any questions.

[dim tree]

thank you so much for the list of guides

i was wondering if anyone made a benchmarks pin? I am in the process of buying a budget GPU for SD and it would be a tremendous help

in truth I already chose the RTX 3060 12GB because well, only budget GPU I could find that has higher than 8GB VRAM, I am interested in training loras and other goodies with it
but then comes the question of OS compatibility, since nvidia has notoriously shaky linux support and I daily-drive linux
if i were to buy said RTX 3060, would I be better off using windows or linux for stable diffusion?

[rugged latch]

yay it worked

i set to 1.1.1.1

[vocal burrow]

no but you can create symbolink hard link instead https://www.tenforums.com/tutorials/131182-create-soft-hard-symbolic-links-windows.html

[crimson grove]

Since the last time ive used SD Months ago, Ive installed a Second GPU so i could have more monitors. Now, when i use Upscaler im getting a long error. i don't know which gpu its trying to use, hopefully its going to GPU 0 which is my original better GPU, But now im getting this below.

Without the highres fix upscaler it works fine

[lunar jolt]

can somebody help me with this?

[ornate elk]

Can you provide a full cmd log ?

Also what 2 GPUs do you use and what's inside your webui-user.bat?

[lunar jolt]

when i try to create video with this it's saying always text2video extension for auto1111 — version 1.2b. The video will be shown below this label when ready

where i can find the version of auto1111 1.2b

[onyx solar]

I'm not aware of a "1.2b" versioning for a1111. (All the releases are here: https://github.com/AUTOMATIC1111/stable-diffusion-webui/releases) If you were to grab an old enough version of a1111 to match the commits for the extension you're looking at, which would be about 7 months ago, you'd find yourself on version 1.7.0. I'd imagine you're just reading that wrong and it's version 1.2b of the extension itself.

[ornate elk]

Try update the extension. And then check on github how old the extension is

If its not updated in the last 5 months. It mostly wont work without some manual fixes

[twin wharf]

"Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver " how do i fix this on rocm

[twin wharf]

hey man sorry to ping but ik your good at this stuff could u help me out please?

[ruby yoke]

you buy an nvidia GPU?

[twin wharf]

i dont have one

i use rocm

on amd

[ruby yoke]

yes, well - you're going to need one if you want to run the program that's telling you it can't find one

[twin wharf]

ive had this error before and i fixed it yet i dont remember how

you can run stable diffusion on rocm]

this is well known

[ruby yoke]

didn't say you couldn't. but the error you are getting is specifically looking for an NVIDIA gpu

[twin wharf]

yeah, i need a way to get rid of this message.. i think it has something to do with the torch version installed

yet im not techy enough to fix this

[ruby yoke]

yeah. pytorch needs cuda and cuda needs nvida

[twin wharf]

oh

how do i uninstall pytorch

so i can install rocm

[ruby yoke]

you'd have to uninstall torch vision

[twin wharf]

how do i do that like whats the command

[ruby yoke]

go to google and search for "how to uninstall torch vision"

[twin wharf]

thanks

[ornate elk]

What's your GPU ?

If your on AMD you need to follow my install guide.

[twin wharf]

im on 6950xt

im using linux so zluda wont work

im tryna use rocm

[ornate elk]

Oh Linux

[twin wharf]

yar

i might just reinstall linux and see if i can do it then

[ornate elk]

Yea would be an idea.
Also checkout AMDs guide how to setup Rocm correctly

[brittle lantern]

Any news on Flux running on AMD GPUs with 8gb of vram (forge and NF4 or sum else) if possible any tutorial?

[ornate elk]

Nope. Best bet would be rocm on Linux or docker with WSL 2 on windows. But idk if any 8gb card could run it.
NF4 works only with bitsandbytes installed. AMD rocm on windows doesn't support it rn

[brittle lantern]

F ty man

[brittle lantern]

rocm on linux does?

[ornate elk]

Yes but its currently in alpha/beta experimental state

[brittle lantern]

i see. idk if its worth in general to dual boot for linux for SD

is it any faster with ROCm on linux vs zluda on windows?

[ornate elk]

What's your gpu?

[brittle lantern]

6600 xt

[ornate elk]

Hmm

[twin wharf]

No module named 'safetensors ' i keep getting this error with rocm

[brittle lantern]

as of now its 6600 xt in a couple of months i may get a used 3090 but i do not know yet if im gonna, theres a chance i do a full pc build with a 4090 but again not 100% yet on that

[ornate elk]

If you familiar with Linux and like to tinker a lot to get it running you can try. But if not I would stick with Zluda.
The new flux model has now a gguf model version and that maybe runs without the need of bitsandbytes.
So in future it could work easily on 8gb. But no guarantee

[twin wharf]

nvm just needed to install requirements

[ornate elk]

Windows, Linux ?

[twin wharf]

pip install -r requirements.txt

yeah

i solved it now

just gotta wait and see if anymore errors

[brittle lantern]

got u so should i just keep win 10 and zluda and when i get the 4090 build or 3090 gpu ill be set to use flux properly

[timber oak]

Hey

i keep getting this error

and idk how to fix it

[ornate elk]

Yep. With a 24gb GPU you won't have problems and maybe in the time it gets more optimised too

[ornate elk]

Would need more info. Like GPU, OS, and the full cmd log

[brittle lantern]

yh and also gives me a massive gaming perf boost

[ornate elk]

True

I'm running a 7900XTX with 24gb

No problems in any game ^^

[brittle lantern]

what cpu should i run 7800x3d or 7900x or intel chip?

