If you're having trouble with TORCH... I had trouble with the webui-user.bat file being unable to locate or install torch. Then I checked the dependencies listed by automatica in their github repository and noticed that it requires python 3.10.6. However, the default install on the python website, is whatever the latest version is. In my case it was python 3.11.something. So, then I uninstalled python AND python launcher. I deleted the stable diffusion folder and its contents, then searched the python website for version 3.10.6 and downloaded and installed that, and then did the git clone step again to create the folders and contents from scratch, put the "model" file in place, and now it's working. Hope that helps someone... EDIT. It takes a lonnnng time for torch to install, and in my case it was early in the morning and I was at work and my boss was not amused by my breakfast margarita. So, you might want to consider the cocktails step as optional.
@@raid5098 hope it works for you. Just be sure when uninstalling python 3.11.x that you uninstall BOTH python apps which show up in your windows app manager. Deleting the entire folder you create for stable diffusuon is also necessary because git will keep looking for the last python version it auto-detected upon installation. So, even if you uninstall python and then install again, git will keep looking for the previous one.
Installing python 3.10.8 did work. At first, I installed the latest version of python(3.11.0), but it didn’t work at all. So I deleted it and installed python 3.10.8 and it worked perfectly. I had to reinstall Stable Diffusion, though. Thank you so much! You are the best! :) Thumbs up!
I was too afraid to even try to look into this, but your tutorial is so accessible, I was able to get it going, and now I am the proud owner of many images of boxes of rocks! Oh, and I am sure you know by now, but I came here from Jazza's channel. Thank you again! :D
Thanks Scott, you removed the fear of a local install. I'm amazed at the speed of growth here. Thank goodness there are people like yourself enabling non-technical people to have access to such fascinating new digital tools. I look forward to you walking us through the many options available now and in the future.
@RealTimeX For me it is the downloading of active scripts from an unknown source. Creating an engine that links your PC to a neural network that, honestly, I know nothing about.
Man, I really enjoyed your video. Although I'm a technical guy, long time developer, I just love when people teach things as easily as you did for non-tech folks. Amazing work.
If you have multiple versions of Python, simply change the line 'set PYTHON=' to 'set PYTHON=py -3.10' of the webui-user.bat. This tells the program to use Version 3.10.x. Make sure you have Version 3.10.6 installed, not a different 3.10 Version. If you have already executed the webui-user.bat: Delete the hole folder, clone the Github repository again and make the changes to the bat-file.
I ran into this problem too, as I ran the .bat file while having the incorrect python version. Although deleting the whole folder works, you can also just remove 2 folders "tmp" and "venv" so you don't need to clone the whole thing again.
If receiving python error 9009 when launching the webui-user .bat file, you must open txt file and paste the python .exe path to the "set PYTHON=" line. To find it, search in windows start menu > "Python" > Open File location > Right click on the Python shortcut it brings you to, Open file location > Shift + Right click on "python" application , "Copy as Path". Then paste the path in the txt file behind "set Python". Save the file and relaunch, error should no longer be present.
OMG, I have gone thru HUNDREDS of videos, help, site, they all seem to make it easy and there is ALWAYS an error, your video ALMOST got it one shot, in the exception of the Python 3.10.6 issue. So I would recommend you to add a text bar during the Python section and WARN the user to READ THE PINNED COMMENT BELOW IF YOU HAVE ENCOUNTED AN ERROR. I will start following you and watch all of the cool tutorials you are teaching us how to use this amazing tool. I want to be a pro at it, thank you so much Scott Derweiler! 5 STAR!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
For me, I had two issues to resolve: 1) In Settings, I had to give a file path for the output directory. Maybe you have to create the txt2img output directory manually? Stable Diffusion wouldn't create the txt2img output directory for me. 2) I had to change the "set COMMANDLINE_ARGS=" line to "set COMMANDLINE_ARGS=--lowvram --precision full --no-half". I found this fix in Reddit. Thanks to the Reddit folks for this fix! I have a Nitro 5 laptop with a n NVIDEO GeForce GTX 1650 video card. (Side note: I probably should have a more powerful computer, but I'm in early days with Stable Diffusion.)
There's a lot of content on internet, I've been searching for a lot of tutorials and you are by far the best of them, I don't knovv your backround but you seem like you've born to be a teacher
I like the detail that you provided before launching WebUI git pull function is important. Do not forget to do this guys, it saved me a lot of trouble!
Thanks a bunch for this video! You did this so well, explaining every step slowly and clearly, so I could follow it and reproduce the steps. I had a problem in the end, but someone in the comments recommended using Anaconda, which solved that problem---obviously atm everything in this area is developing so fast a detail can change in days....thats where the comments section get useful.
Thank you so much I spent like 2 hours trying to figure out error after error. Once I followed the steps in the video it was simple and I had no issues at all
Thank you SO SO MUCH!! I was so overwhelmed about trying to figure this out from other sources. This was super helpful and thank you for putting it simple terms!
I can't thank you enough!! now I'm following up with the ring activated :) You're the best! the way you explain every step and time to take is just perfect! one of the best instructors I came across in my life. keep it up. god bless you!!
Great video! it is always good to download from the source, and you explained clearly the steps on how to do it. I am excited to use stable diffusion now!
For those using the very new python 3.11 pip does not yet support torch as of 10/29/2022 for 3.11. use python 3.8-3.10. Hopefully soon because Python 3.11 can see a 10 to 60% speed increase over previous releases. Cheers!
