Great video. In case anyone runs into the same problem I had, when I launched the "webui-user.bat" file, it couldn't find python, even though I clicked the path box as stated in the video. Good news is that the fix is easy and simple. Just edit the .bat file the same way the video instructs you to add the git pull, but this time find the command line "set PYTHON=". After the =, put the path to your python.exe (wherever you installed it to). For example, "set PYTHON=C:\Users\MainUser\AppData\Local\Programs\Python\Python310\python.exe"
For me the 'set PYTHON' command was already there. So adding another command actually didnt work. But assigning the file path to the set PYTHON command that was there already worked. Thanks for the help
The problem with many videos like this is that they're usually too focused. "ok i'm gonna show you how to do this with a very obscure git project". But you explained WHY we're using this fork, what this fork is, and explained every stop along the way. well done
If you are having trouble installing torch after launching the .bat file try deleting the venv folder in your stable diffusion webui folder and a new one should be generated after launching the webui-user bat file which will install a compatible version of torch.
I am getting this error: ERROR: Could not find a version that satisfies the requirement torch==1.13.1+cu117 (from versions: none) ERROR: No matching distribution found for torch==1.13.1+cu117
Kevin is always my go-to guy for all tech knowledge that I need for the demands of my job. He provides a clear guide, breaks it down into simpler, understandable chunks, and is very straight forward and professional. Thank you, Kevin!
This guy deserves credit no joke, I'm an IT engineer and still didn't know some of the useful keyboard shortcuts he uses so there's plenty of added benefit when watching his tutorials. Subscribed and highly recommend. Kevin, if you haven't already please make a tutorial for M1/2 Macs.
Welcome to IT, where you know all the things you need to know, but you can't keep up with all the damn shortcuts! I never would've thought to use that command prompt shortcut in file explorer. This guy's a genius.
i am a cyber security students and my professor always discuss with his students about new things in cyber industry. even tho you got a title and become a teacher, you always have to learn new things about it, its the fun part of this industry.
Long process, but well and clearly explained. Totally new product for me. The presentation made it quite interesting. For those that may complain about it being difficult to install i would say that it is not an application for everyone, like Word or PowerPoint. Thanks Kevin. Great video.
I would disagree. Having seen what it can do this kind of program is great for many people, especially creatives who are not always very tech savvy. Saying it's not for everyone is just trying to justify gate-keeping.
No reason for ANY application to be difficult to install, it's MAYBE complicated now cause it's not really widespread, in a year or two it will be a one click thing, like every other piece of software.
I disagree with the statement that the installation process for the product is long. It could be made more efficient and user-friendly for a wider audience. Everyone should have access to use it easily, regardless of their technical skill level. It should not be considered an application only for advanced users.
Kevin - thank you so much for the excellent explanation. Like a few people here I had a Python 310/311 issue with the path. I solved it by renaming the Python 310 app folder to 311 and referring to 311 in the BAT file (even though it was Python 310).
Very good instructions. Few things that didn't work: 1) using cmd "git clone" led to "permission denied", so I had to run cmd as admin. 2) double click on webui-user.bat didn't have any effect, so I had to run it via "git CMD". But it worked, thank you very much!
@@krezen3588 I opened "git cmd" as administrator (this is a program you installed), drilled to the right directory (the inner one of stable) and then typed the name of the file (with bat extension)
@@michaell.9027 I tried this but in both CMD and Git CMD (cd d/ X:\...) it gives an error: ====================================== Already up to date Couldn't launch Python exit code: 9009 stderr: Python not found, run without arguments to install from Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases ====================================== I install both Git 64 bit & Python 3.10.7, and checked the Path box. Any idea why it won't launch Python? Is it because it's on a different SSD on Dir X:\ rather than C:\ or D:\ ?
Thank you for the clear and concise explanations. I tried installing this before by myself and gave up with out video. Your video actually made it seem really simple, and thank you for explaining why you need each of the programs.
Thanks for a great vid!!! For all of you who asked about the -"Torch is not able to use GPU" .... what did the job for me was updating my nvidia driver - try it
Awesomesauce 👏🏻👏🏻 very easy to follow. Also if anyone is facing a Runtime error installing torch, delete the venv file which was downloaded for the wrong python version.
@@saltysalt844 It’s suppose to be in the folder where you downloaded stable diffusion. However, if you can’t find it in yours, it might be caused by a different issue.
I can't believe how easy you just made this set-up! Thanks for your clear, easy-to-follow, step-by-step instructions. You skipped me ahead to the fun part! 🙂👍
Thank you so much for this! This is the best tutorial I've seen on getting stable diffusion running. I was so confused on what to do until I found your tutorial
Coming from outside the space this AI-art field seems extremely daunting. Thanks to you I can now try my hands at creating my first image. You are a hero.
Hello there! Very good tutorial! I'm going to link it in my media for the people who ask me how to install my models :) However, just a quick detail, there's no need to rename the downloaded model as 'model', actually that can lead to confusion of people thinking they have to rename any model, and if you use many models keeping original names helps you keep things organised.
