Great overview of installing and running stable diffusion locally, but judging by the issues in the comments I think it needs to cover what to do if a newer version of Python is already installed. I am an absolute beginner :) but the way I got around the PyTorch version issue was to install Python 3.10.6 to a different directory to my current python installation, selecting the Add to Path checkbox during installation. Then, before running webui-user.bat, I had to edit my user environment variables by selecting the Python 3.10.6 path and clicking the Move Up button so that it appeared before my Python 3.12 path in the list. Then stable diffusion could be installed OK, although I had to update my NVidia drivers before it would generate images. I hope this helps somebody!
I searched the internet for 45 minutes and figured out the issue. Then checked the comments and saw this, which would have saved me the 45 minutes. Ha. I ended up uninstalling 3.12 and installing 3.10.6, which worked for me. Version 3.12 gets the Torch error.
@neilie6593 -- I have a feeling that this addresses the problem that I am having after doing the same (installing the latest build of Python, but the command prompt window saying that Python could not be found). Alas, I don't understand most of what you said after describing your problem. I downloaded 3.10.6, but I don't know where I should install it, nor how one edits "user environment variables", etc. Any chance the directions could be broken-down a bit for we mere mortals?😅 Thanks!
I have tried several times and keep getting this message: 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) ERROR: No matching distribution found for torch==2.1.2
@@3sguitar It's telling me "This program is tested with 3.10.6 Python, but you have 3.12.3." I'm going to try uninstalling python and re-downloading the 3.10.6 version
Gotta remove the new version of Python, and pick a 3.10 version. Also remember to delete the venv folder in your stable diffusion ui folder, or else it will continue looking for your Python in a 3.12 folder (or whatever version you previously downloaded). When you double click the bat file to start again, it'll create a new venv folder with the new 3.10 location.
Smart pick using Stable Diffusion and not the commercial/paid tools! 👏🏽👏🏽 Pay no attention to the unsubbers. I'm in the vfx field and aeeing content from a vfx artist trying to integrate ai into their workflow is good to see... because its exactly what the big studios are already doing. So the same short-sighted naysayers here are the same types who were yelling back in the day to shun the Photoshop artists...digital cameras and digital cinematographera...and any artist really who used the computer. Heard these types of complaints thru the years so many times...and they fade away quick because they get left behind. "Adapt or die" in our field has always been true, and still is. Ai isn't going to take your artist job....an artist that knows how to properly use Ai is going to take your artist job. It's an uncomfortable fact, but it's happening.
Sorry but you are wrong. You can’t compare cameras, photoshop with ai. Does cameras used every painting to make photos? Was photoshop built on others people artwork? „Adapt or die“ yes artists are adapting thats why ai techfirms getting sued. And btw your „artist using ai“ will be gone very fast too. If ceos can easy use ai they don’t need an „ai artist“
The thing is though, photoshop and cameras still need talent and need to be used in combination with other skills like lighting, composition, framing, technique, colors, etc. All ai needs is a sentence, its a way different level of effort and very much not the same thing
@@AironyAi Unless we get AGI (real Ai) they are good. These LLM's are not very useful for custom stuff and involves more work to make sure they aren't screwing up. VFX crews use these models for very simple, boring tasks that are repetitive in nature and allows that person to do a more productive job.
Hi There! Thanks for the great and easy to follow video. I have a problem though and I hope you can help, when I run the webui-user (mine does not say .bat) it does say windows batch file after the date and time So when run it I am getting "could not find a version that satisfies the requirements for torch==2.0.1 ( from versions: 2.2.0, 2.2.0+cu118, 2.2.1, 2.2.1+cu118) ERROR: No matching distribution found for torch==2.0.1 Then some white text and ERROR code 1 press any key to continue, I press, it closes down... what have done wrong is my computer missing something to be able to run it? I'm not that technical, I only learn what I have to to do the things I want to do and I would love to try stable diffusion.
confusing. Web UI gives error that it was designed using python 3.10. So I deleted python 3.12.1 and installed 3.10 But now when running Web UI it says it cant find folder of python 3.12.1...which makes sense as I deleted it and installed 3.10 (which is in the proper Programs folder) but apparently this web UI is not compatible with python 3.10? argh. any ideas? thanks
@@chrtsng3722 there are quite a few other videos with slightly different instructions on installing all this so I will try those. If I'm successful I'll let you know.