Also if you already ran the .bat file you will need to delete the venv folder because in will throw an error of missing python folder. Just delete it and run the .bat again to recreate it.
@@sedetweiler That said. Now that i got it all downloaded and started. It says it cant find and launch Python :( Been looking through a few comments. But it just dont seem to wanna work :(
Apparently... to fix it. Check your Python version and be sure it is installed on your machine Check the path environment variable Go to -> "start" and type "Manage App Execution Aliases". Go to it and turn off "Python" That worked.
For anyone getting errors installing or launching Stable Diffusion such as, "requirements.txt file does not exist." What fixed it for me was simple, do not put any spaces in the created folder where you install Stable Diffusion. Instead of calling the folder Stable Diffusion, call the folder: Stable-Diffusion for example, just DO NOT put any spaces. That simple fix and it works fine for me. The creator of this video should update the description to include this information.
I didn't realize it absolutely had to be with an nvidia gpu. I did find a fix to get it to work with amd but it's a little too complicated for me, maybe I'll try later. either way I really enjoyed this video! really good explenation of everything
Thank you! It was smooth sailing for installing for stable diffusion and easier one to follow.... I had so many problems with other installation of this software...
@@NickWho0323 I had to uninstall python , reinstall 3.10.6 and delete the 'venv' directory within my stable diffusion folder. Then when I ran the user batch file it worked.
@@NickWho0323 Yes delete the venv folder. That will have the version of python in the folder that you don't want. When you run the bat file it will recreate the venv folder with the correct version of python
8:35 You can add --medvram in WebUI-user command args to lower VRAM requirements at cost of speed, also Deepbooru can be used to analyze real images or any artwork not just anime, just add --deepdanbooru in WebUI-user command args and restart.
First off: AMAZING TUTORIAL~ Love your style and how simple you made things for people to follow! Thank you so much for this! 2nd: I had a hiccup with the xformers not installing and I'm not sure what that is or does but after a 2nd initialization the whole thing worked sans whatever xformers is.
Hey Scott, before I say what I'm about to say I want to thank you for making time to create this video that LOOKS easy and straightforward. So again thank you, great job! Now I love how you give all this reassurance at the beginning of the video that it's not going to get nerdy, we don't need programing and you will show everything all the way...just for me to do everything right up until the point of launch and then immediately face a myriad of errors...can't find python, can't find torch, torchvision, can't find anything! This of course will launch an investigating session of reddit & youtube that will prob take HOURS! Nice. I am not saying this is your fault or you could have foreseen this! All I'm saying that this stupid BS is why people get so intimidated by stuff like this...this is why I hate programing, despise it & loathe it...I truly do. Rant over, thanks again. :)
sorry, yeah they updated python so the new version breaks this. If you install the version in my video, or the version in the pinned comment and delete your install and re clone it, that will solve it. it was that pesky update to python that broke things. sorry about that!
Hello Scott, Thanks for sharing your knowledge. I´m having the following issue once the TORCH is intalling. " AssertionError: Toch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to desable this check " I have no idea where to put this addition of code on the COMMANDLINE_ARGS I should put it at the same .BAT file that was eddited on this tutorial?
I'm getting the same thing, tried v3.11.1, 3.10.9, 3.10.8 & 3.10.6 of python. Tried reinstalling stable diffusion too and it just keeps coming up with the same thing. Could someone help please?
Scott i followed your directions to the letter But when i run the bat file it says it can't find python any ideas? I get a 9009 error as well. Now the SD install is on a different drive... but i noticed yours was also.
I am having this same issue. Everything was great until the last step and I get "Creating venv in directory venv using python "C:\Program Files\Inkscape\bin\python.exe" venv "D:\Stable Diffusion\stable-diffusion-webui\venv\Scripts\Python.exe" The system cannot find the path specified. Press any key to continue . . ." Yes, I made sure to check the "python is in your path"
@@erikdavis9064 guess Scott hasn't seen this error before. Welp I'm still stalled I'm glad I'm not the only one that has this issue. If you get it working please let me know I'll do the same.
@@erikdavis9064 I uninstalled Python and then reinstalled through custom, using dir C;\Python That worked, so maybe it's just something about weird characters in the pathname, or it being too long
For anyone that has a problem with Python or installing Torch. Deinstall your Python version. Get the 2.10.6 version. Then go to the diffusion main folder and delete the "venv" folder because it still tries to connect to the previous Python version you had installed. After that run the bat again and have fun :)
You can generate a description of the image by loading it into img2img and clicking Interrogate CLIP next to the generate button. Then return to the txt2img. Also use Lexica to find the text people used to generate images.
Great work! The tutorial is easy to understand because the author explains the information in a simple way that anyone can understand. Well probably not everyone but i did
I found that it I didn't have at least one named model, it would yell at me. I have several models now, but I think the initial install wants to see that specifically.
@@sedetweiler That’s strange I never renamed the checkpoint. Just ensured I selected it from the drop-down. I did notice in your video that the drop-down was empty. No checkpoint selected. Maybe if nothing is selected it defaults to ‘model’ hence the error if not renamed.
@@TheDukeVega You can't select it from the dropdown unless you install it first. It's during the first install that it wants to see model.ckpt. otherwise it errors out.
And another tips is that you can have folders in the stable diffusion models folder itself. Very useful to organize many different versions of a specific model, without having to long descriptive filenames
Hi @Scott Detweiler thank you for these helpful videos. I was wondering what do we do if we had v1.4 already in our folder as model.cpkt before this update? do we delete that and change 1.5 to model? I also see some people have multiple checkpoints in their UI and switch back and forth. Right now, I have both versions in my models folder. 1.4 is named model.cpkt and this one is named the original file name. But I chose this checkpoint in the UI. Thank you for your helpful videos.
