NEW: There is a new method of installation that is much simpler and faster - You can find my new 60 second tutorial for it here: ua-cam.com/video/i5hvZvzcxoo/v-deo.html For those of you interested in learning how to train your own AI model with Stable Diffusion FAST, I have a playlist for it right here: ua-cam.com/video/2Nvpbvnj_u0/v-deo.html Also, we have a new discord specifically for anyone new needing general AI help usage/installation/etc here as well - discord.com/channels/1138979360943120445/1204654189897322557 --- If you'd like to try and see what the most common 'installation error' messages are and their solution, you might have some luck with the following: --- For those wondering about my computer's specs, I'm operating on a Windows 10 Pro 64-Bit, AMD Ryzen 7 3700X 8-Core Processor, 32 Gigs of Ram, And a GForce RTX 2070 Super Graphics Card - BIG Shoutouts to Penguin Dragon for taking so much time to walk me through the whole installation process - Here are all the materials demonstrated in the video: Github Website Link: github.com/ Hugging Face Website Link: huggingface.co/ Link to GitBash for windows: gitforwindows.org/ Stable Diffusion WebUI Link: github.com/AUTOMATIC1111/stable-diffusion-webui Hugging Face Stable Diffusion v1-4 Downlaod Link: huggingface.co/CompVis/stable-diffusion-v-1-4-original GFPGAN Face Restoration Link: github.com/TencentARC/GFPGAN Python 3.10.6 Download Link: www.python.org/downloads/release/python-3106/ Code for Gitbash Commands: "cd /c/ai" "git clone github.com/AUTOMATIC1111/stable-diffusion-webui.git" Code Edit for less powerful computers --opt-split-attention --precision full --no-half --medvram ---------------------------------- Common Error Messages and their solutions: 0-Problem: RuntimeError: Couldn't install Torch. 0-Solution: multiple versions of python confusing the system, you need to remove the older version (This is most likely because some versions of Windows come with Python installed by default) --- 1-Problem: when i launch the webui-user.bat file it says: "couldn't launch python" 1-Solution: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python" --- 2-Problem: If you get error code 9009, Python not found. 2-Solution: Uninstall Python using the executable, then go to the windows store and download the same Python version from there. --- 3-Problem: it say "Couldn't launch python" 3-Solution: press "add python to path" when i installed - --- 4-Problem: "No matching distribution found for torch==1.12.1+cu113" 4-Solution: You probably have a previously installed version of python already on my computer, so it was trying to use the old python to run the script. You need to uninstall any and all versions of python, and then delete the ai folder you created and start all over from the beginning. --- 5-Problem: Couldn't launch python, exit code: 9009 5-Solution: going into: start> type: manage app execution aliases> and turn off app installer --- 6-Problem: Creating venv in directory venv using python "C:\Python27\python.exe" Unable to create venv in directory venv exit code: 1 stderr: C:\Python27\python.exe: No module named venv 6-Solution: Try renaming your sd-v1-4.ckpt file to "model.ckpt" --- 7-Problem: it says python not found after installing it 7-Solution: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python". then safe it and it should work (make sure that its the actual python, not some shortcut) Hope that helps :)
@@Dracolichizm People have been successful in making it work on amd , but it's really shoddy... But the beauty of open source is that probably someone will make it work given enough time
This is likely one of the most comprehensive tutorials that I've ever watched for installing any complex software packages in a Linux-UI style manner (no .exe to download by clicking a link, then install by double clicking the file). I've always been confused as to why these layers of complexity are implemented when the dev could just make a downloadable installation file. Like, this is a full on hour, to hour and a half process with a multitude of steps. Think of the combined/condensed days/weeks/years it takes for people to install this, vs the time it'd take for the dev to create an install package.
@@sizlax That's the price of modularity and development environments. None of these things are forced to work with one another and it's super easy to plug in different AI models and do research while having a nice UI on any platform. Windows, Mac, Linux on any hardware. Installers require the developer to make individual installers for every platform and various hardware combinations as well as testing everything. Despite the code not doing anything different. Just because the installers might introduce bugs. In other words, the focus of this software isn't end users. It's developers and AI researchers. And instead of spending tons of time to making install packages they instead spend this time advancing the software.
@@Elanduli Well, I do wish they'd hurry up and perfect that deepfake software. I think it'd be pretty fun to play around with. But I just can't be fucked to go through all these steps just to install it. I mean, I'm consistently dealing with enough headaches trying to learn Autodesk Maya, then getting around all the inevitable bugs and glitches; especially in Xgen (the plug-in within Maya that's used for creating the most realistic game/3D animated hair). And any time I'm not doing that, I'm trying to improve my skills in digital artistry. In all honesty, I've got lots of time, but limited patience for unnecessary complexities.
A note from someone who uses the Windows CMD console a LOT at work: It is safer to press space than enter, if you think the terminal is frozen. Enter may "accept" a command, whereas space will simply just type a space if it has frozen at a command input prompt. Can't tell you how many times I accidentally accepted a default value by pressing enter before I realized any input works to "unfreeze" the screen...
im not gonna install it yet... but i gotta say: I LOVE THIS TUTORIAL!!! I always wish every tutorial is like this. Straight to the point and the instructor knows how to edit. Damn. perfect.
gonna be honest, I don't often give videos likes. This has absolutely earned it and I've installed SD twice on two computers from this tutorial. This is the clearest tutorial I think I've seen on youtube, no nonsense, no bs, just step by step. Thank you a million, man!
Great Tutorial! Unlike other tutorials, this one is actually straight to the point with no useless nonsense information. Great job, this is worth a subscribe!
because a lot of people seem to be having trouble with torch install: The latest supported python version for torch is 3.9.13. If you installed 3.10 like the tutorial suggests, you'll need to uninstall it and use 3.9.13 instead.
Guys, if you seem to get far but then an error message comes up about something not being able to access the GPU, this is something you should try: right-click the 'webui-user' file and select 'edit' this will take you to the 'batch file'. Add the command "--skip-torch-cuda-test" after 'set COMMANDLINE_ARGS='. This worked for me, after being stuck for a pretty long time. (I think maybe 1 hour in total). Good luck guys! I'm gonna finally test it all out!
@@DMLorem No problem bro I remember doing this and being sooo frustrated, glad it's over. I hope you're having a good time with it like I was half a year ago lol
This is the way every tutorial should be! I had no trouble with this and it was all thanks to you and the patron! Had I taken any other tutorial this would've taken me days! Thank you so much! May your bounty be plentiful and your hunger satiated!