@@SurfacedStudio i did that. Completely removed 3.12.1 and installed 3.10. I double checked that a 3.10 folder had been created in the Programs folder… but command prompt web ui says cant install as python 3.12.1. not found. Thats why I wondered if earlier version of web ui is needed instead but the list of past versions doesnt make it easy to see which is the correct to use. thanks. To confusing for me.
I think I found a fix. I had the same thing happen to me. I whet to C:\Users\(YOUR NAME)\AppData\Local\Programs\Python. then i changed the folder name from Python310 to Python312. I think it is looking for the new vision of Python even though it is incompatible so you have to trick it. If u cant tell i don't know much about Python lol. @@pixelkat1819
Great video thank you but I'm having trouble downloading stable diffusion there is no webUI.bat file. There is one called webui-user and it's a windows batch file. When I try to open it windows 11 says defender smart screen prevented an unrecognised app from starting. Running this app might put your PC at risk. The only box I can take on is don't run or the close the window x button. I've installed python. any ideas .thanks
Hi! I got a very similar screen. It wouldn't show the "run anyway" button unless I clicked on the little v arrow to get more information. Once i clicked and it revealed a few lines of information about the name of the program, the button appeared. I guess Windows is trying to make you think harder before you run stuff?
when i down the stable diffusion , it gives me a file called "sd_xl_base_1.0_0.9vae.safetensors" and I cant open it as it is asking for a file application to open it. can someone please help ?
Important! Remember to delete the venv folder in your stable diffusion ui folder, or else it will continue looking for your Python in a 3.12 folder (or whatever version you previously downloaded).
hey, thanks for this amazing tutorial. I've been trying to follow the steps, but when I press run for webui-user.bat I met two errors: [1]installing torch and torch vision - where torch couldn't be fetched. [2] an error checking the latest version of pip. wondering if you have any experience on those two issues? thank you
Hm can’t say I have unfortunately. The failing to fetch sounds like a networking issue and the pip versioning seems like some local tools not working. What OS are you on?
I had the same issue. I just had to delete python 3.12 and install version 3.11.5 instead. Then I also had to install Git. But after that the script ran through. It works for me now. I was just able to generate my first smiling cat 😸
If you need to have multiple versions of python running, you need to run the pip command like: `python -m pip --python \python.exe install pytorch`. Then in webui-user.bat, set the python variable to point to your python.exe
I would love to see how you can input images to help train it to a certain direction, for example as a game developer i'd love to input my previous made textures to get similar results, as my artistic style is quite unique
you use another program in combination with Stable Diffusion called “Kohya” to locally train AI models based on pictures you give it of a specific person or thing. Then you call these models when generating a prompt in Stable Diffusion to generate art with that specific person or thing in it! It’s a bit of a learning curve doing the training but generally you need about a dozen or more pictures to do training! The trained AI files Kohya generates are called “LORA” files. After you create a bunch you can easily browse all your previously trained “LORA” files within Stable Diffusion and click on one to easily load it into your prompt, so you can always create new art of that person or thing contained in that particular LORA file.
The way you explained is perfect for a layman like me. Before coming to your video, I saw many, but none could make a layman understand sd so easily as you did. U deserve a line and su subscribe. Looking forward for more AI stuffs in simple language.
When I try to open webui-user.bat file I get the message saying "No module named pip" and "Launch unsuccessful. Exiting." Does anybody face the same issue? It happens when I use Python 3.11.5 as well as 3.12.3
I am having an issue with the xl model. Stable Diffusion runs fine with the model that it comes along with (1.5V prunned) but once i try to change to xl i get an error "Connection errored out" is there something i am perhaps missing?