[timber oak]

the one piece is real

4070 win 10

[brittle lantern]

doing a platfrom switch rn on R5 5600 (AM4)

[ornate elk]

Definitly the 7800X3D if you go for pure gaming performance.
Anything with more cores if you use render stuff or use it as workstation

[brittle lantern]

got u and 32gb of ram or 64gb of ddr5 (64 kits are expensive)

im also looking at daul channel not quad

idc for looks i want perf on the pc

[ornate elk]

Best would be a dualkit 2x32gb with 6000mhz and low cl

[brittle lantern]

cl28 to cl32 range?

[ornate elk]

I'm running 2x16gb g skill expo 6000mhz with cl30

For AMD, 6000mhz and expo are perfect

For Intel CPUs you can go lower but I would not recommend anyone buying Intel rn.
Their 13/14 gen CPUs have a massive problem

[timber oak]

this is the first module that breaks

[brittle lantern]

750w estimated usage shown on pc part picker so i go 850w or 1000w?

[timber oak]

and all of these too

all other work perfectly

[timber oak]

1k might be overkill

i would personally go for 900

i currently have 850

[ornate elk]

For a 4090 you need a 1000W,
for a 3090 I guess 850 could work

[timber oak]

but tbh

you can go for 1k if you wanna futureproof

[brittle lantern]

do i need pcie5 PSU or any good 1000w will do?

[ornate elk]

Its comfyui. Can't help much with that.
But try ask in #🧣｜comfy-ui

[timber oak]

ok danke. Ill wait for their answer then

[ornate elk]

Any 1000w and it should be modular (which they mostly all are)

[timber oak]

yeah

so far i am pretty happy with my rog strix 850W

just my RAM

😭

[brittle lantern]

https://uk.pcpartpicker.com/product/RHfnTW/nzxt-c1200-1200-w-80-gold-certified-fully-modular-atx-power-supply-pa-2g1bb-us this price is quite nice

[timber oak]

i have 2 times the same ram

but they have different manufacturers

so i cant use xmp bc it crashes my pc every 2 hours

now i have 64 gb ram with 2133MHz speed instead of the 3600MHz

[ornate elk]

1200 is a bit overkill. But NZXT is good. Also BeQuiet is my recommendation

[timber oak]

https://tenor.com/view/spongbob-crying-sad-gif-4246972

[ornate elk]

Thats sad

[timber oak]

yeah

but i bought a pc from my friend

for 10 eur

[ornate elk]

A good deal xD

[timber oak]

has a broken motherboard, intel i stoneage, ddr3 ram,450W noname psu and a gtx 960

this is it

basically i will only use the case : D

which im gonna buy a ddr4 ram compatible shitty mainboard, a cheap compatible cpu. buy ram so i can use xmp.Put my old ram in it and use my old 450W bequiet

and instead of the gtx 960

ill use my old gtx 1060 6gb

[brittle lantern]

£25 diff from 1000w to 1200w

[timber oak]

dont overdo it bro 🫂

[brittle lantern]

ik but the price

[timber oak]

hmm

[brittle lantern]

UK psu price is kinda mid

[timber oak]

ye

[timber oak]

have you googled?

[timber oak]

on the effects

if it actually swallows electricity

bc usually it shouldnt take more than it needs

i know that windows and amd are really power hungry bc they try to keep all processes running at max efficiency

even if they are in the background

[brittle lantern]

yh fair ig

right 1000w it is

[ornate elk]

Nope AMD CPUs need a lot less energy than Intel cpus

[timber oak]

rly?

[brittle lantern]

4k 144hz or save money and go 1440p 144/165hz? both 27 inch IPS panels

[ornate elk]

Yep

[timber oak]

ok thats actually surprising

would you take 1k or 1.2?

[ornate elk]

You should checkout the tdp of the 7800x3d ^^

[timber oak]

cs1o do you have a steam deck?

[ornate elk]

I would use the PSU calc and then see how high the usage is under full load

[ornate elk]

Yes

[timber oak]

which one?

i bought the 512 gb one

[brittle lantern]

£178 for 1440p 180hz IPS panel monitor from MSI thats really solid price no?

[timber oak]

with that special glass

[ornate elk]

The 256gb

[ornate elk]

Thats a very good price.

[ornate elk]

Nicee

[brittle lantern]

https://uk.pcpartpicker.com/product/RktLrH/msi-g272qpf-e2-270-2560-x-1440-180-hz-monitor-g272qpf-e2

£379 for cheapest ips 4k 160hz from MSI

this isnt so bad

[timber oak]

ye

[brittle lantern]

yk ill go 1440p as i am still a comp gamer

[ornate elk]

Dont go for 4k

[brittle lantern]

prolly wont do 4k often either

[timber oak]

damn

i tried

but games like valorant/cs or cod are just not for me : D

[ornate elk]

Wqhd is the sweetspot of quality and performance

[brittle lantern]

https://uk.pcpartpicker.com/list/vR6PjH @ornate elk hows this massive upgrade looking?

[timber oak]

league is kinda fun

bc path of exile is my favourite game

[brittle lantern]

was immo 3 on val and was pro league challengers on FaceIt back in CSGO

[timber oak]

is that good?

i cant tell : P

bc i am a noob

[ornate elk]

Looks solid!

[brittle lantern]

immo 3 is one rank below Top 500 in the region and i was basically online pro player for CS

[brittle lantern]

ty

[timber oak]

nice

thats huge

i am plastic V in league of legends

[ornate elk]

But why micro atx mainboard in an atx case ?