This tutorial is so good, I am using it as a reference how to properly, precisely and quickly present an information. Every tutorial should be like that. Instead of playing it with x1.5 speed, pausing to note down necessary content. PS: In my cause, I had to pause it for a week at 2:41 to buy an RTX to handle Stable Diffusion :))
followed the tutorial but when i launch the webui-user.bat file it says: "couldn't launch python" i reinstalled python and made sure it's the exact same path as it should, but the batch file won't work EDIT: found the solution in the comment section: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python"
so, when i do that it says Couldn't launch python exit code: 9009 stderr: 'C:\Users\User\AppData\Local\Programs\Python\Python310' is not recognized as an internal or external command, operable program or batch file. Launch unsuccessful. Exiting. Press any key to continue . . . please help its very frusterating
The longest 3 mins of my entire life. Although I must say...it worked in the end. If you are having problems with python (like me) install from the microsoft store instead...thats what worked for "failed launch of python" when trying to run the batch file. Hope this helps someone, because I was stressing over it lol
Stable Diffusion is one of the best image generators I have ever seen, and it is free and open source. It is astonishing that something like this is possible.
Hey, I just wanted to say thank you so much for making this tutorial. Without dropping names, I tried a few other tutorials and they didn't work/give me what I want, and I just happened to stumble upon this one. I have been trying to set this up for a week and I finally got it working with this video. I will definitely be looking through the rest of your Stable Diffusion playlist as I want to learn more about this program and how to use it. Awesome video! Thank you so much again ♥️
hello, what should I fix, if i have this: ERROR: Could not find a version that satisfies the requirement torch==2.1.2 (from versions: 2.2.0, 2.2.0+cu121, 2.2.1, 2.2.1+cu121, 2.2.2, 2.2.2+cu121) ERROR: No matching distribution found for torch==2.1.2
bro love your guide. I spent the whole day with some guys who post their work and skip alot of things. Going around in circle, had to delete alot of things and gave them a thumbs down.
For the guys that have AMD and when doing the last step it gives you error ''torch can't use gpu'' you just have to copy de prompt that it gives you (in my case ''--skip-torch-cuda-test''), then go to the ai folder, open webui-user.bat with notepad++ (or any other, just open it with notepad) and copy the prompt in command line args (just, copy paste next to the = sign), save the file and you are good to go. I know it sounds obvious but it took me a while to understand xd.
Dude, thank you for making a no-nonsense, quick, easy to understand tutorial. So tired of videos that have 3 minutes of promoting the channel and blah blah, detours, etc. Thanks man!
very straightforward tutorial, found no issues and the only problem is the waiting part. which to me took some 2 hours... but thats due to my sucky internet provider. Thanks for the simple tutorial!
I came across two issues that I could eventually fix. I'll share what I did maybe it'll help you too. First, after the installation, I had issues with the local URL not working and I fixed that (following another comment in another youtube video) by uninstalling Python and re-installing it with CUSTOM INSTALLATION and everything ticked (I had to delete the files and restart the process for it to work as well) and then the URL file worked. Secondly, SD kept generating black images no matter what I did, the only solution that fixed it for me (following a comment on Reddit) was typing --lowvram --precision full --no-half after set COMMANDLINE_ARGS= with no space, ctrl+save, and reopening the webui, and IT GOT FIXED. Hope this can be helpful!
i cant believe i actually managed to follow a tutorial with this many steps, i ran into some problems with it, but at the end it actually worked and im so happy it did
Great tutorial, after I've installed it successfully, I've closed everything, how do I restart everything again? what prompt do I use, instead of re-installing everything again?
i followed all of the instructions presented by the video but whenever i launch webui it gets stuck on installing torch and torchvision, can somebody please help me
Heads up the share option actually exposes the application to the entire internet and allows for remote code execution. I would highly recommend no one use it under any circumstance.
It's working! Thank you so much every other tutorial was either convuluded or hard to follow i.e giving wrong or over simplified instructions. It's downloading now! I just had to screw around with the .cfg file to get the batch file to find the right version of python after I corrected it. So if anyone is watching this, make sure you follow his instructions EXACTLY (though I did download the newer version of stable diffusion as he said to; at the time the newest was 1.4 for him but it was 2.1 for me). The version of Python he got working for him was the EXACT one you need, don't look for the newer one. Thank you so much Royal Skies! You're a life saver. Wish me luck with my writing!
I ran into the same thing and grabbed a newer version of Python, and got errors and had to start over the install. Follow exact same versions of the files listed, and you should be golden
okay ive got 1 important question, can i simply make a shortcut link to the generated link from that git command prompt? or if i close it do i have to generate a new link every time? and if so how would i go about doing that?
Thanks a lot for the tutorial!!! Is it possible to add more models? Does it require some additional steps, or just putting models into a folder just like shown on this video?
I ran into multiple errors. It turned out I had Python 3.7 already preinstalled, and continued to have errors when I uninstalled it and installed 3.10. Finally I deleted everything, and redownloaded the webgui thing, ran it all again, and it worked! Amazing work here, thank you.
Just want to give you a big thank for making this video and channel. Dummy like myself was ready to pay some freelance to install Stable Diffusion on my computer and pay them to teach me how to use it. Most videos on YT were made for people who already have a certain level of knowledge. I have never seen a video so effective and easy to follow. I am supporting you, keep making good new content, so dummies like me can take the AI ride too
Honestly, I'm so impressed at how well SD operates. Just reading the comments it looks like the tradeoff between having a less beefy computer is just the time, which honestly seems negligible when the thought of this hitting mainstream (based on the pacing of development on hardware & software side) is within five to seven years. I'd wager that the only bottleneck once it hits mainstream would be the VRAM but it's too early (for me) to tell. I'm running on a Lenovo T7i (i7-11700K, 16GB RAM, RTX 3070) and it takes me less than 15s to generate one 512x512 image using LMS on 20 step samples
@@anthonycampbell422 Nice! I was initially thinking of upgrading my RAM to see if I can justify SD but as long as you have a GPU with a lot of VRAM you should be good to go. I've been using Euler_a on 80-100 samples, CFG at 15-20, and have been creating them one by one lately but the output quality has not really faltered.
How the heck do you reenter the software though? The link no longer works and I don't want to reinstall anything since I have it already. I had to exit as I can't leave it running forever but Idk how to get back in
I followed the instructions to the letter on Windows 10, but when I run the batch file I get this error: Couldn't launch python exit code: 9009 stderr: Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases. Launch unsuccessful. Exiting. Press any key to continue . . . Can you please tell me how I can fix this?
My friend linked me this guide and boy I was never disappointed All guides I've seen are hours and etc This one is straightforward easy and works like a charm Thank u
Once installed, how do you update the automatic1111 installation via git? Seems kind of important for an everchanging project and as much as I appreciate the instructions to the successful one time installation - it leaves me clueless on how to keep the automatic1111 web-ui updated (preferably without loosing my settings and/or custom ckpt-models. Worth to make a youtube video about it? Highly appreciated in advance...
Worth mentioning. Most of this machine learning stuff does not play well with AMD GPUs, if you can get it to work at all. You'll need an Nvidia card or a _very_ powerful CPU for decent performance.