I didn't understand the hype reading the comments but now I see what it's all about. Thanks to your clear and concise tutorial I had it running in less than 10 minutes! Keep doing what you're doing!
Question. How do I know that it's actually doing something after I click 'Generate'? Is there some sort of a progress bar or some 'Please wait...' animation... thing? I just wanna know that it's processing something and not me waiting for nothing like an idiot for hours.
Wonderful precise step by step process with explanation. It was an adventure for me as I had to walk through it several times because of a python problem. It finally worked after i got rid of anaconda3, anaconda2, miniconda3, and their env variables.
I'm guessing I have the same problem... When executing the bat file, it won't run because it tells me python is not found. Sadly I need anaconda for my studies....
THANK YOU so much for the easiest and clearest step-by-step guide to installing this. I'm a novice myself, and following your steps made so much more sense than all the other videos and documents I found online.
If you have an error with "torch==1.13.1+cu117 ", it's because you downloaded a "too recent" version of python. Follow these steps : -Delete the "venv" folder in stable diffusion -Uninstall python -Install python 3.10.6
unfortunately did not work ERROR: Could not find a version that satisfies the requirement torch==1.13.1+cu117 (from versions: none) ERROR: No matching distribution found for torch==1.13.1+cu117
@@bluecobaltbackup woops, I forgot to mention that you have to uninstall python. It seems that stable diffusion auto chooses the latest version of python. Try again with these steps
I'm pretty computer illiterate and this seemed like a really daunting endeavor. However, it worked the first time through for me. Thank you for the video!!
Kevin, always like your presentation knowledge base. Just installed Stable Diffusion and ran into a glitch. Installed the model but it failed not finding a torch 2.0.1. Wondering if my python might be a later version than 3.10.6?
Thanks for the hard work on putting this together! One thing. None of the URLs will be copy-pasteable because of YT "needing" to screw them all up. You'll probably need to just right-click them and open in a new tab/window and copy them that way.
What a great tutorial! 👍Not just instructions, but also explanations. I did have some errors. But it was all my faults, e.g. didn't ensure my old Python version having been removed. Again, that's a super good video. Thanks for making such great effort to illustrate the process 👍👍👍
Kevin - this is the best stable diffusion beginner's tutorial I've found. Incredible work and thank you for being so precise with your words and instructions.
Does your GPU have to be an Nvidia GPU to use stable diffusion? Mine is an AMD Radeon GPU but it has 12 gig memory and should be more than capable of running stable diffusion. I keep getting an error saying torch is unable to use my GPU.
A week ago my car broke down in the middle of the night and a car full of mobggers pulled up and started giving me trouble. Thankfully Kevin was driving by. He pulled over and came to help. Not only did he save my life but he got those mobster muggers to turn their life around. Thank you Kevin
00:00 Learn how to install and use Stable Diffusion for AI-generated images 01:49 Two pre-requisites required for Stable Diffusion installation 03:24 Install Stable Diffusion and WebUI for easy interaction. 05:03 Download the base model for Stable Diffusion. 06:35 Install and launch Stable Diffusion 08:02 Using Stable Diffusion to generate photorealistic images from text prompts 09:33 Adjust parameters for Stable Diffusion image generation 10:59 Stable Diffusion generates different images for each prompt Crafted by Merlin AI.
@@KevinStratvert What is "a while" to complete? Seems to be stuck on "Model loaded in 121.4s (load weights from disk: 82.4s, create model: 4.6s, apply weights to model: 4.7s, apply half(): 3.6s, apply dtype to VAE: 0.2s, load VAE: 0.5s, move model to device: 13.1s, hijack: 0.3s, load textual inversion embeddings: 2.3s, calculate empty prompt: 9.6s)." and I am not seeing much use in CPU, GPU,. or disk.
I've been trying to get stable diffusion over the past few days. I keep running into an issue that doesn't allow me to download torch. I'm not sure if I messed with the code so that it doesn't work or if the torch download folder changed locations, preventing the previous path from working. I'm not sure what I'm doing wrong here. ~~~ Edit: I figured it out. I had the current version of Python, but it needed to be the 3.10.6 version. In order for the program to work, I also needed to redownload the stable diffusion folder so the system wouldn't check for the current version. I still don't know how to update to the 2.0 or 2.1 version of stable diffusion without causing crashes, but I'll figure it out by the time the next version comes out.
Not sure if you forgot to mention this in the video or not but does this tutorial cover how to install this for non NVIDIA GPU users? Is there anyway to get SD running natively with an AMD GPU or are we stuck with long wait times using the CPU? Could you do a tutorial on this if thats the case? I appreciate your help.
I tried to follow the video, which was very detailed, however, when I started stable diffusion I got an error about creating torch. Any suggestions. I put the stable diffusion on and external SDD.
Make sure you install Python 3.10.6 The link is in his description. Also you have to uninstall newer versions of python before you use the "git clone" function. Hope this helps
As always a nice and helpful video, this is no the first time I'm installing a software and initiate with a new journey of knowledge thanks to your tutorials. Thank you very much for sharing your knowledge, you do a great job guiding us step by step, to get to the final result.