[brittle lantern]

ye now im tryna get Top 250 in CoD warzone

[brittle lantern]

iwl micro ATX cheaper than atx but ill swap it out

[timber oak]

atx has better heat distribution

than micro atx

[brittle lantern]

https://uk.pcpartpicker.com/product/CvcgXL/gigabyte-b650-eagle-ax-atx-am5-motherboard-b650-eagle-ax swapped it for this

[timber oak]

every capacitor has their own place to sleep on the bed

imagine having to cuddle in summer

[ornate elk]

Ah okay.
Also I don't like Gigabyte (they are not good with BIOS updates), Would recommend MSI or Asus. But not needed if your fine with it.

[timber oak]

my atx board has some blocks to protect the capacitors around the cpu

and the one M.2

[brittle lantern]

https://uk.pcpartpicker.com/product/szfxFT/msi-b650-gaming-plus-wifi-atx-am5-motherboard-b650-gaming-plus-wifi swapped it again

hope that mobo is better

[timber oak]

i think mine is called Aorus V2 sth sth

i forgor

[brittle lantern]

i may get nvme should i get 2tb or 4tb?

[timber oak]

hmm

if you have enough slots

id say smaller is faster

[brittle lantern]

(not my only drive just gonna be another i buy)

[timber oak]

put windows on one nvme

and all your games on the other

for super fast gaming and system programs

stable diffusion etc

[brittle lantern]

ill have 3tb total for nvme if i buy a 2tb nvme

[ornate elk]

Thats nice. I'm using a MSI MPG B650 Carbon WiFi.
Both are good

[river plover]

Would also choose something different then the Gainward 4090, Tend to have a lot of coilwhine from that producent. Zotac, Msi, Asus rather

[brittle lantern]

i have a 2tb hdd for non important files rn

[timber oak]

and in the other ssds you put files and stuff

[timber oak]

good

[ornate elk]

2tb is much cheaper and you can always buy a second if needed

[timber oak]

NEVER EVER DEFRAGMENT AN SSD

it drastically reduces the lifetime of it

[brittle lantern]

https://uk.pcpartpicker.com/product/BQXJ7P/pny-xlr8-gaming-verto-epic-x-rgb-geforce-rtx-4090-24-gb-video-card-vcg409024tfxxpb1 this better?

[brittle lantern]

ik

[timber oak]

🫂👍

[brittle lantern]

ill grab a sn770 2tb for a £100 then

[ornate elk]

Nah gainward would be better than pny

[river plover]

i don't have any personal experience with pny so cannot say about that card

[ornate elk]

But Asus and MSI would be better than gainward

[brittle lantern]

ill just get ventus x3 then

[ornate elk]

But gainward is nice. But maybe read up a bit on that

[timber oak]

i have a zotac 4070

very funny

they sent me stickers and a tiny "pappaufsteller" aka paper figurine

[brittle lantern]

https://uk.pcpartpicker.com/list/H2LqcH should be final? any changes needed?

[river plover]

i have returned 3 gainward RTX cards, (they used to make good cards back in the days when they made the HD3870, HD4870 )

[brittle lantern]

yh

i could also scrap this whole build and spend £580 on a refurb 3090 and call it a day xD

[ornate elk]

My GTX1080 was great from them

[river plover]

I do understand the "want to get the best" but sometimes going for the cheapest options just to get it build is also not the best idea.
So if spending 580 on a 3090 and that will cover all you need now, might be the best option and save up more cash to spend on a pc at a later upgrade.
that would probably be the best option

[brittle lantern]

yh and if i do need more perf i can just get a 5700x3d and 64gb of ddr4 ram for £288 ish total

looking at that "upgrade" that i made before is just a new pc build

[river plover]

yeah i swapped over to Nvidia founders edt at the 10x series. i think the 4090 is the first "newer" card i bought from another vendor and ended with 3 returs of gainward, then got the asus rog strix and have been happy with that for now

[ornate elk]

Looks good

[brittle lantern]

https://uk.pcpartpicker.com/list/2KQMRK like this @river plover

[river plover]

Both options are good, it all comes down to how much you want to spend tbh. so the Final CS1o replied to and this one.

wait, missed the psu. 750w is abit low

would go atleast 850w on that

[brittle lantern]

i think for now ill do the 3090 one ill run it with my 5600 and see how it goes if i actually need more cpu power then ill buy new cpu cuz i dont mind spending £2k+ but in also 19 and i will be kinda "hard" getting £2k+ back

[river plover]

for 5800x3d (5700x3d is not in the calculator) and a 3090

don't need to go for bequiet psu, but an 850w one

[brittle lantern]

https://uk.pcpartpicker.com/product/zF4Zxr/msi-mag-a850gl-pcie5-850-w-80-gold-certified-fully-modular-atx-power-supply-mag-a850gl-pcie5 swapped it for this then

yh finacially wise 3090 option seems best for me

[river plover]

scan has an offer for https://www.scan.co.uk/products/850w-corsair-rmx-series-rm850x-fully-modular-80plus-gold-single-rail-708a-135mm-fan-atx-psu?utm_source=pc part picker&utm_medium i prefeer Corsair over MSI but both are known brands (except GamersNexus got MSI to lit on fire)

[earnest cedar]

hola

[crimson grove]

That is the whole thing aside from 2 lines where it is showing that it's generating normally. Originally I had a 1070 which worked fine. Now that I've added the 1660 super just to have more monitor outputs it's giving this issue. Ofc I want it to still run on the 1070, not sure how to make sure if it is or not

[manic matrix]

Is anyone online to help me rn

It’s basically saying that I don’t have git installed and I do

Python, git, typing cmd in the ai folder, and when the terminal is brought up and I copy paste the command it says “git not an internal or external command”

I did

I’m thinking I only have git desktop not 64 bit

That was the issue I only had the desktop git and not 64 bit

I hope I don’t run into more shit

I’ve tried doing this ten times now 🙂

Am I supposed to open the batch file next that says webui-user

I’m going to cry maybe it’s working

In case it works, where do I go to install the proper models I want to use?