Is the URL always available to launch, or is there something you need to execute manually to generate the URL? My plan was to just get it installed then create a shortcut to open the GUI in my browser.
am getting the runtime error: torch is not able to use gpu any advice thanks very much Traceback (most recent call last): File "C:\ai\stable-diffusion-webui\launch.py", line 48, in main() File "C:\ai\stable-diffusion-webui\launch.py", line 39, in main prepare_environment() File "C:\ai\stable-diffusion-webui\modules\launch_utils.py", line 384, in prepare_environment raise RuntimeError( RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check Press any key to continue . . .
you have to rightclick "webui-user.bat and add --skip-torch-cuda-test to COMMANDLINE_ARGS that it should look like this @echo off set PYTHON= set GIT= set VENV_DIR= set COMMANDLINE_ARGS= --skip-torch-cuda-test if you keep getting some error about half or something, add --precision full --no-half that it looks like call webui.bat --precision full --no-half
How do you, re-open it? That is to say, once I've restarted my computer the local URL no longer works and I just ended up having to do the last part on CMD again. Is this normal or is there a way around this?
Trying to figure it out as well. Like this tutorial ends with how to start it and it worked. But now I have no idea how to turn it on again. I tried running webui-user.bat again but it stops at saying "To create a public link, set `share=True` in `launch()`." I have no idea what I'm supposed to do. I went into a file called launch and opened with notepad but there's no "share=false" or something like that.
This is great! It's very easy to dismiss SD at a glance beacuse of the FREE pricetag. You might find it cheap in quality and lacking features. This however is the reverse when you start to dive in. You'll soon start to see how jam packed of features SD has and how fast it's overtaking the paid alternatives with it's open source community behind it.
Costing money is not at all connected to quality, especially in the FOSS community. A lot of people do what you said, they dismiss open source stuff because it looks finicky and different, but its almost always an improvement to whatever proprietary product they're using.
Heyo I did the exact steps stated in this video but had to add --precision and --no-half to get it all running. Yet it is very slow and doesnt use my GPU (only 7% Load) can sb help me fix this? I hvae a AMD RX6750XT... Thank you in advance!
I don't think it works on macbooks, but unlike the other tutorials I watched, this one was straight to the point and simple, so thanks for making it this way👍
This is hands down the best How To video ever. The world would be a better place, almost like a heaven, if all tutorials would look like this. Keep up the good work mate!
will you be going into detail of the stable diffusion web UI interface? or how to use scripts and libraries and that stuff? The first and only prompt I did turned out really weird and I feel kinda dumb lol it's not as easy as a browser based image generator lol
so when i click on the webui-user it load up and gets the point where it says "installing torch and torchvision" but it doesn't look like its doing anything. how do i know if its working... its been quite a long time and its just stuck on that screen. how long does this step normally take? i have a very beefy pc setup
at 2.40 on the cmd, it block at ""Textual inversion embeddings loaded(0): Model loaded in 76.0s (calculate hash: 15.2s, load weights from disk: 50.0s, create model: 4.3s, apply weights to model: 1.0s, apply half(): 1.0s, load VAE: 0.1s, move model to device: 1.0s, load textual inversion embeddings: 3.5s).""""
Some remarks GFPGAN is optional, Automatic's repo has CodeFormer that will download and install by itself, that imo works better. --precision full no-half is only required for cards series GTX 16XX, that have rpoblems with half point precision, resulting in black or green image. Remove it normally as it will just increase memory usage. --medvram can be changed to --lowvram if you still get CUDA out of memory errors lastly, adding 'git pull' before 'call webui bat' will automatically download updates and it tends to be added something every day. Your version for example don't have the tab for training custom style or objects. AI can be extended by downloading additional objects or styles 4kb files in 'embeddings' folder.
Hey Robert seems like you know what you are talking about so I kindly asking here. Im looking for a decent tutorial on how to use the best of this WebUI I just installed in this video. Im looking around but most tutorial are for the web version and not local or for another UI. What would you recommend me? Thanks in advance for your help !
Thanks for taking the time to clarify for everyone! I didn’t know about the -lowvram argument. I have Nvidia 16XX (i think 1680) and it gave me black screens but still took forever to create, then I put in the magical first 3 arguments found on a reddit, and my ai now dreams of penguin dragons and cielmaids 🐧🐉
@@Yotanou In my experimenting, what makes or breaks the ai’s content is your use of artist names. I’m personally looking forward to the “important artist names” bonus mission from RoyalSkies this month. If you haven’t checked out RoyalSkies previous video on stable diffusion’s official site, start there!
i have the black square image i see what you said but how do i actually perform that instruction? do i need to edit a text file do i need to do it in cmd or is it one of the settings on the web version?
@@motionsick Great question, figured it out? I have the same problem and there are no answers to it anywhere. People just try it once and that's it or what?
I have trouble to the step after clicking on the webui.user. there is an error and it is written : [notice] A new release of pip available: 22.2.1 -> 23.1.2 [notice] To update, run: C:\IA\stable-diffusion-webui\venv\Scripts\python.exe -m pip install --upgrade pip of course, I have no others versions of python and when I go to python command prompt and paste "-m pip install --upgrade pip", it says: invalid syntax..
Dude... It works... *press play* * listen* *pause* * do what you're told* *press play* *repeat*! Links in the description in order! Brilliant job! Thank you very much!
Great Tutorial Thx ! Installed without any problem. But i have maybe a dumb question, how can i relaunch it ? Yesterday its worked fine but today off course the link isn't good anymore. Do i need to do again every step of the tutorial ? Thanks
Looks intimidating at first but it was stupidly easy to do. I am absolutely horrible at computer related stuff but this was super easy to follow. Keep up the great videos 👍
Thank you for your work. I'm stuck at 2min13, when I launch "webui-usr", the window starts and says "unable to create venv in directory venv exit code 3 stderr the specified path is not found." I have however respected all the folder positioning in the tutorial. Thanks for your help
Didn't work for me unfortunately. It gives me the URL but it leads to nothing. I left the webui batch file open, but still nothing. Anyone got any advice as to why it won't work? I got no errors at all
ERROR: Could not find a version that satisfies the requirement torch==2.1.2 (from versions: 2.2.0, 2.2.0+cu121, 2.2.1, 2.2.1+cu121, 2.2.2, 2.2.2+cu121, 2.3.0, 2.3.0+cu121, 2.3.1, 2.3.1+cu121) ERROR: No matching distribution found for torch==2.1.2
This only works with Python 3.10.6. If you try to use a later version you will get the "torch error". - Uninstall Python - Install the proper version of Python - Delete repo - Clone repo - Move chck files and .pth files to respective places - Profit.
ERROR: Could not find a version that satisfies the requirement torch==1.13.1+cu117 (from versions: 2.0.0, 2.0.0+cu117) ERROR: No matching distribution found for torch==1.13.1+cu117
Hey, I am stuck at "Installing torch and torchvision". I already tried both variants, one for the "normal" computers and one for the ones with less processing power but both stay stuck at that command. What could I do? Every help would be greatly appreciated
Jeez. I got it to work. My problems were nothing to do with your tutorial - kudos for the work you put into this. My issue was my Python version was 3.9 so I aborted this process after reading the initial warning and installed 3.10 - my PC refused to let go of the previous path to the Python.exe and the .BATch file wouldn't work even if I specified the new path explicitly. So I rolled back to 3.9 and it worked fine. Of course prompt engineering is another mountain to climb but at least I can play around free of nags for now. Massive thank you!