At the top there is a drop down and you can change the model. Because you have more than one, you can select the one you want for that session. You just need to have at least one named model.
@@KadayiPolokov I have not given up my 1.4 file yet, as it remains to be seen what might have been nerfed or changed in a significantly negative way. Hands still suck, but architecture and cars are much better.
WOW thanks mate - really good video and you made it super easy! I had problems with the latest Python so I went back and got an older version (10.8) as you did in the video and it worked. Cheers.
This was amazing and extremely helpful, thank you so much for taking the time to do this video! The latest version of python doesn't work, I tried with the one in your video and all is well. Offtopic, you have a great voice for radio. Cheers!
@@sedetweiler I don't know but the one you presented seems just fine to me. The creative possibilities are so vast, my brain hurts. It's like I'm lost in the infinite (oh, the drama :)) )
Useful instructions! Also, I would like to correct Scott on one thing. The model does not have to be named model.ckpt. You can simply put them in there as is and they will work.
Amazing tutorial, can't wait to dive in. Would love to see a tutorial on how to do the image sequence morph animations. These are becoming more and more popular and there are virtually no tutorials for it.
the funniest thing is the first thing I learned in this video was that you can open task manager with cntrl+shift+esc lmao i cant believe i haven't known that and I've hated not having it available in the task bar since my pc has been updated to windows 11 THANK GOD
I'm still having trouble with it. I installed Python 3.10.6 and Stable Diffusion 1.5, did all the steps shown in the video, and I'm still unable to install torch. An 'Error Code 1'. It also says there's 'No module named pip'. Does anybody have any advice?
I hit a snag, of some sort. After the git pull and the launch at 9:00 mark, my computer did not start downloading and instead loaded for a second before the black screen disappears. I get no local link. Does anyone know what went wrong here, or how to fix it?
Wonderful:) I had previously downloaded the Zip and was missing the extensions. I followed your lesson and now see the extensions, and can work with dream booth:) thanks so much for the great and clear information :)
When I follow the steps until 9:20, I get this: Creating venv in directory C:\Stable Diffusion\stable-diffusion-webui\venv using python "C:\Users\(private)\AppData\Local\Programs\Python\Python310\python.exe" Unable to create venv in directory "C:\Stable Diffusion\stable-diffusion-webui\venv" But my "(private)" user folder contains a non-latin character. Instead of the character there is a "÷" divideing sign. Is that the source of the issue? I have also tried to manually create the ...\venv\scripts\python.exe folders and then copied the exe into the last folder. Didn`t work either. Any ideas?
Do I need to keep the Command Window open, from the webui-user batch file, to be able to use Stable Diffusion in my browser? When I closed it, my browser would not open Stable Diffusion...
im trying to download the CKPT file and it gets stuck at 0.0GB and refuses to download. or itll download up to around 1GB in a few seconds then come to a halt. my download and internet speeds are fine and nothing ive done to troubleshoot has helped (changing browsers, updating drivers, restarting my computer, etc) anyone know whats going on?
RuntimeError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 2.00 GiB total capacity; 1.71 GiB already allocated; 0 bytes free; 1.75 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF. I am having this problem, Does anyone can help me?
For the high res fix. It works well At 0.15 and using the original size used for the generation of the smaller version originally (and same sees of course). Same prompt. I think it's just like img2img with low denoising strength and a higher resolution.
Newb question. When I launched the program for the first time (8:58), and I was waiting and waiting and waiting while "Installing torch and torchvision" was being displayed in the cmd box, I eventually realized that there were references in the box to an older version of Python (3.7.4) that I didn't even remember having been installed on my Win7 Pro 64-bit PC. Was I supposed to somehow direct proceedings to the newer version of Python (3.10.6)? [EDIT] After about 45 minutes of "Installing torch and torchvision" being displayed, something finally happened... and it wasn't good. Several errors were reported, including "Couldn't install gfpgan." Would my best option be to simply uninstall the earlier version of Python and try this again?
I would start there. Keep things current including your video driver. You can also just install some of those models directly. There is a dependency area on the GitHub page with links to the models.
@@sedetweiler Thanks for your reply. I uninstalled both versions of Python that had been installed, re-installed 3.10.6, but am now getting the error message (at 9:12)... "python is not recognized as an internal or external command, operable program or batch file". I had put the tick in the box during installation that you had mentioned regarding the path, but for some reason there seems to be a problem. I've tried to rectify the problem with what I've found online by going into the Environmental Variables, but so far no luck. I'll give further reports in case this helps anyone else who may be facing similar challenges.
This getting really frustrating. Because of this path issue, I decided to uninstall Python 3.10.6 again, and then re-install it again. However, now it refuses to install, saying that Windows 8.1 is required! 😮 I don't understand this at all. I downloaded the version of Python 3.10.6 that was specifically mentioned for Windows 7, installed it twice previously... and now the installer is refusing to install it. Going in circles here!