Isn’t it called like civ or something

Do you know where the links to that are? Thank you

What do you use?

It says in the cmd it’s already downloading pruned emaonly safetensors v1.5

Is that good

Okay what is a base and checkpoint

[ruby yoke]

a checkpoint usually covers a wider range of information than a lora does

[manic matrix]

Wait, so I didn’t download a good base model?

What requires less specific prompting to get it to work

I want it to be as user friendly as possible

I’m not a tech wizard obviously

[ruby yoke]

you did. however you're probably going to want to go to civitAI and look for checkpoints and loras for 1.5 - after you work with 1.5 for a bit and get a feel for how it thiks

[ruby yoke]

you want to work with AI - you're going to get to turn into one, like it or not

[manic matrix]

At first I just want to create realistic people, I’ll figure out other kinds of things eventually but that’s my main priority

[manic matrix]

Epic

[ruby yoke]

if you don't want to do that, you need to run your image gens on someone's website so all you ahve to do is give it a prompt - because running it at home is going to require you to turn into a programmer of sorts and learn a lot of technical stuff

[manic matrix]

I’m fine with that

As long as I have help

🥲

[ruby yoke]

you've picked the absolute hardest of the interfaces, just an FYI - you'd be a lot happier with comfyUI

[manic matrix]

..:

[ruby yoke]

we're right here and will help as much as we can

[manic matrix]

I wasn’t given any advice on what to pick

It was the first option in the guide

So I clicked it

[ruby yoke]

it's fine. you can learn how to use it and you'll be fine with it, just expect a learnig curve

[manic matrix]

Maybe I’ll download that too idk

[ruby yoke]

learn the one you have first

[manic matrix]

Alrighty 🔥🥰🔥

Not burning down my house or anything

[ruby yoke]

once you get a place to actually put in a prompt, just put in the word apple and generate

see how long that takes on your machine

[manic matrix]

Alrighty

I’m still downloading

Gonna take like an hour or something

[ruby yoke]

be aware that stable diffusion 1.5 is happiest with images that are 512x512

[manic matrix]

Good to know

[ruby yoke]

yeah, some of the models are huge

hopefully you have enough drive space

[manic matrix]

What’s the best graphics card for this?

For nvidia

[ruby yoke]

the best - welllllll you really want an intel processor and the top of the line home Nvidia GPU - so at least a 4060

as fast as stuff is changing though i wouldn't be buying anything new right now

not unless your gens are glacial in speed

[manic matrix]

😟

[ruby yoke]

you asked. but work with what you have. the hardware field is changing as fast as the software field

[manic matrix]

Alright I’m getting scared

But I’ll try

As long as it doesn’t take ages to do anything I’ll be fine

[ruby yoke]

there was a time when i'd kick off a render and come back a week later to see if it was done...

[manic matrix]

Whaaaaaaaaaat

[ruby yoke]

was a long time ago. i doubt you'll have renders that slow

[manic matrix]

Is 50 gigs enough space

Or do I need to buy a stupid ass ssd

[ruby yoke]

nope. not even. you need to keep about that much free for system and other things. unless you're installing to some drive that's not your system drive

[manic matrix]

No I mean

On top of everything else installed on my pc

This is my second drive

I still have 50 gigs

[ruby yoke]

https://www.newegg.com/sandisk-extreme-v2-1tb-usb-3-2-gen-2-usb-c/p/N82E16820173503 these aren't that expensive

[ruby yoke]

that's going to be about one model...

[manic matrix]

??????

This model is 4 gigs

What ssd do you use?

[ruby yoke]

the one i gave you the link to

[ruby yoke]

sd1.5 is small. some stuff can get very large.

[manic matrix]

Would I be able to trust the 130 dollar 2TB at Walmart

[ruby yoke]

probably. but i still by my stuff from newegg

[manic matrix]

Why specifically from them

Also, is 1 TB enough?

[ruby yoke]

cause tigerdirect went out of business and they're my second choice

[manic matrix]

Also also, would I be able to move what I have running right now to my dad after I buy it

Fuck

SSD

Not dad

[ruby yoke]

i dunno. probably not without reinstalling. not sure how webui handles stuff getting moved around

[manic matrix]

Jesus Christ I hate computers now

Ten minutes ago I loved them

GET MORE STORAGE BOZO 🤡🤡🤡

GET BETTER SPECS BOZO 🤡🤡🤡

BECOME A CODING GOD BOZO 🤡🤡🤡

JUST TO SEE SOME PIXELS ON A SCREEN

Now that I’m done losing my mind, does that ssd come with all the cables it needs to work

[ruby yoke]

the ones i got did. small USB cable. it plugs into a usb port. and the one i got came with a universal adapter in case my usb ports were the updated versions

[ruby yoke]

ther'es always using mage.space. they ahve free accounts, with unlmited generations

[manic matrix]

Why isn’t anyone talking about it lmao

[ruby yoke]

most of the people here are using their own machines - because most of the people here are more into tech stuff i guess

[manic matrix]

Lol anything mature requires a subscription for that’s good to know

I think I know why more people are running their own machines now

[rare summit]

how do i fix this?

No Python at 'C:\Users\PC\AppData\Local\Microsoft\WindowsApps\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\python.exe'

[untold hollow]

thank you so much this worked!!