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument index in method wrapper__index_select)
This model is the reason there's a discord server called unstable diffusion dedicated to NSFW generations ( and like always, nsfw drives the industry, it has quite a big R&D community in it )
@@Mandraw2012 oh good. I'm happy about that. While I'm not planning to generate NSFW stuff the fact things I make can get misconstrued as it is a major turn off
Getting an error that says "NansException: A tensor with all NaNs was produced in Unet. This could either be because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in settings>Stable Diffusion or using the no-half commandline argument to fix this issue..... Any help? I have a 1660ti graphics card, and I downloaded stable diffusion v1.5
It won't let me install torch, it's saying "No module named pip'". I've made sure there's no previous versions of Python on my computer and followed everything as instructed twice. Still doesn't work when I double click the batch file, any help?
I searched a lot but pytorch is just not compatible with python 3.11 so the only solution is to uninstall python. and install an old version. seems like 3.9.13 works for me.
for problem 1 "1-Problem: when i launch the webui-user.bat file it says: "couldn't launch python" 1-Solution: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python" i cant find the "set PYTHON=" within the webui-user file. im downloading this to my local drive (D:), so I'd like to change the directory if possible.
This would be great if everything python related was as easy as it was made out in the video. A fresh install of the python version found in the video will just lead to a bunch of new errors for windows users. Anyone getting a "no module named encodings" error, search up System Variables, click Environment Variables, and in User variables for [user], click New and type PYTHONHOME with the Variable Value being your Python's directory (for me it's C:Users\[user]\Local\Programs\Python\Python310). Now that I've done this, I get a new error: (AttributeError: 'str' object has no attribute 'isascii') At this point I've given up. I officially hate python.
I was getting "no module 'xformers'. Processing without.." three times. The program still launches, but I don't know if CFPGAN works. Also it said it couldn't load styles
I have a different error. First of all it shows me Downloading torch==2.0.1 Downloads something weighing 2.6 GB then when it finishes it's just on 2.6 / 2.6 and nothing happens.
When i´m at the process at 2:31 it says ERROR: could not find a version that satisfies the requirement torch 2.0.1. ERROR No matching distribution found for torch 2.0.1. What could the error be? Otherwise i completely followed your instructions.
NEW: There is a new method of installation that is much simpler and faster - You can find my new 60 second tutorial for it here:
ua-cam.com/video/i5hvZvzcxoo/v-deo.html
For those of you interested in learning how to train your own AI model with Stable Diffusion FAST, I have a playlist for it right here:
ua-cam.com/video/2Nvpbvnj_u0/v-deo.html
Also, we have a new discord specifically for anyone new needing general AI help usage/installation/etc here as well -
discord.com/channels/1138979360943120445/1204654189897322557
---
If you'd like to try and see what the most common 'installation error' messages are and their solution, you might have some luck with the following:
---
For those wondering about my computer's specs, I'm operating on a Windows 10 Pro 64-Bit, AMD Ryzen 7 3700X 8-Core Processor, 32 Gigs of Ram, And a GForce RTX 2070 Super Graphics Card -
BIG Shoutouts to Penguin Dragon for taking so much time to walk me through the whole installation process -
Here are all the materials demonstrated in the video:
Github Website Link:
github.com/
Hugging Face Website Link:
huggingface.co/
Link to GitBash for windows:
gitforwindows.org/
Stable Diffusion WebUI Link:
github.com/AUTOMATIC1111/stable-diffusion-webui
Hugging Face Stable Diffusion v1-4 Downlaod Link:
huggingface.co/CompVis/stable-diffusion-v-1-4-original
GFPGAN Face Restoration Link:
github.com/TencentARC/GFPGAN
Python 3.10.6 Download Link:
www.python.org/downloads/release/python-3106/
Code for Gitbash Commands:
"cd /c/ai"
"git clone github.com/AUTOMATIC1111/stable-diffusion-webui.git"
Code Edit for less powerful computers
--opt-split-attention --precision full --no-half --medvram
----------------------------------
Common Error Messages and their solutions:
0-Problem: RuntimeError: Couldn't install Torch.
0-Solution: multiple versions of python confusing the system, you need to remove the older version (This is most likely because some versions of Windows come with Python installed by default)
---
1-Problem: when i launch the webui-user.bat file it says: "couldn't launch python"
1-Solution: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python"
---
2-Problem: If you get error code 9009, Python not found.
2-Solution: Uninstall Python using the executable, then go to the windows store and download the same Python version from there.
---
3-Problem: it say "Couldn't launch python"
3-Solution: press "add python to path" when i installed -
---
4-Problem: "No matching distribution found for torch==1.12.1+cu113"
4-Solution: You probably have a previously installed version of python already on my computer, so it was trying to use the old python to run the script. You need to uninstall any and all versions of python, and then delete the ai folder you created and start all over from the beginning.
---
5-Problem: Couldn't launch python, exit code: 9009
5-Solution: going into: start> type: manage app execution aliases> and turn off app installer
---
6-Problem: Creating venv in directory venv using python "C:\Python27\python.exe"
Unable to create venv in directory venv
exit code: 1
stderr:
C:\Python27\python.exe: No module named venv
6-Solution: Try renaming your sd-v1-4.ckpt file to "model.ckpt"
---
7-Problem: it says python not found after installing it
7-Solution: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python". then safe it and it should work (make sure that its the actual python, not some shortcut)
Hope that helps :)
I some how got an error :(
@@DrFeho Then say what it is ¯\_(ಠ_ಠ)_/¯
From what I recall this will only work with Nvidia cards currently
@@Dracolichizm People have been successful in making it work on amd , but it's really shoddy... But the beauty of open source is that probably someone will make it work given enough time
@@Dracolichizm i have that. i got an error when you have to let it download some thing
This should be the worldwide baseline standard for how tutorial videos are made. Thank you for this comprehensive, easy-to-follow setup guide.
This is likely one of the most comprehensive tutorials that I've ever watched for installing any complex software packages in a Linux-UI style manner (no .exe to download by clicking a link, then install by double clicking the file). I've always been confused as to why these layers of complexity are implemented when the dev could just make a downloadable installation file. Like, this is a full on hour, to hour and a half process with a multitude of steps. Think of the combined/condensed days/weeks/years it takes for people to install this, vs the time it'd take for the dev to create an install package.
@@sizlax That's the price of modularity and development environments. None of these things are forced to work with one another and it's super easy to plug in different AI models and do research while having a nice UI on any platform. Windows, Mac, Linux on any hardware. Installers require the developer to make individual installers for every platform and various hardware combinations as well as testing everything. Despite the code not doing anything different. Just because the installers might introduce bugs.