I successfully entered the 8:50 webpage ui, but when importing v1-5-pruned-emaonly, I got an error Expecting value: line 1 column 1 (char 0). Does anyone know what I should do?
So I followed every step so far. I click on webui, it runs till the end. But it's telling me "Installing torch and torchvision", "Couldn't install torch",
I follow the steps but looks like it's not working well with the last version of Python 3.12.3 because can't install it: 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 . . .
Thanks for your excellent video. I was able to get everything installed despite the python version and torch snafu's how after entering a very simple prompt all I ever get is "in queue" and nothing happens after an hour or so. So after an install that took about two hours, I am simply giving up and going back to some free web based AI generators. Thanks again.
I have a problem that when I try to run the program, this appears: Couldn't launch python exit code: 9009 stderr: Python se nepovedlo najÝt. Spu Launch unsuccessful. Exiting. Press any key to continue . . . does anyone know how to solve it?
uninstall your copy of python and use a lower version. I went with 3.11.10 and it worked. Just delete your webui-master folder and start with a new one
You might not have the right version of Python installed. Check with the SD version you downloaded what version of python you need. Then uninstall any you might already have and only install that specific one - that seems to resolve it for most people
Thanx for this tutorial. On Linux it installs like a charm, there is a script that makes it easy, even with a higher version of python it works perfect 👌
I keep getting the error: "File "C:\Users yan1\OneDrive\Documents\stableDiffusion\stable-diffusion-webui-master\stable-diffusion-webui-master\launch.py", line 48, in main() File "C:\Users yan1\OneDrive\Documents\stableDiffusion\stable-diffusion-webui-master\stable-diffusion-webui-master\launch.py", line 39, in main prepare_environment() File "C:\Users yan1\OneDrive\Documents\stableDiffusion\stable-diffusion-webui-master\stable-diffusion-webui-master\modules\launch_utils.py", line 386, 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 . . ." whenever i run the webui-user batch file, any ideas why?
I try to installed again but looks like it's not working anymore with the recently version of Python: Creating venv in directory C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv using python "C:\Users\Kedem Dray\AppData\Local\Programs\Python\Python312\python.exe" venv "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv\Scripts\Python.exe" ============================================================================================================================= INCOMPATIBLE PYTHON VERSION This program is tested with 3.10.6 Python, but you have 3.12.3. If you encounter an error with "RuntimeError: Couldn't install torch." message, or any other error regarding unsuccessful package (library) installation, please downgrade (or upgrade) to the latest version of 3.10 Python and delete current Python and "venv" folder in WebUI's directory. You can download 3.10 Python from here: www.python.org/downloads/release/python-3106/ Alternatively, use a binary release of WebUI: github.com/AUTOMATIC1111/stable-diffusion-webui/releases/tag/v1.0.0-pre Use --skip-python-version-check to suppress this warning. ============================================================================================================================= Python 3.12.3 (tags/v3.12.3:f6650f9, Apr 9 2024, 14:05:25) [MSC v.1938 64 bit (AMD64)] Version: 1.9.3 Commit hash: Installing torch and torchvision Looking in indexes: pypi.org/simple, download.pytorch.org/whl/cu121 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) ERROR: No matching distribution found for torch==2.1.2 Traceback (most recent call last): File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 48, in main() File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 39, in main prepare_environment() File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\modules\launch_utils.py", line 380, in prepare_environment run(f'"{python}" -m {torch_command}', "Installing torch and torchvision", "Couldn't install torch", live=True) File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\modules\launch_utils.py", line 115, in run raise RuntimeError(" ".join(error_bits)) RuntimeError: Couldn't install torch. Command: "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv\Scripts\python.exe" -m pip install torch==2.1.2 torchvision==0.16.2 --extra-index-url download.pytorch.org/whl/cu121 Error code: 1 Press any key to continue . . .
Now said that I don't have Python 3.10.6 installed: 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 . . .
I guess I'm not an absolute beginner! LOL I'm trying to come up with the more appropriate adjectives but somehow my choices are probably not considered acceptable language on RUclips.