Thank you! You are right, I couldn't find it at all :( Where should I change these settings and run a model again? Sorry for asking, I wish to be as smart as you guys! Will appreciate any help. Thanks to Kevin that even people like me can at least try their chances here!
Grateful for this video. Not only did it really dumb down (lol) Stable Diffusion installation, I now understand how the hell I am supposed to download anything on Git Hub. I have wanted to download things from there before but I'd open the window, see a bunch of random files with extensions I am not even familiar with, and I'd say nevermind and close it. lol. I am going to watch your other videos on it. Much appreciated!
I had a problem at 7:43 in the video where i waited very long and nothing happened (i checked my D and C drive data transfer on the task manager, but there was no data transfer). So the hint i can give you is to restart your computer after installing everything before doing the cmd thing at 7:43. Now it worked for me. And i set the name of the data "webui-user" to "webui-user.bat" because in my folder the name of the data was "webui-user". I hope this helps someone with the same problem.
I have the same problem, that step takes forever, with no progress. I restarted my pc, and the file is indeed called "webui-user.bat''. But I am still stuck...
I really like how you explain each part short and clear! I've followed another DS installation video before for my old PC but it didn't cover any explaination.
IF DON'T HAVE GPU: Very good, 4/5 just because I had to reclone the git repository for having python 3.11 and not 3.10. Also, I'm a laptop user and have no GPU in my laptop, following the tutorial has helped my greatly but I got stuck on PyTorch not letting me proceed and all I had to do was to change in the webui-user.bat file the set COMMANDLINE_ARGS= --skip-torch-cuda-test --no-half And everything worked perfectly fine.
PSA: I got a permission error while trying to clone, I had my SD folder inside of my program files folder, which didn't have permission. I made a new folder in C and it worked perfectly. Thanks for the video, it was a great help!
I'm running a desktop 3080, but out of curiosity since you only mentioned NVIDIA at 1:20, would someone running an AMD Radeon discrete GPU be able to use stable diffusion?
Hello, please I need your help, I tried to install 3 versions of stable diffusion 1.4, 1.5, and 2.0 and yet I have the same error message ( Stable diffusion model failed to load, exiting ). Thank you in advance for your help.
raise RuntimeError("Cannot add middleware after an application has started"_ RuntimeError: Cannot add middleware after an application has started. Local URL: wont launch. Pls help
Automatic1111 includes a nice UX and also contains performance optimizations for consumer grade hardware. Base Stable Diffusion is all command line and runs into performance issues.
really good video .I was stuck while downloading stable diffusion , but this video saved a lot of my time , plus it was easy as everything was so systematic...
When trying to run the webui-user.bat file these errors pop up ERROR: Could not find a version that satisfies the requirement torch==2.0.1 (from versions: none) ERROR: No matching distribution found for torch==2.0.1 Its also detecting python version 3.12.1 even though I have version 3.10.6 installed onto my computer alreadt. Anyone know the fix for this?
I tried this and got further webui-user.bat file and under the arguments type "--skip-torch-cuda-test"....but then got Warning: caught exception 'Found no NVIDIA driver on your system. Basically think the video fails to stress this won't work on AMD!
This was pretty difficult to get working for me since I have an AMD GPU and the PyTorch that came with stability diffusion didn't want to work with my version of python even though I made sure to uninstall the current version/s I had and removed everything from the environment variables before installing. The solution I found was that PyTorch actually has a version 2.0 now that works with the new versions of python and it's completely backwards compatible with the pervious versions of PyTorch. So instead of using the provided PyTorch I used the new version. Search for "stable diffusion pytorch 2.0"
Thank you so much, really helpful and easy to follow video. Honestly you got me into AI and this will change my life so thank you again. Only negative I would say which isn't huge is just get all downloads at the start so you don't have to wait when following the tutorial but other than that absolutely perfect.
I had error message :[notice] A new release of pip is available: 23.2.1 -> 23.3.2 [notice] To update, run: C:\Stable diffusion\stable-diffusion-webui\venv\Scripts\python.exe -m pip install --upgrade pip. if you have the same issue open a CMD in the C:\Stable diffusion\stable-diffusion-webui\venv\Scripts then type python.exe -m pip install --upgrade pip
thank you so much man. incredible tutorial. i am familiar with installing things on a pc but i know that there are some people out there that struggle with file placement and i know that even this is about as simple and great of a tutorial as there can be,
im getting this error when I run the .bat file AssertionError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
thanks for this , saved me the headache of trying to get it working on my own . plz make follow ups for more ways to uses and different models and so on
Great video, but I am getting 2 errors that are preventing me from continuing. 1. Fatal error about dubious ownership in repository. 2. The "Python was not found" error. I have tried re-installing the version in the video multiple times.
Thanks Kevin for this basic tutorial. its good knowledge in case you forget what is installed where. I have advanced my AI journey to new levels beyond the scope of this tutorial such as muti-gpu AI training and larger data sets that require coding, But outside of that thanks again