I am a Complete novice when it comes to computers. How did you create a new drive and directory for stable diffusion to be installed to? Can you put it on your main drive (C) or is that harmful/not recommended? You cover it like it should be second nature. Please someone help
hit the windows button (or click the start button at the bottom left of the desktop) and type in 'disk manager'. There should be a program called 'windows disk manager' or something similar. You can use that to create partitions in disks. The only issue is the size of the disk, make sure you have enough free space for the install. Also I'm not sure if there is an issue with partitioning the windows install but I think it'll already be its own thing? Maybe.
is anybody else getting the error when running batch file, can't install torch :( stderr: ERROR: Could not find a version that satisfies the requirement torch==1.12.1+cu113 (from versions: none) ERROR: No matching distribution found for torch==1.12.1+cu113
Stable diffsion 1.5 works great thanks for all your videos! Inidentally I came across a super interesting combination of artists, and prompt, it creates some really amazing scenes: normandy landing during World War II with Giant Robots, wide shot, detailed, intricate, sharp focus, art by Myasoyedov and alphonse mucha and greg rutkowski and repin
Worked like a champ, thank you very much. Now I need a new video card. My 1080 does ok though. I just cut and paste my prompts from elsewhere and I was back in action. Now if I only had a rich relative to buy me a 4090.
I'm getting this error when trying to run it the bat file. Help? RuntimeError: Error running command. Command: "C:\Stable Diffusion\stable-diffusion-webui\venv\Scripts\python.exe" -c "import torch; assert torch.cuda.is_available(), 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'" Error code: 1 stdout: stderr: Traceback (most recent call last): File "", line 1, in AssertionError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
stderr: ERROR: Could not find a version that satisfies the requirement torch==1.12.1+cu113 (from versions: none) ERROR: No matching distribution found for torch==1.12.1+cu113 - pls help!
If you're having trouble with TORCH...
I had trouble with the webui-user.bat file being unable to locate or install torch. Then I checked the dependencies listed by automatica in their github repository and noticed that it requires python 3.10.6. However, the default install on the python website, is whatever the latest version is. In my case it was python 3.11.something.
So, then I uninstalled python AND python launcher. I deleted the stable diffusion folder and its contents, then searched the python website for version 3.10.6 and downloaded and installed that, and then did the git clone step again to create the folders and contents from scratch, put the "model" file in place, and now it's working.
Hope that helps someone...
EDIT. It takes a lonnnng time for torch to install, and in my case it was early in the morning and I was at work and my boss was not amused by my breakfast margarita. So, you might want to consider the cocktails step as optional.
Trying this out thank you, had the same issue.
LOL breakfast margarita. I too am doing this during a slow period at work hahahaha.
Thanks a lot amigo, I had the same issue. This comment should get pinned!
@@raid5098 hope it works for you. Just be sure when uninstalling python 3.11.x that you uninstall BOTH python apps which show up in your windows app manager. Deleting the entire folder you create for stable diffusuon is also necessary because git will keep looking for the last python version it auto-detected upon installation. So, even if you uninstall python and then install again, git will keep looking for the previous one.
@@justinmccabe8708 you're very welcome. Have fun playing with it. I'm on 500 images created ao far since installing it yesterday. Lol
Installing python 3.10.8 did work.
At first, I installed the latest version of python(3.11.0), but it didn’t work at all.
So I deleted it and installed python 3.10.8 and it worked perfectly.
I had to reinstall Stable Diffusion, though.
Thank you so much! You are the best! :)
Thumbs up!
Super solid tutorial. I appreciate that you're self aware and that you talk to us realistically, without lots of assumptions that we're developers
Glad you enjoyed it!
I was too afraid to even try to look into this, but your tutorial is so accessible, I was able to get it going, and now I am the proud owner of many images of boxes of rocks! Oh, and I am sure you know by now, but I came here from Jazza's channel. Thank you again! :D
Thanks Scott, you removed the fear of a local install. I'm amazed at the speed of growth here. Thank goodness there are people like yourself enabling non-technical people to have access to such fascinating new digital tools. I look forward to you walking us through the many options available now and in the future.
Oh awesome! Well, so many videos are coming on that to be sure.
@RealTimeX For me it is the downloading of active scripts from an unknown source. Creating an engine that links your PC to a neural network that, honestly, I know nothing about.
btw now you have to register at hugging face in addition for downloading the weights
That has always been the case. You must agree to the usage terms in order to download it.
@@sedetweiler ah i c. thx for celearing that up.
Man, I really enjoyed your video. Although I'm a technical guy, long time developer, I just love when people teach things as easily as you did for non-tech folks. Amazing work.
This worked flawlessly. Thank you for bringing ease of installation to folks who just wanna make some art.
That's great news!
@@sedetweiler How do you apply this to a video instead of images?
Tks Scott! Very easy guide to follow! Up and running in a couple of minutes. Very straightforward! Thank you for sharing this!
Thank you so much! Very kind of you! 🥂
If you have multiple versions of Python, simply change the line 'set PYTHON=' to 'set PYTHON=py -3.10' of the webui-user.bat. This tells the program to use Version 3.10.x. Make sure you have Version 3.10.6 installed, not a different 3.10 Version. If you have already executed the webui-user.bat: Delete the hole folder, clone the Github repository again and make the changes to the bat-file.
Lifesaver. Worked a charm.
I ran into this problem too, as I ran the .bat file while having the incorrect python version. Although deleting the whole folder works, you can also just remove 2 folders "tmp" and "venv" so you don't need to clone the whole thing again.
This needs more likes, this absolutely saved my ass I was so confused. Thank you
Yes that worked for me thank you!
python for some reason isnt letting me download 3.10
Please make tutorials for local install of PyTTI & DIsco Diffusion as well. Your tutorial style is way easier to follow than most
Dying, DYING for a disco diffusion installation that works.