I think knowing how to use these tools to get great output is the real magic. But I'm trying to make up for my lack of artist's skills, rather than using it for an ideas system. Very interested in any follow ups you do
@@SurfacedStudio this is the most recent reply i could find, mine seems to download fine but becomes stuck once its 100% downloaded, just ssaying theyres 2.5gb/2.5gb ? eta: 0:00:00
Can you be more specific? “My car doesn’t start” is hard to diagnose via internet. Where is the error, what does it look like, what have you tried, what happened when you tried those solutions, what are you trying next?
I'm not at all impressed by the results shown in the video. Is this worth the hustle? I've used several paid AI image generators that produce truly amazing results and take your feedback to improve. Does Stable Diffusion take any form of feedback from the user to regenerate an image? Would I be able to train a model?
@@SurfacedStudio Lol I can imagine! With SD specifically it seems like every single process has a billion steps to get through but you're right, I guess that's everything
well took me a few hrs of troubleshooting but I finally got it work. the only downside is I am having to use my CPU instead of my GPU and renders take forever. Anyone know how to get it to work with AMD GPUs? I have A RX 6600.
venv "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv\Scripts\Python.exe" Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: 1.9.3 Commit hash: Traceback (most recent call last): File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 48, in main() File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 39, in main prepare_environment() File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\modules\launch_utils.py", line 386, 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 . . .
Interesting! However, with all of the issues I've read below, I think I'll stick with traditional art, like a certain Jack 'King' Kirby. All he needed was a pencil, paper and his cigar. 'nuff said!
I don’t think that’s a thing. But I might be wrong, there seems to be a market for anything these days. It’s pretty easy to install yourself so not sure if people would pay for that
Your explanation that neural networks don't store images isn't correct. They absolutely do! They are essentially compression algorithms. It's like saying a JPEG image isn't the same as the original RAW file. If you give a neural network the right prompt you can get a pretty faithfully recreated copy of the original (depending on the network and the particular image of course).
My generations all come out looking terrible. Just tons of weird stuff. Additional hands, legs, extra heads, weird psychedelic glitchiness all around, I can't get it working right at all to produce anything halfway useful.
Python not found.. press key to exit. So before you waste your time like I did, this doesn't seem to work anymore, the right version of python doesn't seem supported anymore so I can't download it, so all this is for nothing thanks.
I don't think the results were very representative of the capabilities - the video was very much focused on how to get started rather than about 'how to get amazing images' :)
@@SurfacedStudio Thanks, I did a little research into connecting an external GPU rig into my Mac Mini but feel it will just be bottlenecking the GPU. So I'm going to upgrade my PC build instead, I've been spoiled by the glamorous life of low spec Macs. 😂
AI isn’t ‘smart’. It’s good at identifying and applying patterns very very efficiently. It does not ‘understand’ how many fingers humans have because it doesn’t understand what humans are ;)
Omg I'm so excited, a right-wing AI that's not censored and controlled by the disgusting left wing monsters, and it's totally decentralized, I love this, I'm gonna make so much money creating funny AI images haha. These guys that are developing these open-source tools are the real heroes.
delete this video, no lesson learned here, just frustration, i been wanting to install this stable diffusion since 2022, but still no youtuber helped me,
Hello brother. I'm new to AI. I'm curious about why some people are claiming this to be unethical. Is it Stable Diffusion or AI in general? I'm new to this, can you give some insight? I want to know if I'm missing something before getting involved.
You do realize, it's not really ai like in the movies, right ? It's another computer program, like all the other programs that your using, where programmers sat down and hard coded like 20% to 50%, and the rest of the code was completed by libraries of code the program has in its database, developed from humans in the past. Programmers have been doing this for decades, although it's not quite as easy for the coder as these programs are making it for general user... still, it's JUST a program that interprets a user's simplified code words and then references actions it should take based on those words. These "ai" programs aren't exactly intelligent and are more like really decent at decoding adjectives and nouns, and then referencing a database of what to do with those words. So, really, your "moral" opposition to this, is really towards computers in total, and you should go back to the 1940's before there were computers.