In other words, the focus of this software isn't end users. It's developers and AI researchers.
And instead of spending tons of time to making install packages they instead spend this time advancing the software.
@@Elanduli Well, I do wish they'd hurry up and perfect that deepfake software. I think it'd be pretty fun to play around with. But I just can't be fucked to go through all these steps just to install it.
I mean, I'm consistently dealing with enough headaches trying to learn Autodesk Maya, then getting around all the inevitable bugs and glitches; especially in Xgen (the plug-in within Maya that's used for creating the most realistic game/3D animated hair).
And any time I'm not doing that, I'm trying to improve my skills in digital artistry.
In all honesty, I've got lots of time, but limited patience for unnecessary complexities.
you don't like indian accents and shitty mics
@@sizlax brother the steps take
A note from someone who uses the Windows CMD console a LOT at work: It is safer to press space than enter, if you think the terminal is frozen.
Enter may "accept" a command, whereas space will simply just type a space if it has frozen at a command input prompt.
Can't tell you how many times I accidentally accepted a default value by pressing enter before I realized any input works to "unfreeze" the screen...
you should migrate to powershell and start using test-connection instead of ping. Join me :)
this actually unpaused the download for me when hitting enter didn't, thanks
The download isnt paused
Yep, this helped unpause the download while clicking didn't do anything for me. Thanks!
thank you
im not gonna install it yet... but i gotta say: I LOVE THIS TUTORIAL!!! I always wish every tutorial is like this. Straight to the point and the instructor knows how to edit. Damn. perfect.
Well said :)
agger
ua-cam.com/video/qe9PEJo3_VE/v-deo.html
Fr
As a tech writer, I 100% agree. Straight and to the point! Excellent tutorial.
gonna be honest, I don't often give videos likes. This has absolutely earned it and I've installed SD twice on two computers from this tutorial. This is the clearest tutorial I think I've seen on youtube, no nonsense, no bs, just step by step. Thank you a million, man!
Great Tutorial! Unlike other tutorials, this one is actually straight to the point with no useless nonsense information. Great job, this is worth a subscribe!
One of the best tutorials of any kind. Straight to the point, no useless talking. Thank You!
because a lot of people seem to be having trouble with torch install: The latest supported python version for torch is 3.9.13. If you installed 3.10 like the tutorial suggests, you'll need to uninstall it and use 3.9.13 instead.
bro wtf people say this for every fucking python version. why do most people seem to get it right with 3.10.6?
Mine had the torch install error too but underneath it it said i needed to download windows distributable C++ and gave me a link. Then it worked fine.
yup, work for me, thx bro.
ill try it. thanks
@@ARegularPerson-bs2hc thank you
Great tutorial: Straight to the point, no filler, everything you need is shown and explained and links are in the description. Wow! Thank you so much!
Deserves way more subs tbh
how do u copy the ai file address if using win11?
Finally a tutorial that is concise, doesn't skip important details, and doesn't bombard me with information I didn't click for. Thanks so much!
Yesss!!! #RespectOurVitalTime 😊👍
Guys, if you seem to get far but then an error message comes up about something not being able to access the GPU, this is something you should try:
right-click the 'webui-user' file and select 'edit' this will take you to the 'batch file'. Add the command "--skip-torch-cuda-test" after 'set COMMANDLINE_ARGS='. This worked for me, after being stuck for a pretty long time. (I think maybe 1 hour in total).
Good luck guys! I'm gonna finally test it all out!
Thanks bro worked for me
@@DMLorem No problem bro
I remember doing this and being sooo frustrated, glad it's over. I hope you're having a good time with it like I was half a year ago lol
This is the way every tutorial should be! I had no trouble with this and it was all thanks to you and the patron! Had I taken any other tutorial this would've taken me days! Thank you so much! May your bounty be plentiful and your hunger satiated!
it looked intimidating at first but it's super easy to follow. thanks!
ua-cam.com/video/qe9PEJo3_VE/v-deo.html
That was the most straight to the point pain free tutorial I've followed in my entire life. Thank you!
It truly is, it respects our brains and not treat us like 5 year old as all other tutorials.
This tutorial is so good, I am using it as a reference how to properly, precisely and quickly present an information.
Every tutorial should be like that. Instead of playing it with x1.5 speed, pausing to note down necessary content.
PS: In my cause, I had to pause it for a week at 2:41 to buy an RTX to handle Stable Diffusion :))
do we need to copy the url everytime and run in browser when i open my desktop the next day?
followed the tutorial but when i launch the webui-user.bat file it says: "couldn't launch python"
i reinstalled python and made sure it's the exact same path as it should, but the batch file won't work
EDIT:
found the solution in the comment section:
you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python"
so, when i do that it says
Couldn't launch python
exit code: 9009
stderr:
'C:\Users\User\AppData\Local\Programs\Python\Python310' is not recognized as an internal or external command,
operable program or batch file.
Launch unsuccessful. Exiting.
Press any key to continue . . .
please help its very frusterating
Problem was not selecting “add to path” when installing python. Now webui user runs but says cannot use gpu
@@ogonbio8145 yeah same im stuck there
@@ogonbio8145 what is your pc spec
you need to register python with enviromental variable
The longest 3 mins of my entire life. Although I must say...it worked in the end. If you are having problems with python (like me) install from the microsoft store instead...thats what worked for "failed launch of python" when trying to run the batch file. Hope this helps someone, because I was stressing over it lol
Good looking out! got it working :D
Stable Diffusion is one of the best image generators I have ever seen, and it is free and open source. It is astonishing that something like this is possible.
Hey, I just wanted to say thank you so much for making this tutorial. Without dropping names, I tried a few other tutorials and they didn't work/give me what I want, and I just happened to stumble upon this one. I have been trying to set this up for a week and I finally got it working with this video. I will definitely be looking through the rest of your Stable Diffusion playlist as I want to learn more about this program and how to use it. Awesome video! Thank you so much again ♥️
hello, what should I fix, if i have this: ERROR: Could not find a version that satisfies the requirement torch==2.1.2 (from versions: 2.2.0, 2.2.0+cu121, 2.2.1, 2.2.1+cu121, 2.2.2, 2.2.2+cu121)
ERROR: No matching distribution found for torch==2.1.2
I have followed many other installation tutorials trying to set it up..
Yours work perfectly! Thanks a lot!
Short, straight to the point, actually has useful extra information...You really nailed the tutorial format, worth a subscribe!
bro love your guide. I spent the whole day with some guys who post their work and skip alot of things. Going around in circle, had to delete alot of things and gave them a thumbs down.
For the guys that have AMD and when doing the last step it gives you error ''torch can't use gpu'' you just have to copy de prompt that it gives you (in my case ''--skip-torch-cuda-test''), then go to the ai folder, open webui-user.bat with notepad++ (or any other, just open it with notepad) and copy the prompt in command line args (just, copy paste next to the = sign), save the file and you are good to go. I know it sounds obvious but it took me a while to understand xd.