If receiving python error 9009 when launching the webui-user .bat file, you must open txt file and paste the python .exe path to the "set PYTHON=" line.
To find it, search in windows start menu > "Python" > Open File location > Right click on the Python shortcut it brings you to, Open file location > Shift + Right click on "python" application , "Copy as Path". Then paste the path in the txt file behind "set Python". Save the file and relaunch, error should no longer be present.
thank you!
OMG, I have gone thru HUNDREDS of videos, help, site, they all seem to make it easy and there is ALWAYS an error, your video ALMOST got it one shot, in the exception of the Python 3.10.6 issue. So I would recommend you to add a text bar during the Python section and WARN the user to READ THE PINNED COMMENT BELOW IF YOU HAVE ENCOUNTED AN ERROR. I will start following you and watch all of the cool tutorials you are teaching us how to use this amazing tool. I want to be a pro at it, thank you so much Scott Derweiler! 5 STAR!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
For me, I had two issues to resolve:
1) In Settings, I had to give a file path for the output directory. Maybe you have to create the txt2img output directory manually? Stable Diffusion wouldn't create the txt2img output directory for me.
2) I had to change the "set COMMANDLINE_ARGS=" line to "set COMMANDLINE_ARGS=--lowvram --precision full --no-half". I found this fix in Reddit. Thanks to the Reddit folks for this fix! I have a Nitro 5 laptop with a n NVIDEO GeForce GTX 1650 video card. (Side note: I probably should have a more powerful computer, but I'm in early days with Stable Diffusion.)
As someone who didn't need nearly this much instruction I still find you made a wonderful video I feel my daughter could have installed this
⚠Update Info⚠
The current version of Python that works with Automatic1111's web ui is:
3.9.13
❤
There's a lot of content on internet, I've been searching for a lot of tutorials and you are by far the best of them, I don't knovv your backround but you seem like you've born to be a teacher
aww, thank you! you made my day! 🙂
@@sedetweiler Hi, do you knovv about an easy vvay to use this stable difussion in video? :)
You are doing a great job making this accessible to many more people. Well done and thank you!!
You are quite welcome!
@@sedetweiler would love an updated dreambooth tutorial aswell. I wanna make my wife in SD :)
@@MysteryFinery that is coming, but the tech needs a bit more love. Give it a week.
I like the detail that you provided before launching WebUI git pull function is important. Do not forget to do this guys, it saved me a lot of trouble!
I see so many videos with people using the zip file and I am totally lost as to why. This is so much easier and keeps things current.
Thanks a bunch for this video! You did this so well, explaining every step slowly and clearly, so I could follow it and reproduce the steps. I had a problem in the end, but someone in the comments recommended using Anaconda, which solved that problem---obviously atm everything in this area is developing so fast a detail can change in days....thats where the comments section get useful.
Thank you so much I spent like 2 hours trying to figure out error after error. Once I followed the steps in the video it was simple and I had no issues at all
Awesome! Glad it helped!
Thank you SO SO MUCH!! I was so overwhelmed about trying to figure this out from other sources. This was super helpful and thank you for putting it simple terms!
Glad it was helpful!
Shad recommended your video. Totally worth it.
I can't thank you enough!! now I'm following up with the ring activated :) You're the best! the way you explain every step and time to take is just perfect! one of the best instructors I came across in my life. keep it up. god bless you!!
Wow! Thank you for that compliment! Cheers!
Great video! it is always good to download from the source, and you explained clearly the steps on how to do it. I am excited to use stable diffusion now!
For those using the very new python 3.11 pip does not yet support torch as of 10/29/2022 for 3.11. use python 3.8-3.10. Hopefully soon because Python 3.11 can see a 10 to 60% speed increase over previous releases. Cheers!
Also if you already ran the .bat file you will need to delete the venv folder because in will throw an error of missing python folder. Just delete it and run the .bat again to recreate it.
thanks to the both of you. I was fn pulling my hair out trying to find out why I wasn't getting the same results.
@@ramiroandresvidela6592 Ah thank you very much, saved me a lot of time looking for a solution
@@ramiroandresvidela6592 I'm sorry but how do i delete the venv folder, i don't really know where it is ?
@@Arthur-nb7ph it's in the same folder where you clone the project
Man... You have a gift for making confusing things super easy. Ive met 2 other people that had this gift.
Glad it was easy to understand! 🥂
@@sedetweiler That said. Now that i got it all downloaded and started. It says it cant find and launch Python :( Been looking through a few comments. But it just dont seem to wanna work :(
Apparently... to fix it.
Check your Python version and be sure it is installed on your machine
Check the path environment variable
Go to -> "start" and type "Manage App Execution Aliases". Go to it and turn off "Python"
That worked.
This is what I was looking for....super easy to follow.
For anyone getting errors installing or launching Stable Diffusion such as, "requirements.txt file does not exist." What fixed it for me was simple, do not put any spaces in the created folder where you install Stable Diffusion. Instead of calling the folder Stable Diffusion, call the folder: Stable-Diffusion for example, just DO NOT put any spaces. That simple fix and it works fine for me. The creator of this video should update the description to include this information.
Youre so sweet and caring at the start of the video, it made me like and subscrible to your channel instantly, thank you for you guidance Sir !
Awww! Thank you!
I didn't realize it absolutely had to be with an nvidia gpu. I did find a fix to get it to work with amd but it's a little too complicated for me, maybe I'll try later.
either way I really enjoyed this video! really good explenation of everything
Thank you a lot! Especially for the small details that make the results better.