@tyrzxv lol lmao imagine being such a dork and defending this shit, yes I know it's not actual Ai like "the movies" I know it was programmed by unethical losers who don't have an imagination and train their loser program on other people's work to generate this shit. Go touch grass.
Why do all of you think everyone is living under a rock because they don't know ai? Maybe people actually have skill and prefer to draw by hand or actually do photography, being able to put words in a prompt literally requires no talent.
There's a difference between 'not knowing AI' and 'not having heard of AI'. I think if you fall into the latter category and you've somehow come to this video then some news might have passed you by. Also, do note that I deliberately never call it AI 'art' anywhere in this video. It generates AI images. Whether it's 'art' is a separate discussion
Great overview of installing and running stable diffusion locally, but judging by the issues in the comments I think it needs to cover what to do if a newer version of Python is already installed. I am an absolute beginner :) but the way I got around the PyTorch version issue was to install Python 3.10.6 to a different directory to my current python installation, selecting the Add to Path checkbox during installation. Then, before running webui-user.bat, I had to edit my user environment variables by selecting the Python 3.10.6 path and clicking the Move Up button so that it appeared before my Python 3.12 path in the list. Then stable diffusion could be installed OK, although I had to update my NVidia drivers before it would generate images. I hope this helps somebody!
I searched the internet for 45 minutes and figured out the issue. Then checked the comments and saw this, which would have saved me the 45 minutes. Ha. I ended up uninstalling 3.12 and installing 3.10.6, which worked for me. Version 3.12 gets the Torch error.
@neilie6593 -- I have a feeling that this addresses the problem that I am having after doing the same (installing the latest build of Python, but the command prompt window saying that Python could not be found).
Alas, I don't understand most of what you said after describing your problem. I downloaded 3.10.6, but I don't know where I should install it, nor how one edits "user environment variables", etc.
Any chance the directions could be broken-down a bit for we mere mortals?😅 Thanks!
worked! just had to move the path up
this tutorial works as of today w/ no issues !! good video tutorial
@@bricology i don't understand either having a lot of trouble
Nice video thank you. But when I run web user sh it opens Xcode, I have mac. Also I tried changing to terminal but nothing happens.
same here
I have tried several times and keep getting this message:
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)
ERROR: No matching distribution found for torch==2.1.2
I get this same error message even if I try it on the 3.12.1 or 3.12.3 version of Python.
@@3sguitar It's telling me "This program is tested with 3.10.6 Python, but you have 3.12.3." I'm going to try uninstalling python and re-downloading the 3.10.6 version
Gotta remove the new version of Python, and pick a 3.10 version. Also remember to delete the venv folder in your stable diffusion ui folder, or else it will continue looking for your Python in a 3.12 folder (or whatever version you previously downloaded). When you double click the bat file to start again, it'll create a new venv folder with the new 3.10 location.
@@MrBrockHeinzyeah just read it on the cmd.. was not obvious to me as all of this is new (I am a noob) on this.. but thank you 👍
@@audreyandremington5265 latest windows installer version of the 3.10.x is 3.10.11 if you wish to update the python
Smart pick using Stable Diffusion and not the commercial/paid tools! 👏🏽👏🏽
Pay no attention to the unsubbers. I'm in the vfx field and aeeing content from a vfx artist trying to integrate ai into their workflow is good to see... because its exactly what the big studios are already doing. So the same short-sighted naysayers here are the same types who were yelling back in the day to shun the Photoshop artists...digital cameras and digital cinematographera...and any artist really who used the computer. Heard these types of complaints thru the years so many times...and they fade away quick because they get left behind. "Adapt or die" in our field has always been true, and still is.
Ai isn't going to take your artist job....an artist that knows how to properly use Ai is going to take your artist job. It's an uncomfortable fact, but it's happening.
Sorry but you are wrong. You can’t compare cameras, photoshop with ai. Does cameras used every painting to make photos? Was photoshop built on others people artwork?