Thank you, helped me very much!
thank you somuch :)!
Dude, thank you for making a no-nonsense, quick, easy to understand tutorial. So tired of videos that have 3 minutes of promoting the channel and blah blah, detours, etc. Thanks man!
very straightforward tutorial, found no issues and the only problem is the waiting part. which to me took some 2 hours... but thats due to my sucky internet provider.
Thanks for the simple tutorial!
Thanks for the quick and easy walkthrough, truly appreciate your effort in making this.
I came across two issues that I could eventually fix. I'll share what I did maybe it'll help you too. First, after the installation, I had issues with the local URL not working and I fixed that (following another comment in another youtube video) by uninstalling Python and re-installing it with CUSTOM INSTALLATION and everything ticked (I had to delete the files and restart the process for it to work as well) and then the URL file worked. Secondly, SD kept generating black images no matter what I did, the only solution that fixed it for me (following a comment on Reddit) was typing --lowvram --precision full --no-half after set COMMANDLINE_ARGS= with no space, ctrl+save, and reopening the webui, and IT GOT FIXED. Hope this can be helpful!
Thanks for the comment here. Im setting mine up right now, so it's helpful to know this while setting this whole thing up now rather than later.
Where do you type these lines sorry? Same issue
@@theresident1 you need to edit webui-user.bat
Followed instructions using python 3.11.4 and no issues so far using the tool or generating images. Thank you for the guide!
i cant believe i actually managed to follow a tutorial with this many steps, i ran into some problems with it, but at the end it actually worked and im so happy it did
Great tutorial, after I've installed it successfully, I've closed everything, how do I restart everything again? what prompt do I use, instead of re-installing everything again?
i followed all of the instructions presented by the video but whenever i launch webui it gets stuck on installing torch and torchvision, can somebody please help me
I have exact same issue
same issue
Heads up the share option actually exposes the application to the entire internet and allows for remote code execution. I would highly recommend no one use it under any circumstance.
It's working! Thank you so much every other tutorial was either convuluded or hard to follow i.e giving wrong or over simplified instructions. It's downloading now! I just had to screw around with the .cfg file to get the batch file to find the right version of python after I corrected it. So if anyone is watching this, make sure you follow his instructions EXACTLY (though I did download the newer version of stable diffusion as he said to; at the time the newest was 1.4 for him but it was 2.1 for me). The version of Python he got working for him was the EXACT one you need, don't look for the newer one.
Thank you so much Royal Skies! You're a life saver. Wish me luck with my writing!
I ran into the same thing and grabbed a newer version of Python, and got errors and had to start over the install. Follow exact same versions of the files listed, and you should be golden
I don't know why but somehow this video could the most simple-and-clear tutorial instruction I've ever seen, great job and thanks a lot dude !
okay ive got 1 important question, can i simply make a shortcut link to the generated link from that git command prompt? or if i close it do i have to generate a new link every time? and if so how would i go about doing that?
Thanks a lot for the tutorial!!!
Is it possible to add more models? Does it require some additional steps, or just putting models into a folder just like shown on this video?
Thank you for the great tutorial! Just in case someone is trying: Python 3.11 is not working. You get the torch errors.
Thank you so much! I've been wondering if it would work with 3.11
@@eliherrin3997 Same for me. Thanks M. Tutorial!
You have no idea how much I appreciate a tutorial that gets to the point. Thank you!
I ran into multiple errors. It turned out I had Python 3.7 already preinstalled, and continued to have errors when I uninstalled it and installed 3.10. Finally I deleted everything, and redownloaded the webgui thing, ran it all again, and it worked! Amazing work here, thank you.
Just want to give you a big thank for making this video and channel. Dummy like myself was ready to pay some freelance to install Stable Diffusion on my computer and pay them to teach me how to use it. Most videos on YT were made for people who already have a certain level of knowledge. I have never seen a video so effective and easy to follow. I am supporting you, keep making good new content, so dummies like me can take the AI ride too
Honestly, I'm so impressed at how well SD operates. Just reading the comments it looks like the tradeoff between having a less beefy computer is just the time, which honestly seems negligible when the thought of this hitting mainstream (based on the pacing of development on hardware & software side) is within five to seven years. I'd wager that the only bottleneck once it hits mainstream would be the VRAM but it's too early (for me) to tell.
I'm running on a Lenovo T7i (i7-11700K, 16GB RAM, RTX 3070) and it takes me less than 15s to generate one 512x512 image using LMS on 20 step samples
I upgraded to a 3090 and 32gb of ram just in case for stable diffusion :D
@@anthonycampbell422 Nice! I was initially thinking of upgrading my RAM to see if I can justify SD but as long as you have a GPU with a lot of VRAM you should be good to go.
I've been using Euler_a on 80-100 samples, CFG at 15-20, and have been creating them one by one lately but the output quality has not really faltered.
How the heck do you reenter the software though? The link no longer works and I don't want to reinstall anything since I have it already. I had to exit as I can't leave it running forever but Idk how to get back in
@@Subutai_Khan Run the file from 2:16. It should start quickly - the downloading is a one-time thing.
I followed the instructions to the letter on Windows 10, but when I run the batch file I get this error:
Couldn't launch python
exit code: 9009
stderr:
Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases.
Launch unsuccessful. Exiting.
Press any key to continue . . .
Can you please tell me how I can fix this?
got the same problem :/
Struggling with this too :/
i fixed this by going into: start> type: manage app execution aliases> and turn off app installer
@@tordg1356 Thanks, only one that worked (:
My friend linked me this guide and boy I was never disappointed
All guides I've seen are hours and etc
This one is straightforward easy and works like a charm
Thank u
This is one of the best installation tutorials I've ever watched. Concise and coherent. Thank you for your time and effort.
penguin dragon is a GigaChad
o7
De nada
maybe yesterday, but tonight i have made too many comments for this reality to be true 😆🐧🐉❄️❄️❄️❄️
Once installed, how do you update the automatic1111 installation via git? Seems kind of important for an everchanging project and as much as I appreciate the instructions to the successful one time installation - it leaves me clueless on how to keep the automatic1111 web-ui updated (preferably without loosing my settings and/or custom ckpt-models. Worth to make a youtube video about it? Highly appreciated in advance...
This probably won't solve your trouble but 2 months after you commented that I saw this video, followed all the steps and I got the thing to work :)
I did everything you said exactly as you said it. When I entered the URL in my browser, I got the prompt "This site can't be reached."
I had this happen. I rebooted my computer and it worked afterward.
@@theroyalbureau so, I tried this. And it didn’t work :(
When you say “reboot,” you don’t mean “restart” do you..?
Like, you ACTUALLY mean “reboot?”
@@sanjaydutt678 I went to Windows Start Menu > Power > Restart and that solved it for me. I don't know why, but it worked.