I think they are a nice step in the right direction for sure.
Thank you! It was smooth sailing for installing for stable diffusion and easier one to follow.... I had so many problems with other installation of this software...
That's great news!
Doesn't work. Installed python 3.10.6 , Did everything as he says. Atlast the webui-user.bat says "Couldn't install torch" and "No module named pip".
This was great. I've had issues with git before due to some inaccurate instructions. Thanks for making it super clear.
You're welcome!
Hey Scott, Great video! I do have a question, I am receiving an error when it comes to installing torch and torchvision. What did I do wrong?
I had the same problem. You have to install python 3.10.8. It did not work with python 3.11.0
@@XENoVAR Thanks for helping! However, replaced the 3.11.0 version with the 3.10.8 version, now it's telling my it can't find python! Any advice?
@@NickWho0323 I had to uninstall python , reinstall 3.10.6 and delete the 'venv' directory within my stable diffusion folder. Then when I ran the user batch file it worked.
@@vovi547 Thank you sir! I'll give that a try
@@NickWho0323 Yes delete the venv folder. That will have the version of python in the folder that you don't want. When you run the bat file it will recreate the venv folder with the correct version of python
Very precise and to-the-point tutorial.
8:35 You can add --medvram in WebUI-user command args to lower VRAM requirements at cost of speed, also Deepbooru can be used to analyze real images or any artwork not just anime, just add --deepdanbooru in WebUI-user command args and restart.
Thanks for the tips!
@@sedetweiler You are welcome 👍
@@blackvx Haven't tried this method (I just edit the file itself) but if it adds them to the command line args then yeah, that's it!
First off: AMAZING TUTORIAL~ Love your style and how simple you made things for people to follow! Thank you so much for this!
2nd: I had a hiccup with the xformers not installing and I'm not sure what that is or does but after a 2nd initialization the whole thing worked sans whatever xformers is.
Also, important note. These instructions fail with python 3.11.0.
Thank you so much for the easy guide I followed along and am now having a lot of fun with SD.
Glad to hear!
that was cool thank you Scott you did a great job explaining this to us
Happy to help!
Hey Scott, before I say what I'm about to say I want to thank you for making time to create this video that LOOKS easy and straightforward. So again thank you, great job!
Now I love how you give all this reassurance at the beginning of the video that it's not going to get nerdy, we don't need programing and you will show everything all the way...just for me to do everything right up until the point of launch and then immediately face a myriad of errors...can't find python, can't find torch, torchvision, can't find anything! This of course will launch an investigating session of reddit & youtube that will prob take HOURS! Nice.
I am not saying this is your fault or you could have foreseen this! All I'm saying that this stupid BS is why people get so intimidated by stuff like this...this is why I hate programing, despise it & loathe it...I truly do.
Rant over, thanks again. :)
sorry, yeah they updated python so the new version breaks this. If you install the version in my video, or the version in the pinned comment and delete your install and re clone it, that will solve it. it was that pesky update to python that broke things. sorry about that!
Can You Also Do A Video For AMD video card?
Hello Scott, Thanks for sharing your knowledge.
I´m having the following issue once the TORCH is intalling.
" AssertionError: Toch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to desable this check "
I have no idea where to put this addition of code on the COMMANDLINE_ARGS
I should put it at the same .BAT file that was eddited on this tutorial?
resolved this error by updating the NDVIDIA DRIVE
When I run Stable Diffusion it sayd error code 9009 can't run Python
I'm getting the same thing, tried v3.11.1, 3.10.9, 3.10.8 & 3.10.6 of python. Tried reinstalling stable diffusion too and it just keeps coming up with the same thing. Could someone help please?
That was easy. Thank you Scott! I got Rebelle Pro 5 too and I'm ready to go! looking forward for your next videos! 👍
Awesome! That was just an insane sale!
Scott i followed your directions to the letter But when i run the bat file it says it can't find python any ideas? I get a 9009 error as well. Now the SD install is on a different drive... but i noticed yours was also.
Door you check the box to make sure python is in your path?
@@sedetweiler yes i did I even uninstalled it and reinstalled it just to make sure
I am having this same issue. Everything was great until the last step and I get "Creating venv in directory venv using python "C:\Program Files\Inkscape\bin\python.exe"
venv "D:\Stable Diffusion\stable-diffusion-webui\venv\Scripts\Python.exe"
The system cannot find the path specified.
Press any key to continue . . ."
Yes, I made sure to check the "python is in your path"
@@erikdavis9064 guess Scott hasn't seen this error before. Welp I'm still stalled I'm glad I'm not the only one that has this issue. If you get it working please let me know I'll do the same.
@@erikdavis9064 I uninstalled Python and then reinstalled through custom, using dir C;\Python
That worked, so maybe it's just something about weird characters in the pathname, or it being too long
Hello Scott! Thank you so much! On 9:10 I have a problem cmd write me exit code 9009. That`s wrong?
For anyone that has a problem with Python or installing Torch. Deinstall your Python version. Get the 2.10.6 version. Then go to the diffusion main folder and delete the "venv" folder because it still tries to connect to the previous Python version you had installed. After that run the bat again and have fun :)
Did you mean 3.10.6 or 2.10.6?
I don't have a venv folder and when i try to launch i get an error. C:\Python27\python.exe: No module named venv
thank you, good sir. tried with python 3.11 and didnt work, then went to 3.10.8 you you used and it worked like a charm.
cheers!
You can generate a description of the image by loading it into img2img and clicking Interrogate CLIP next to the generate button. Then return to the txt2img. Also use Lexica to find the text people used to generate images.