„Adapt or die“ yes artists are adapting thats why ai techfirms getting sued.
And btw your „artist using ai“ will be gone very fast too. If ceos can easy use ai they don’t need an „ai artist“
The thing is though, photoshop and cameras still need talent and need to be used in combination with other skills like lighting, composition, framing, technique, colors, etc. All ai needs is a sentence, its a way different level of effort and very much not the same thing
@@gingerdog8203 still takes some talent to get decent images just writing a sentence does not get the perfect image.
@@AironyAi Unless we get AGI (real Ai) they are good. These LLM's are not very useful for custom stuff and involves more work to make sure they aren't screwing up. VFX crews use these models for very simple, boring tasks that are repetitive in nature and allows that person to do a more productive job.
Never clicked faster on one of your videos
LOL! You definitely were FAST! 😀
@SurfacedStudio dad you keep comming 😊 i hope you're always fine
😊
@@SurfacedStudio
Hi There!
Thanks for the great and easy to follow video.
I have a problem though and I hope you can help, when I run the webui-user (mine does not say .bat) it does say windows batch file after the date and time
So when run it I am getting "could not find a version that satisfies the requirements for torch==2.0.1 ( from versions: 2.2.0, 2.2.0+cu118, 2.2.1, 2.2.1+cu118)
ERROR: No matching distribution found for torch==2.0.1
Then some white text and ERROR code 1 press any key to continue, I press, it closes down... what have done wrong is my computer missing something to be able to run it? I'm not that technical, I only learn what I have to to do the things I want to do and I would love to try stable diffusion.
same
same.
Same issues.
Good thing he told people to leave comments so he could answer questions and then ignore them
Install an older version of Python and you'll be alright
confusing. Web UI gives error that it was designed using python 3.10. So I deleted python 3.12.1 and installed 3.10
But now when running Web UI it says it cant find folder of python 3.12.1...which makes sense as I deleted it and installed 3.10 (which is in the proper Programs folder) but apparently this web UI is not compatible with python 3.10?
argh. any ideas? thanks
same here couldnt launch python
@@chrtsng3722 there are quite a few other videos with slightly different instructions on installing all this so I will try those. If I'm successful I'll let you know.
Do try installing python 3.10 directly to see if that works. Might have to uninstall 3.12 fully to be sure it’s only using the older version
@@SurfacedStudio i did that. Completely removed 3.12.1 and installed 3.10. I double checked that a 3.10 folder had been created in the Programs folder… but command prompt web ui says cant install as python 3.12.1. not found. Thats why I wondered if earlier version of web ui is needed instead but the list of past versions doesnt make it easy to see which is the correct to use. thanks. To confusing for me.
I think I found a fix. I had the same thing happen to me. I whet to C:\Users\(YOUR NAME)\AppData\Local\Programs\Python. then i changed the folder name from Python310 to Python312. I think it is looking for the new vision of Python even though it is incompatible so you have to trick it. If u cant tell i don't know much about Python lol. @@pixelkat1819
Nice, hope youll cover more advanced topics like training and using custom models
Does it need an external GPU to run Stable Diffusion? Can it be used without external GPUs?
I’ve never had an external GPU. You need a decent graphics card but you won’t need anything beyond what you can get in basic laptops and desktops
@@SurfacedStudio As I have only Integrated Intel GPU on my PC. Is it possible to run Stable Diffusion with HotshotXL on the existing system?
Great video thank you but I'm having trouble downloading stable diffusion there is no webUI.bat file. There is one called webui-user and it's a windows batch file. When I try to open it windows 11 says defender smart screen prevented an unrecognised app from starting. Running this app might put your PC at risk. The only box I can take on is don't run or the close the window x button. I've installed python. any ideas .thanks
Hi! I got a very similar screen. It wouldn't show the "run anyway" button unless I clicked on the little v arrow to get more information. Once i clicked and it revealed a few lines of information about the name of the program, the button appeared. I guess Windows is trying to make you think harder before you run stuff?
been goin round in circles with other suppossed tutorials, about to give up until U explain just how & where to put the downloaded files, thanking U
when i down the stable diffusion , it gives me a file called "sd_xl_base_1.0_0.9vae.safetensors" and I cant open it as it is asking for a file application to open it. can someone please help ?