[spring pike]

tried to ask for help before but none has helped, my problem is when i make ai and put the hires fix too anything but 0 my whole pc freeses when the image is at like 99% and then its done it unfreezes

plz help

[remote elm]

Need help; I'm stuck here, the last thing that I did was update some extensions.

[ornate elk]

What's in your webui-user.bat?

[ornate elk]

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

[ornate elk]

Uninstall python 3.9 and install python 3.10.11 64bit (and not from windows store).
Then delete the venv folder and relaunch.

[ornate elk]

You paused the webui by clicking into the cmd.
Press space to unpause it or relaunch.

[remote elm]

Thanks. I just left the window open, did nothing, and just fixed itself.

[fair oxide]

help, please - Ive been following the guidelines for AMD GPU installation but every time I reach some error concerning torch. Ive finally fixed that and now, this. Please help, thank yo for any support you can give

if it's no use and should not proceed, let me know 🙂

[spring pike]

iv got a ryzen 3060 and is it this ""C:\Ai\stable-diffusion-webui\venv\Scripts\Python.exe"" ?

[ornate elk]

Hey, can you show the full cmd log?

[ornate elk]

No I mean what's inside the webui-user.bat when you right click and edit it

[spring pike]

i dont see edit

[ornate elk]

Windows 11 user? Then click on more options

[spring pike]

ok i got to edit

[ornate elk]

You need to add --xformers to the line commandline_args=

[spring pike]

mine looks like this "set COMMANDLINE_ARGS= --xformers --no-half-vae --medvram"

[ornate elk]

You have a 3060 with 12gb vram right?

Then you dont need --medvram

[spring pike]

im pretty sure my vram is 16gb

[fair oxide]

of course, here - reran it today because other screenshot was from yesterday when I gave up! :

[ornate elk]

Nope a 3060 has either 8 or 12 depending if its a mobile or PC version

[spring pike]

i can look at vram in task manager or no?

[ornate elk]

What guide did you followed and what's your GPU again?

[ornate elk]

Yes under performance tab then click on gpu

[spring pike]

my gpu is a nvidia gefore rtx 3060

[ornate elk]

Look at dedicated memory when clicking on the GPU

[spring pike]

ok

16.0 Gb on memory

[ornate elk]

Don't mix ram and vram xD

[spring pike]

oh my b

so what do i do?

[ornate elk]

On taskmanager click on the performance tab. Then click on the GPU. Then scroll down and you see dedicated memory

[verbal fractal]

Hello I am a fooocus user. The command tells me that I am in lowram mode. I would like to use all the capabilities of my computer to speed up the process. Can I configure RAM usage for Fooocus? How to do it? Thank you so much

[ornate elk]

Hey, what's your GPU?

[spring pike]

8.0 gb on dedicated gpu memory

[ornate elk]

Alright then change the --medvram to --medvram-sdxl
And then save and close

[spring pike]

alright is that it?

[ornate elk]

Relaunch and then set hires steps to 10 and try again

[fair oxide]

the AMD one pinned, as well as this video, when torch wouldnt work (I forgot to install python and git - new computer 😑 )

GPU is AMD RADEON RX 6600

[spring pike]

alright thank you

[ornate elk]

Please follow the latest guide from my link. The other one is a bit older.

The link is the top link on the pinned messages

[verbal fractal]

Nvidia geforce GTX 1060

[fair oxide]

ah got it, will do

[verbal fractal]

6043 MB

[ornate elk]

Okay, with that GPU you shouldnt use Fooocus. Its not optimized for 6gb vram GPUs

As its only using sdxl models

You should use Automatic1111 for the best experience/performance

[verbal fractal]

Ok thanks

[ornate elk]

Let me know if any step isnt clear

[verbal fractal]

I tried with others SD itération but the result wasnt as good as fooocus

[timber oak]

is onyx used in normal sd aswell?

[ornate elk]

Its model dependent and prompt dependent. With Automatic1111 you have much more settings to Generate what you want. Also it has an fp8 mode for sdxl models.
With that you can generate much faster

Follow my install guide from the pinned messages to get it running with the needed performance arguments

[timber oak]

it says i have onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : LoadLibrary failed with error 126

is there a place i can see what error 126 is?

[ornate elk]

When do you get this?

[timber oak]

its still the same issue i had 12h ago

i tried to solve it

but its not getting any better

[verbal fractal]

I dont understand. If i use Automatic1111 why i would use SDXL model ??

[timber oak]

people on the internet are just confused as me from what i read in forums

[verbal fractal]

Actually i use SDXL model because i cant use others

[timber oak]

mostly just forums being closed due to inactivity

didnt get any answer in #🧣｜comfy-ui yet

[ornate elk]

In Auto1111 you can use any model you want

[timber oak]

if your gpu can run it SDXL is cooler

[verbal fractal]

Ok

[timber oak]

bc it is made for 1024x1024 images

unlike sd with 521x521 native

i personally recommend pony diffusion

or if you want very detailed stuff check out flux

(i havent yet)

[verbal fractal]

So i understand that Automatic1111 + SDXL will get me same results but faster ??

[timber oak]

i have all my stuff focussed around pony

[ornate elk]

Yes

[verbal fractal]

Nice

[timber oak]

automatic is just the program

[verbal fractal]

Yes

[timber oak]

like comfy ui

[verbal fractal]

I tried to use it

But it was more complicated and worst results

[ornate elk]

Important is that you have --xformers --medvram --no-half-vae
In your webui-user.bat and that you enable FP8 mode for sdxl in the settings

[timber oak]

fp8?

what was that again?