@@theroyalbureau hmm.. weird.. didn’t work for me. Tried it like three times (sigh)…
stucked in a cdm line: model loaded in ...... load textual inversion embeddings:...
what should i do?
you are awesome, best installation guide ever. I'd never figure that out without help
Worth mentioning.
Most of this machine learning stuff does not play well with AMD GPUs, if you can get it to work at all.
You'll need an Nvidia card or a _very_ powerful CPU for decent performance.
By not play well you mean it will get a error message or long waiting time or you will get bad image results?
@@lightbel18 either not work or long time wait
@@lightbel18 The rendering will be done on your CPU, not on your AMD GPU. So its a couple of minutes vs a couple of seconds per image.
Is the URL always available to launch, or is there something you need to execute manually to generate the URL? My plan was to just get it installed then create a shortcut to open the GUI in my browser.
am getting the runtime error: torch is not able to use gpu any advice thanks very much
Traceback (most recent call last):
File "C:\ai\stable-diffusion-webui\launch.py", line 48, in
main()
File "C:\ai\stable-diffusion-webui\launch.py", line 39, in main
prepare_environment()
File "C:\ai\stable-diffusion-webui\modules\launch_utils.py", line 384, in prepare_environment
raise RuntimeError(
RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
Press any key to continue . . .
you have to rightclick "webui-user.bat and add --skip-torch-cuda-test to COMMANDLINE_ARGS that it should look like this
@echo off
set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS= --skip-torch-cuda-test
if you keep getting some error about half or something, add --precision full --no-half
that it looks like
call webui.bat --precision full --no-half
THANKSSSSSS@@terrorhuhn9192
@@terrorhuhn9192 God bless this man
Thought something in this process would most probably go wrong - but it was indeed just easy like this. Done after 30 minutes. Thank you so much!
How do you, re-open it? That is to say, once I've restarted my computer the local URL no longer works and I just ended up having to do the last part on CMD again. Is this normal or is there a way around this?
Trying to figure it out as well.
Like this tutorial ends with how to start it and it worked. But now I have no idea how to turn it on again. I tried running webui-user.bat again but it stops at saying "To create a public link, set `share=True` in `launch()`."
I have no idea what I'm supposed to do. I went into a file called launch and opened with notepad but there's no "share=false" or something like that.
Even though I installed python, it says "Couldn't launch Python" 9009 code. What do I do?
Any tips for stuck at installing torch & vision torch?
This is great! It's very easy to dismiss SD at a glance beacuse of the FREE pricetag. You might find it cheap in quality and lacking features. This however is the reverse when you start to dive in. You'll soon start to see how jam packed of features SD has and how fast it's overtaking the paid alternatives with it's open source community behind it.
Costing money is not at all connected to quality, especially in the FOSS community.
A lot of people do what you said, they dismiss open source stuff because it looks finicky and different, but its almost always an improvement to whatever proprietary product they're using.
Heyo I did the exact steps stated in this video but had to add --precision and --no-half to get it all running. Yet it is very slow and doesnt use my GPU (only 7% Load) can sb help me fix this? I hvae a AMD RX6750XT... Thank you in advance!
I'm surprised that someone didn't just compile all of it into one program and installer by now
I hope someone does cause this is pretty stressful as is
I don't think it works on macbooks, but unlike the other tutorials I watched, this one was straight to the point and simple, so thanks for making it this way👍
There is one out now that works on M1 and M2 machines. Expect your computer to turn into a space heater.
@@sub-jec-tiv Is there a link or something to the M1 version? Been having trouble finding it.
@@sub-jec-tiv where is the guide?
This is hands down the best How To video ever. The world would be a better place, almost like a heaven, if all tutorials would look like this. Keep up the good work mate!
Perfect; straight to the point, step by step, and no yapping. Awesome! Have this like and sub, good sir!
Do i have to redownload every time I want to use the locally sourced generator again? Or can I use the same IP/website?
I got the "RuntimeError: CUDA out of memory. Tried to allocate ..." maybe because I use a 8ram laptop. Any help to fix this?
Same for me
will you be going into detail of the stable diffusion web UI interface? or how to use scripts and libraries and that stuff? The first and only prompt I did turned out really weird and I feel kinda dumb lol it's not as easy as a browser based image generator lol
so when i click on the webui-user it load up and gets the point where it says "installing torch and torchvision" but it doesn't look like its doing anything. how do i know if its working... its been quite a long time and its just stuck on that screen. how long does this step normally take? i have a very beefy pc setup
at 2.40 on the cmd, it block at ""Textual inversion embeddings loaded(0):
Model loaded in 76.0s (calculate hash: 15.2s, load weights from disk: 50.0s, create model: 4.3s, apply weights to model: 1.0s, apply half(): 1.0s, load VAE: 0.1s, move model to device: 1.0s, load textual inversion embeddings: 3.5s).""""
Some remarks
GFPGAN is optional, Automatic's repo has CodeFormer that will download and install by itself, that imo works better.
--precision full no-half is only required for cards series GTX 16XX, that have rpoblems with half point precision, resulting in black or green image. Remove it normally as it will just increase memory usage.
--medvram can be changed to --lowvram if you still get CUDA out of memory errors
lastly, adding 'git pull' before 'call webui bat' will automatically download updates and it tends to be added something every day. Your version for example don't have the tab for training custom style or objects.
AI can be extended by downloading additional objects or styles 4kb files in 'embeddings' folder.
Hey Robert seems like you know what you are talking about so I kindly asking here.
Im looking for a decent tutorial on how to use the best of this WebUI I just installed in this video.
Im looking around but most tutorial are for the web version and not local or for another UI.
What would you recommend me? Thanks in advance for your help !
Thanks for taking the time to clarify for everyone! I didn’t know about the -lowvram argument. I have Nvidia 16XX (i think 1680) and it gave me black screens but still took forever to create, then I put in the magical first 3 arguments found on a reddit, and my ai now dreams of penguin dragons and cielmaids 🐧🐉
@@Yotanou In my experimenting, what makes or breaks the ai’s content is your use of artist names. I’m personally looking forward to the “important artist names” bonus mission from RoyalSkies this month. If you haven’t checked out RoyalSkies previous video on stable diffusion’s official site, start there!
what's annoying is I have like 8 gigs of vram and I keep getting CUDA problems
i have the black square image i see what you said but how do i actually perform that instruction? do i need to edit a text file do i need to do it in cmd or is it one of the settings on the web version?
Best tutorial clarity and pace i've seen in a while. Thank you so much
Is there a easy way to re-launch stable diffusion or do you always have to go though webui-user?
Why is there no answer to this anywhere?
@@motionsick Great question, figured it out? I have the same problem and there are no answers to it anywhere. People just try it once and that's it or what?
I have trouble to the step after clicking on the webui.user. there is an error and it is written :
[notice] A new release of pip available: 22.2.1 -> 23.1.2
[notice] To update, run: C:\IA\stable-diffusion-webui\venv\Scripts\python.exe -m pip install --upgrade pip
of course, I have no others versions of python and when I go to python command prompt and paste "-m pip install --upgrade pip", it says: invalid syntax..