Important! Remember to delete the venv folder in your stable diffusion ui folder, or else it will continue looking for your Python in a 3.12 folder (or whatever version you previously downloaded).
this helped me! It was keep looking for python in C drive, when i had to reinstall 3.10 version in the E Drive
followed your instruction, but the interface im seeing is different.
what do you do if you want better resolution but keep the generated image in the local ui
hey, thanks for this amazing tutorial. I've been trying to follow the steps, but when I press run for webui-user.bat I met two errors: [1]installing torch and torch vision - where torch couldn't be fetched. [2] an error checking the latest version of pip. wondering if you have any experience on those two issues? thank you
Hm can’t say I have unfortunately. The failing to fetch sounds like a networking issue and the pip versioning seems like some local tools not working. What OS are you on?
I had the same issue. I just had to delete python 3.12 and install version 3.11.5 instead. Then I also had to install Git. But after that the script ran through.
It works for me now. I was just able to generate my first smiling cat 😸
Hehe, glad to hear you got it working!
If you need to have multiple versions of python running, you need to run the pip command like: `python -m pip --python \python.exe install pytorch`. Then in webui-user.bat, set the python variable to point to your python.exe
@@peculiarhumanbiped Thanks, it worked for me.
I would love to see how you can input images to help train it to a certain direction, for example as a game developer i'd love to input my previous made textures to get similar results, as my artistic style is quite unique
you use another program in combination with Stable Diffusion called “Kohya” to locally train AI models based on pictures you give it of a specific person or thing. Then you call these models when generating a prompt in Stable Diffusion to generate art with that specific person or thing in it! It’s a bit of a learning curve doing the training but generally you need about a dozen or more pictures to do training! The trained AI files Kohya generates are called “LORA” files. After you create a bunch you can easily browse all your previously trained “LORA” files within Stable Diffusion and click on one to easily load it into your prompt, so you can always create new art of that person or thing contained in that particular LORA file.
The way you explained is perfect for a layman like me. Before coming to your video, I saw many, but none could make a layman understand sd so easily as you did. U deserve a line and su subscribe. Looking forward for more AI stuffs in simple language.
Thank you for the kind words! Very happy to hear you enjoy my way of teaching :)
I'm using this from long time, this is very powerful.
why I can not run the webui.bat?Help
the links doesnt match to your video, can you update the tutorial
When I try to open webui-user.bat file I get the message saying "No module named pip" and "Launch unsuccessful. Exiting."
Does anybody face the same issue? It happens when I use Python 3.11.5 as well as 3.12.3
This was exactly what I was looking for to start, and I bet for others too. Thanks!❤
Glad it was helpful!
I am having an issue with the xl model. Stable Diffusion runs fine with the model that it comes along with (1.5V prunned) but once i try to change to xl i get an error "Connection errored out"
is there something i am perhaps missing?
Does this run on Mac?
If your Mac runs python then yes :)
I successfully entered the 8:50 webpage ui, but when importing v1-5-pruned-emaonly, I got an error Expecting value: line 1 column 1 (char 0). Does anyone know what I should do?
So I followed every step so far. I click on webui, it runs till the end. But it's telling me "Installing torch and torchvision", "Couldn't install torch",
I follow the steps but looks like it's not working well with the last version of Python 3.12.3 because can't install it:
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 . . .
I had this issue. Uninstalled that version of python and then installed python 3.10.6 and it worked.
@@Mirikashh This worked for me as well. You may need to delete 'venv' folder as well.
@@PuneetChhikara where is that folder
Thanks for your excellent video. I was able to get everything installed despite the python version and torch snafu's how after entering a very simple prompt all I ever get is "in queue" and nothing happens after an hour or so. So after an install that took about two hours, I am simply giving up and going back to some free web based AI generators. Thanks again.
thanks for sharing.