[ornate elk]

Its float point precision, models are either fp16 or fp32 but fp8 reduces vram usage a lot

[timber oak]

🫡

ok thats good

[timber oak]

apparently error 126 means it has missing permissions

but when i launch it as admin

i get this

so if i lauch normal it works, but launching as admin crashes it

well...it doesnt start at all

[spring pike]

@ornate elk the first image i made didnt freeze me but when i made a second one it did

[ornate elk]

Can you show me the txt2img settings you used ?

Like resolution, steps, upscaler etc

[tender jay]

Hi, I'm trying to install Comfy manager but it just stop loading my ComfyUI after a while for no reason.

[spring pike]

i took ss of it and its not in my folder 🤨

[tender jay]

Can anyone help me fix Comfy Manager?

[spring pike]

@ornate elk I dm you a pic of the stuff

[ornate elk]

Do you have git installed?

[tender jay]

I have it but I don't know how to use it. I didn't install Comfy using Git.

[ornate elk]

Try installing the comfyui manager using git clone like its stated in their guide on github

[tender jay]

Where is the Git page I got the comfy manager from Civitai.

[ornate elk]

Here:
https://github.com/ltdrdata/ComfyUI-Manager

[tender jay]

Do I have to use Git to download it? Can't I just copy the files myself to where it has to go? I don't really understand how to use the Git application.

I don't see the instructions on how to use this?

[onyx solar]

You don't typically use that application window at all. Once git is installed, you use a command line window and follow the instructions on the extension page to install the ComfyUI Manager. If you follow the link you were given and scroll down, there are instructions that walk you through it.

[tender jay]

so how to I tell it where to install on my PC? I don't want my files downloading to my C drive??

[onyx solar]

Are you asking where to install ComfyUI Manager or Git itself, because it seems like you already have Git.

[tender jay]

I just want to install Comfy manager. I don't want to have to deal with Git at all.

Is there a way to do this with out Git.

[onyx solar]

Git is a component that is required by ComfyUI Manager to install and update all of the add-ons. Once you have the manager add-on working, you won't have to worry about using git manually.

[tender jay]

But how to I tell Git to copy the files to my Z drive? How do git know where to put the files? There's not setting interface for me to tell it where to place download files? There's no instructions?

[onyx solar]

Ok, back up a second.

Where do you have ComfyUI itself installed?

[tender jay]

On my Z drive

[onyx solar]

Ok. So do you have the page for ComfyUI Manager open in your browser?

[tender jay]

https://github.com/ltdrdata/ComfyUI-Manager?tab=readme-ov-file

this right

[onyx solar]

Yes, that.

[tender jay]

I'm on it.

[onyx solar]

Scroll down to the section that says Installation.

[tender jay]

I'm there.

[onyx solar]

Ok...step 1 in the first installation instructions shows you exactly where all the stuff goes.

Open a terminal/command line window.

[tender jay]

this?

[onyx solar]

Yes

[tender jay]

so I can just download the zip and follow that?

[onyx solar]

No, just follow me here...

Open a terminal/command line window

[tender jay]

you mean CMD?

[onyx solar]

Yes. CMD is short for command

[tender jay]

ok

[onyx solar]

When you opened it, I assume it put you in your C: drive if you look at the text where the blinking cursor is?

[tender jay]

yeah it's on my c: drive

[onyx solar]

ok, so type in Z: and hit enter

[tender jay]

done

[onyx solar]

That should change the drive you're on

[tender jay]

yeah I'm on z now

[onyx solar]

Is your ComfyUI folder directly on Z or in another folder?

[tender jay]

one sec

it's in some folders

where should I change the path to?

[onyx solar]

Ok, so go find the full path for where ComfyUI is and type in CD followed by a space, followed by the full path and hit enter

[tender jay]

should I change to the path that has the "run_nvidia_gpu.bat"

[onyx solar]

Ok, so in that folder, do you have a ComfyUI folder where that bat file is?

[tender jay]

this is where I changed to.

[onyx solar]

Go into that ComfyUI folder at the top

Then you should see a custom_nodes folder inside that one.

[tender jay]

so change my command prompt there?

[onyx solar]

Yeah, you want to be inside the custom_nodes folder as shown in instruction 1 for installation.

[onyx solar]

Great.

Now copy the command from instruction 2 and put that in there

[tender jay]

??

I don't see any command in instruction 2?

[onyx solar]

Scroll up

[tender jay]

sorry wrong one.

[onyx solar]

When it's done, if it worked correctly, you'll have a ComfyUI-Manager folder inside the custom_nodes folder.

[tender jay]

ohhhh I think I see what's wrong

[cerulean hemlock]

i keep getting this error while training a model anyone know why

[tender jay]

the video I watch had me move the files in the Custom nodes not the folder for it.

[onyx solar]

Do you have the folder for ComfyUI-Manager inside the custom_nodes folder now?

[tender jay]

yeah

[onyx solar]

Ok. Great. Is ComfyUI currently running on your machine or do you have the browser and terminal window for it closed?

[tender jay]

before I had the Manager files in the Custom node folder

not in it's own folder in that folder.

[onyx solar]

Yeah, that'd definitely be an issue. 🙂

[tender jay]

that's what the youtube video told me to do.

[onyx solar]

Be sure to fully shut ComfyUI and start it again fresh.

When it starts up, if everything worked right, you'll get the Manager button in your UI.

[tender jay]

yeah this one is working better than the last one.

[onyx solar]

Excellent.

[tender jay]

now the update works.

now I just need to figure out how to get Flux working that's what I was trying to do.

I'll try that later though. I'm just happy I finally got this working.

Thanks for the help!