Great work! The tutorial is easy to understand because the author explains the information in a simple way that anyone can understand. Well probably not everyone but i did
You don’t need to rename the checkpoint file. There is now a drop down so you can add different models to the SD models directory.
I found that it I didn't have at least one named model, it would yell at me. I have several models now, but I think the initial install wants to see that specifically.
@@sedetweiler That’s strange I never renamed the checkpoint. Just ensured I selected it from the drop-down.
I did notice in your video that the drop-down was empty. No checkpoint selected. Maybe if nothing is selected it defaults to ‘model’ hence the error if not renamed.
@@TheDukeVega You can't select it from the dropdown unless you install it first. It's during the first install that it wants to see model.ckpt. otherwise it errors out.
And another tips is that you can have folders in the stable diffusion models folder itself.
Very useful to organize many different versions of a specific model, without having to long descriptive filenames
This is for NVidia, better add it in the name of the video. Lets not asume everyone have or like to have nvidia video card :) good tutorial btw.
Hi @Scott Detweiler thank you for these helpful videos. I was wondering what do we do if we had v1.4 already in our folder as model.cpkt before this update? do we delete that and change 1.5 to model? I also see some people have multiple checkpoints in their UI and switch back and forth. Right now, I have both versions in my models folder. 1.4 is named model.cpkt and this one is named the original file name. But I chose this checkpoint in the UI. Thank you for your helpful videos.
At the top there is a drop down and you can change the model. Because you have more than one, you can select the one you want for that session. You just need to have at least one named model.
You can move between them. It's worth hanging onto 1.4 becasuse you might find it works better with a prompt versus 1.5.
@@KadayiPolokov I have not given up my 1.4 file yet, as it remains to be seen what might have been nerfed or changed in a significantly negative way. Hands still suck, but architecture and cars are much better.
After whole night I finally got it running, Thanks man u are a life saver. I subbed and liked.
Glad I could help
7:30 - I don't have permissions? how to run cmd from that place as admin?
WOW thanks mate - really good video and you made it super easy! I had problems with the latest Python so I went back and got an older version (10.8) as you did in the video and it worked. Cheers.
This was amazing and extremely helpful, thank you so much for taking the time to do this video! The latest version of python doesn't work, I tried with the one in your video and all is well. Offtopic, you have a great voice for radio. Cheers!
Thank you, and I am glad you figured it out! I wonder how long it will be until they get the current version working?
@@sedetweiler I don't know but the one you presented seems just fine to me. The creative possibilities are so vast, my brain hurts. It's like I'm lost in the infinite (oh, the drama :)) )
brilliant, thank you Scott.
Useful instructions!
Also, I would like to correct Scott on one thing. The model does not have to be named model.ckpt. You can simply put them in there as is and they will work.
Amazing tutorial, can't wait to dive in. Would love to see a tutorial on how to do the image sequence morph animations. These are becoming more and more popular and there are virtually no tutorials for it.
Noted! I will see what I can do!
Scott, your video is clear and simple and such a big help. A very big thank you.
You're welcome!
Could not get it to install torch with most recent Python (3.11.0) as it kept giving me errors. Version 3.10.8 works as shown in the video.
the funniest thing is the first thing I learned in this video was that you can open task manager with cntrl+shift+esc lmao i cant believe i haven't known that and I've hated not having it available in the task bar since my pc has been updated to windows 11 THANK GOD
Do you know how to add models? I see so many on Hugging Face but not much talk about like, HOW you add them to this
I'm still having trouble with it. I installed Python 3.10.6 and Stable Diffusion 1.5, did all the steps shown in the video, and I'm still unable to install torch. An 'Error Code 1'. It also says there's 'No module named pip'. Does anybody have any advice?
I have the same issue, did you managed to solve it?
@@marcoconti313 Still fidgeting around with everything, redid the process, but I couldn't fix it
I hit a snag, of some sort. After the git pull and the launch at 9:00 mark, my computer did not start downloading and instead loaded for a second before the black screen disappears. I get no local link. Does anyone know what went wrong here, or how to fix it?
Thanks so much for this. I've been meaning to wean myself from Midjourney and this may do it!! Very clear and easy to follow instructions. Rock on.
You can do it!
Wonderful:) I had previously downloaded the Zip and was missing the extensions. I followed your lesson and now see the extensions, and can work with dream booth:) thanks so much for the great and clear information :)
Glad it helped!
man love your tutorial so simple and streightforward.
Glad you found it helpful!
When I follow the steps until 9:20, I get this:
Creating venv in directory C:\Stable Diffusion\stable-diffusion-webui\venv using python "C:\Users\(private)\AppData\Local\Programs\Python\Python310\python.exe"
Unable to create venv in directory "C:\Stable Diffusion\stable-diffusion-webui\venv"
But my "(private)" user folder contains a non-latin character. Instead of the character there is a "÷" divideing sign. Is that the source of the issue?
I have also tried to manually create the ...\venv\scripts\python.exe folders and then copied the exe into the last folder. Didn`t work either.
Any ideas?
1:00 I'm only seeing 512MB on my laptop. Should I not even bother trying with this computer?
thank you soo soo much! other videos intimidated me but you made this very simple. Thanks! Waiting for it to load up hehe
Thank you! It worked on 6/25/2024. I was very skeptical going into this.... LOL.. fun stuff!!!