Dude... It works... *press play* * listen* *pause* * do what you're told* *press play* *repeat*! Links in the description in order! Brilliant job! Thank you very much!
Great Tutorial Thx ! Installed without any problem. But i have maybe a dumb question, how can i relaunch it ? Yesterday its worked fine but today off course the link isn't good anymore. Do i need to do again every step of the tutorial ?
Thanks
BUMP. Did you find the solution? I'm in the same boat. Presumably we would need to run something along with our web browser.
Such great content! Wanted to start generating my own art and I came across this video and it helped so much. Keep up the great work!
Never followed a better tutorial than this. Usually i hate watching tutorials because of all the slow and meaningless talk, but this was amazing.
i dont even know what kind of videos you make, but this was so pain free i subscribed anyways. Thank you
Looks intimidating at first but it was stupidly easy to do. I am absolutely horrible at computer related stuff but this was super easy to follow. Keep up the great videos 👍
Thank you for your work.
I'm stuck at 2min13, when I launch "webui-usr", the window starts and says "unable to create venv in directory venv
exit code 3
stderr
the specified path is not found." I have however respected all the folder positioning in the tutorial. Thanks for your help
i have the same problem
my mistakes was inserting the wrong pathway to the python exe
maybe it doesnt have enough permission, try to run it as administrator
Same problem, did you found a solution. The one in the description ( renaming it ) didn't do it for me.
@@planebreach I solved it by creating a different user with a name in English, I had the same problem and a name in a different language
Didn't work for me unfortunately. It gives me the URL but it leads to nothing. I left the webui batch file open, but still nothing. Anyone got any advice as to why it won't work? I got no errors at all
I am having the same issue
ERROR: Could not find a version that satisfies the requirement torch==2.1.2 (from versions: 2.2.0, 2.2.0+cu121, 2.2.1, 2.2.1+cu121, 2.2.2, 2.2.2+cu121, 2.3.0, 2.3.0+cu121, 2.3.1, 2.3.1+cu121)
ERROR: No matching distribution found for torch==2.1.2
My pc crash in this point of the instalation (2:17)
My pc specs: 12gb ram
rx570 8gb vram
cpu ryzen 5 1600af
i installed in HDD 1TB
I'm stuck at 2:18
It says "...Torch is not able to use GPU.." sad.. :(
me too
late but amd gpu
This only works with Python 3.10.6. If you try to use a later version you will get the "torch error".
- Uninstall Python
- Install the proper version of Python
- Delete repo
- Clone repo
- Move chck files and .pth files to respective places
- Profit.
ERROR: Could not find a version that satisfies the requirement torch==1.13.1+cu117 (from versions: 2.0.0, 2.0.0+cu117)
ERROR: No matching distribution found for torch==1.13.1+cu117
Does this work with a normal laptop?? Or will I need a super fast computer?
Yes, finally!. (THANK you penguin dragon and royals skies ) now my ai art skills could be a bit more.. better :)
Penguin dragon is a chad !
De nada
Hey, I am stuck at "Installing torch and torchvision". I already tried both variants, one for the "normal" computers and one for the ones with less processing power but both stay stuck at that command. What could I do? Every help would be greatly appreciated
même thing
same thing
what are you do?
Great tutorial Sir! Pure info. No 10min intro about what is a computer...
Jeez. I got it to work. My problems were nothing to do with your tutorial - kudos for the work you put into this. My issue was my Python version was 3.9 so I aborted this process after reading the initial warning and installed 3.10 - my PC refused to let go of the previous path to the Python.exe and the .BATch file wouldn't work even if I specified the new path explicitly. So I rolled back to 3.9 and it worked fine. Of course prompt engineering is another mountain to climb but at least I can play around free of nags for now. Massive thank you!
Haha I already had it installed myself and it works great even if you only have a 1050TI
I got it working with a lame integrated intel card lmao
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument index in method wrapper__index_select)
I had the same problem. Did you manage to solve it?
Oh good this way we don't have to deal with the censor blur problem right?
Yup
This model is the reason there's a discord server called unstable diffusion dedicated to NSFW generations ( and like always, nsfw drives the industry, it has quite a big R&D community in it )
@@Mandraw2012 oh good. I'm happy about that. While I'm not planning to generate NSFW stuff the fact things I make can get misconstrued as it is a major turn off
Getting an error that says
"NansException: A tensor with all NaNs was produced in Unet. This could either be because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in settings>Stable Diffusion or using the no-half commandline argument to fix this issue.....
Any help? I have a 1660ti graphics card, and I downloaded stable diffusion v1.5
It won't let me install torch, it's saying "No module named pip'".
I've made sure there's no previous versions of Python on my computer and followed everything as instructed twice.
Still doesn't work when I double click the batch file, any help?
im also having this issue
i need help with this too please. anyone
same probleme over here
I searched a lot but pytorch is just not compatible with python 3.11 so the only solution is to uninstall python. and install an old version. seems like 3.9.13 works for me.
for problem 1
"1-Problem: when i launch the webui-user.bat file it says: "couldn't launch python"
1-Solution: you have to the edit webui-user file via note editor or some text editor, then you need to enter the file path of python to where it says "set PYTHON=" should look like : "set PYTHON=C:\Users\urusername\AppData\Local\Programs\Python\Python310\python"
i cant find the "set PYTHON=" within the webui-user file.
im downloading this to my local drive (D:), so I'd like to change the directory if possible.
This would be great if everything python related was as easy as it was made out in the video.
A fresh install of the python version found in the video will just lead to a bunch of new errors for windows users.
Anyone getting a "no module named encodings" error, search up System Variables, click Environment Variables, and in User variables for [user], click New and type PYTHONHOME with the Variable Value being your Python's directory (for me it's C:Users\[user]\Local\Programs\Python\Python310).
Now that I've done this, I get a new error: (AttributeError: 'str' object has no attribute 'isascii')
At this point I've given up. I officially hate python.
There is a version of Python older than 3.7 installed on your machine if you have that error. And it ended up being used for something.
I was getting "no module 'xformers'. Processing without.." three times. The program still launches, but I don't know if CFPGAN works. Also it said it couldn't load styles
I have a different error.
First of all it shows me Downloading torch==2.0.1
Downloads something weighing 2.6 GB then when it finishes it's just on 2.6 / 2.6 and nothing happens.
Im having the same exact problems
When i´m at the process at 2:31 it says ERROR: could not find a version that satisfies the requirement torch 2.0.1. ERROR No matching distribution found for torch 2.0.1. What could the error be? Otherwise i completely followed your instructions.
very sad to see the ''Torch is not able to use GPU'' Error after over an hour of trying to get this to run, truly an AMD GPU moment
What to do in these cases??
same. i did all this shit just to find out AMD cards don't work