[onyx solar]

Patience and communication can solve many things. Thanks for being patient and communicating well.

[tender jay]

Thank you too. I'm not good with command line stuff, that's why I was trying to avoid git. But you made it seem easier that the youtube video's I watch.

[cerulean hemlock]

guys please help me i've been stuck on this for 2 hour

i tried googling and asking in other servers but got no answer

[vocal burrow]

what's the full log

[pale echo]

anyone knows how i fix that error?
RuntimeError: Input type (float) and bias type (struct c10::Half) should be the same

[vocal burrow]

same thing, what's the full log

[vocal burrow]

the full log

there's stuff before that

[pale echo]

before that are just my promts i was using

[vocal burrow]

and the whole initialization log

[fair oxide]

Can Automatic 1111 run locally on linux?

is there a big performance drop in comparison to windows?

[ornate elk]

If your on AMD it can even be faster on linxu

For nvidia it doesn't matter

[fair oxide]

this has been the only thing holding me back
time to switch ig

thx

[ornate elk]

No problem

[fair oxide]

Does it matter which distro performance wise?

[turbid carbon]

I was thinking of migrating to Linux after Microsoft trying their best to shove as many bloatware and spying stuff in Windows 11. Which Distro is the best purely for Stable Diffusion?
I'm thinking of Fedora or Mint.

[ornate elk]

Yea both are good, would use Mint or maybe EndavourOS

[quick roost]

I can't see the previews. How do I fix it?

[wind chasm]

Can anyone translate some Kohya error ? Im trying to work out whats broken

[lost crescent]

Is textual Inversion still available in A1111 at all?

There is - clip files go in clip folder - clip_vision files go in clip_vision folder

[fair oxide]

it worked, thank you very much for correcing me! Trns out I just suck at reading things properly

[ornate elk]

No problem! Yea most people go to fast over instructions ^^

[fair oxide]

when using directml it gives this error when running, and it doesnt use gpu at all and only uses cpu

[ornate elk]

What's the full cmd log?

And your GPU

[fair oxide]

it does it on my rx 6600 and 5600xt

[ornate elk]

With a 6600xt you should install the Zluda Version of automatic1111 for the best performance.

[fair oxide]

its not a xt version

just 6600

but ok

[ornate elk]

If it has 8gb its still good

[fair oxide]

5600xt work on that or not?

6gb card

[ornate elk]

Which one?

[fair oxide]

5600xt

but i have 2

so i can use both if that would allow me to use it with dual 5600xt

[ornate elk]

The 5600 could work too with Zluda. If it has the same gfx version as the rx5700

I checked it. Yes it should work

[fair oxide]

ok nice

ill give it a try later tonight

[ornate elk]

Okay, let me know if any step isnt clear. The latest guide is on the first link of the pinned messages

First start with the 6600 and get it working. after that you need to do some additional steps to get the 5600 working.
Just @ me then.

[fair oxide]

ok

[ornate elk]

Also that log says it found some old nvidia driver on your system.
Make sure you don't have any nvidia driver left. Also make sure to have the latest AMD Adrenalin driver

[fair oxide]

I used ddu to uninstall drivers

I can check again later

I also have a 6900xt coming, I'm going to use that for ollama to so that will be fun

[ornate elk]

Ouh nice. That also works good with stable-diffusion zluda

[fair oxide]

Yes and better support, for whenever rocm starts working

[ornate elk]

Rocm (hip SDK) works on windows right now. Zluda makes use of it and also ollama and other llm software

[fair oxide]

But the highwr end cards support the sdk

My 6600 doesn't

Just the runtime

[ornate elk]

There are community made files for the lower end cards. My guide covers that

[fair oxide]

Oh ok cool

Can you dm me whatever guide you want me to follow later?

[ornate elk]

For anything AMD related you can ask me

[fair oxide]

Awesome

[celest slate]

Hello everyone, I'm encountering a small problem in sd3 that I can't find diffusion_pytorch_model.safetensors file anywhere

Does anyone know where I can find it or how to solve it?, sd3 keeps giving me error about the file

It only asks for the file, but I can't find it anywhere

[heavy sigil]

heya all, quick one... using a PC/Edge browser and i can drag images into comfy image nodes, no probs. But if i try and do this on my Macbook/Safari browser it wants to try and open the image as a workflow and not load the image into the node. Any way around this?!

[ornate elk]

Hey, What's in your webui-user.bat?

Its unlikely that pony would run on your setup

[edgy kelp]

Yo i am using automatic1111

[ornate elk]

What's your GPU?

[edgy kelp]

but my lora's do not show up

i mean models

[ornate elk]

Loras are made for specific model versions. Like 1.5 or sdxl and only show up when a compatible model is selected

Ah alright. Then you can use pony

[edgy kelp]

and how may i do that?

[edgy kelp]

I'm on stability matrix

[ornate elk]

You should enable the FP8 mode for sdxl in the settings of auto1111

Sure np

[edgy kelp]

...

[ornate elk]

Download a compatible model from Civitai.com
Also read the description box of the loras. There you see with what model version they are compatible

[edgy kelp]

ye i know it's sdxl and sd 1.5

but how do i do that

[ornate elk]

Download a model and put it into the stable-diffusion-webui/models/stable-diffusion folder

Then select it in the dropdown

[edgy kelp]

i already downloaded the model

it just does not show up when i run automatic1111

[ornate elk]

What do you mean?

Best would be to not use stability matrix

[edgy kelp]

i do not see it here

how do i get sd 1.5?