Is there a away to benchmark this? I got a 1060 6gb and I find it very slow to do a batch with almost nothing changed. I got about 20/20 [01:35
Do I need to keep the Command Window open, from the webui-user batch file, to be able to use Stable Diffusion in my browser? When I closed it, my browser would not open Stable Diffusion...
im trying to download the CKPT file and it gets stuck at 0.0GB and refuses to download. or itll download up to around 1GB in a few seconds then come to a halt. my download and internet speeds are fine and nothing ive done to troubleshoot has helped (changing browsers, updating drivers, restarting my computer, etc)
anyone know whats going on?
i cant seem to install torch
New folder\stable-diffusion-webui\venv\Scripts\python.exe: No module named pip
i get a error code 1
RuntimeError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 2.00 GiB total capacity; 1.71 GiB already allocated; 0 bytes free; 1.75 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF. I am having this problem, Does anyone can help me?
For the high res fix. It works well At 0.15 and using the original size used for the generation of the smaller version originally (and same sees of course). Same prompt.
I think it's just like img2img with low denoising strength and a higher resolution.
Thanks for making an easy to follow tutorial! Really appreciated the pacing
latest version of python does not work - roll it back one 10.8
Hey Scott thanks for the guide, but i keep getting this error after running the bat file
AttributeError: 'NoneType' object has no attribute 'keys'
Thank you for this, very smart solution to check for updates upon launch!
I think so. It keeps the goodness coming!
How do you use the TRAIN tab?
I keep getting: No module named pip error help!
Cheers for this tutorial! Clear concise instructions that are easy to follow. Perfect.
Newb question. When I launched the program for the first time (8:58), and I was waiting and waiting and waiting while "Installing torch and torchvision" was being displayed in the cmd box, I eventually realized that there were references in the box to an older version of Python (3.7.4) that I didn't even remember having been installed on my Win7 Pro 64-bit PC. Was I supposed to somehow direct proceedings to the newer version of Python (3.10.6)?
[EDIT] After about 45 minutes of "Installing torch and torchvision" being displayed, something finally happened... and it wasn't good. Several errors were reported, including "Couldn't install gfpgan." Would my best option be to simply uninstall the earlier version of Python and try this again?
I would start there. Keep things current including your video driver. You can also just install some of those models directly. There is a dependency area on the GitHub page with links to the models.
@@sedetweiler Thanks for your reply. I uninstalled both versions of Python that had been installed, re-installed 3.10.6, but am now getting the error message (at 9:12)... "python is not recognized as an internal or external command, operable program or batch file". I had put the tick in the box during installation that you had mentioned regarding the path, but for some reason there seems to be a problem. I've tried to rectify the problem with what I've found online by going into the Environmental Variables, but so far no luck.
I'll give further reports in case this helps anyone else who may be facing similar challenges.
This getting really frustrating. Because of this path issue, I decided to uninstall Python 3.10.6 again, and then re-install it again. However, now it refuses to install, saying that Windows 8.1 is required! 😮 I don't understand this at all. I downloaded the version of Python 3.10.6 that was specifically mentioned for Windows 7, installed it twice previously... and now the installer is refusing to install it. Going in circles here!
I did my install as I was watching you!
awesome! get things working?
I am a Complete novice when it comes to computers. How did you create a new drive and directory for stable diffusion to be installed to? Can you put it on your main drive (C) or is that harmful/not recommended? You cover it like it should be second nature. Please someone help
hit the windows button (or click the start button at the bottom left of the desktop) and type in 'disk manager'. There should be a program called 'windows disk manager' or something similar. You can use that to create partitions in disks. The only issue is the size of the disk, make sure you have enough free space for the install. Also I'm not sure if there is an issue with partitioning the windows install but I think it'll already be its own thing? Maybe.
is anybody else getting the error when running batch file, can't install torch :(
stderr: ERROR: Could not find a version that satisfies the requirement torch==1.12.1+cu113 (from versions: none)
ERROR: No matching distribution found for torch==1.12.1+cu113
Try using the version of Python I show in the video, as the new one might be a bit broken at the moment.
@@sedetweiler thank you Scott, that fixed it :)
Stable diffsion 1.5 works great thanks for all your videos! Inidentally I came across a super interesting combination of artists, and prompt, it creates some really amazing scenes:
normandy landing during World War II with Giant Robots, wide shot, detailed, intricate, sharp focus, art by Myasoyedov and alphonse mucha and greg rutkowski and repin
image size was 832 wide by 512 tall
Such a helpful easy to follow tutorial. Thank you for the effort.
Worked like a champ, thank you very much. Now I need a new video card. My 1080 does ok though. I just cut and paste my prompts from elsewhere and I was back in action. Now if I only had a rich relative to buy me a 4090.
someone who explain github so ANYONE can understand... im seriously shocked. you fuckin rock
works on Python 3.10.8 / GTX3070 too! - great tutorial thanks!
What does the face restoration at 14:31 do if i turn this on?
i cant get past 9:27 because i keep getting errors in the cmd window, i've followed to a tea
I'm getting this error when trying to run it the bat file. Help?
RuntimeError: Error running command.
Command: "C:\Stable Diffusion\stable-diffusion-webui\venv\Scripts\python.exe" -c "import torch; assert torch.cuda.is_available(), 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'"
Error code: 1
stdout:
stderr: Traceback (most recent call last):
File "", line 1, in
AssertionError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
stderr: ERROR: Could not find a version that satisfies the requirement torch==1.12.1+cu113 (from versions: none)
ERROR: No matching distribution found for torch==1.12.1+cu113 - pls help!