[ornate elk]

You can download any model from Civitai.com manually

You don't need the base 1.5 model

[edgy kelp]

okay how do i put it in automatic1111

using stability matrix

[ornate elk]

Dont know sry, never used stability matrix

[edgy kelp]

damn

[manic matrix]

Da fuq this mean, everything was working last night

I deleted my old stuff to redownload it on the new SSD I bought and this happened

[ornate elk]

Dis you moved the webui to another drive?

[manic matrix]

No I deleted it first

Then I reinstalled it in another drive

[ornate elk]

The drive needs to be formated as GPT

[manic matrix]

Uh how do I do that

[ornate elk]

In the disk management

But formatting a drive will erase all data on it

So backup your stuff first

[manic matrix]

I have no data on it I bought it today

[ornate elk]

Ah alright

[manic matrix]

But I’m in the manager and I don’t see a place to format it

[ornate elk]

From Microsofts site:
Use Disk Management to convert an MBR disk

To complete the disk conversion by using Disk Management, follow these steps.

Before conversion, back up or move the data on the MBR disk.

In the search box on the taskbar, enter Create and format hard disk partitions to launch Disk Management.

Delete all partitions and volumes on the MBR disk.
    For each partition or volume, select and hold (or right-click) the item, and select Delete Partition or Delete Volume.

Select and hold (or right-click) the MBR disk to convert to the GPT format, and select Convert to GPT Disk.

The process notifies you when the conversion completes.

Do this only to the new drive and not to any other drive

[manic matrix]

I deleted the volume for it and now it’s not there for me to change

Do I need to unplug it and plug it back in

[ornate elk]

Right click and select new volume

[manic matrix]

Right clicking isn’t doing anything

[ornate elk]

Ok then yea plug it out and in

It should then ask

[manic matrix]

Now it only shows up in the manager as Sandisk locker

Even though I typed in the password and it’s unlocked it’s not showing up in the drive manager now

[ornate elk]

Can you screenshot it ?

How it looks in the disk management

[ornate elk]

Scroll down on the side

If its possible

[manic matrix]

Should I just eject and try yet again

I removed the password so I shouldn’t need the unlocker at all

[ornate elk]

So it was not locked before?

[manic matrix]

It was

Now it isn’t

[ornate elk]

Okay then replug

[manic matrix]

Nothing happens when I scroll

[ornate elk]

What's disk 2 ?

[manic matrix]

That’s this

But it’s not showing up in the volumes

[ornate elk]

Because its unallocated

You have to right click on it

[manic matrix]

Do I right click and say convert to MBR?

[ornate elk]

No

It needs to be GPT

[manic matrix]

That’s not an option I only get dynamic disk and mbr

[ornate elk]

Okay then it should be GPT now

You can create a new volume then

[manic matrix]

I looked it up and it says dynamic disk can run gpt?

But I didn’t convert it into gpt

I touched nothing

Just unplugging it turned it into gpt?

[ornate elk]

You deleted the volumes

Maybe it was MBR beforw

[manic matrix]

So what do I do with volumes now

[ornate elk]

You need to make one

Create new volume

[manic matrix]

How 🙂

New spanned or striped volume?

I have two options

[ornate elk]

Yea normaly it should just say new volume.
I have to read it up myself

[manic matrix]

I’m just gonna click spanned it seems pretty straightforward on google

Never mind

It’s telling me to do striped lol

[ornate elk]

Ah okay

If possible set it to GPT and then NTFS

[manic matrix]

It’s not letting me add it now 😭

It won’t let me click next

[ornate elk]

Thats strange

Are you familiar with Diskpart ?

[manic matrix]

No

Why is it telling me that I need to combine 2 drives

[ornate elk]

Something is wrong with that drive

[manic matrix]

???

[ornate elk]

You may need to use Diskpart to fix it

[manic matrix]

Awesome how do I use that

[ornate elk]

Its a CMD thing.
Open up a cmd and type
Diskpart

Then hit enter

[manic matrix]

Ok I’m there

[ornate elk]

In Diskpart type
List disk

[manic matrix]

It says disk 2 online with 931 gb free

[ornate elk]

Can you screenshot it?

[ornate elk]

How much drives do you have?

Physical

[manic matrix]

The OC, the Drive, and this one

What does that mean

Aren’t they all physical

[ornate elk]

It shows 3 disks

Is that correct ?

[manic matrix]

Yep

[ornate elk]

Okay then type
Select disk 2

[manic matrix]

I

[ornate elk]

Then type again
List disk

[manic matrix]

It showed up the exact same thing

[ornate elk]

Now it should show a *

In front of disk 2

[manic matrix]

Yes

[ornate elk]

Okay now type
Clean

[manic matrix]

Succeeded

Now what

[ornate elk]

Now replug the drive

The PC should show a message to format it

[manic matrix]

No but when I right click in the manager now it’s asking me to initialize

[ornate elk]

Okay sounds good

[manic matrix]

Set it to GPT

[ornate elk]

Sry for causing so much trouble 😅

[manic matrix]

Bro you’re good I’m trying to get it work

So now should I recreate the ai folder and go through all the steps again

[ornate elk]

Now checkout my install guide from the pinned messages.

There its all explained

[manic matrix]

Yep I memorized them that’s how much I’ve tried it 😭

[ornate elk]

You need to git clone the repo again

Onto the drive

[manic matrix]

I think I was gonna do that anyway

Wait

Now it’s not showing up in my drives list in the file explorer

I’m gonna break this damn desk dude

[ornate elk]

Did you made a new volume ?

[manic matrix]

It’s doing the same thing as before not letting me click next

It wants me to have 2 disks to combine

[ornate elk]

Okay we can try diskpart again

To fix this