Great material, great explanation, thanks!!
Hey, My buddy you are my first teacher, you taught me how to get a witch out of TV😅. Thank you for everything. you did and have been doing.
I have a problem that when I try to run the program, this appears: Couldn't launch python
exit code: 9009
stderr:
Python se nepovedlo najÝt. Spu
Launch unsuccessful. Exiting.
Press any key to continue . . .
does anyone know how to solve it?
uninstall your copy of python and use a lower version. I went with 3.11.10 and it worked. Just delete your webui-master folder and start with a new one
if I want to use ComfyUI, should I install Stable Diffusion XL Model either?
thank you
You can use any stable diffusion model you like with Comfy UI. What you "should" and "should not" isn't something I could tell you :)
the sxdl link doesnt provide download button, i thought this is for beginer video, the site is overestimating the visitors, update your video please,
Thank you Dr. Hans Volter
i fixed my previous problem but now it doesn't load up the generator and i don't know what is wrong
My system had nvidia GeForce gtx 1650 it is taking 10 mins 😢
I am getting error “could not find version that satisfies the requirement torch==2.1.2
You might not have the right version of Python installed. Check with the SD version you downloaded what version of python you need. Then uninstall any you might already have and only install that specific one - that seems to resolve it for most people
Thanx for this tutorial. On Linux it installs like a charm, there is a script that makes it easy, even with a higher version of python it works perfect 👌
Great, concise video. You've earned a like and sub.
Thanks for the like and sub!
This work for Intel processor PC or just for AMD processor PC because can't find Torch file for Intel just AMD and Linux.
Works fine for intel / AMD processors. But you need to make sure your graphics card can handle it
@@SurfacedStudio I'm using: NVIDIA GeForce RTX3060 Laptop GPU
woww thank you so much. really appreciate your help
You're very welcome!
is it possible to make 3d object from 2d picture?
Thnx, very easy and informative guide!
What kind of monster uses light mode?
is it possible on a local machine without GPU?
I keep getting the error:
"File "C:\Users
yan1\OneDrive\Documents\stableDiffusion\stable-diffusion-webui-master\stable-diffusion-webui-master\launch.py", line 48, in
main()
File "C:\Users
yan1\OneDrive\Documents\stableDiffusion\stable-diffusion-webui-master\stable-diffusion-webui-master\launch.py", line 39, in main
prepare_environment()
File "C:\Users
yan1\OneDrive\Documents\stableDiffusion\stable-diffusion-webui-master\stable-diffusion-webui-master\modules\launch_utils.py", line 386, 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 . . ."
whenever i run the webui-user batch file, any ideas why?
‘Torch is not able to use GPU’. What graphics card do you have?
@@SurfacedStudio I got this issue as well - it seems to occur with AMD GPUs
why does mine take so long to generate images even if i use your same prompts?
Are you running the exact same hardware?
Wow. Thanks. I'm also interested in the video game you are working on.
Thanks! More to come :)
I try to installed again but looks like it's not working anymore with the recently version of Python:
Creating venv in directory C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv using python "C:\Users\Kedem Dray\AppData\Local\Programs\Python\Python312\python.exe"
venv "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv\Scripts\Python.exe"
=============================================================================================================================
INCOMPATIBLE PYTHON VERSION
This program is tested with 3.10.6 Python, but you have 3.12.3.
If you encounter an error with "RuntimeError: Couldn't install torch." message,
or any other error regarding unsuccessful package (library) installation,
please downgrade (or upgrade) to the latest version of 3.10 Python
and delete current Python and "venv" folder in WebUI's directory.
You can download 3.10 Python from here: www.python.org/downloads/release/python-3106/
Alternatively, use a binary release of WebUI: github.com/AUTOMATIC1111/stable-diffusion-webui/releases/tag/v1.0.0-pre
Use --skip-python-version-check to suppress this warning.
=============================================================================================================================
Python 3.12.3 (tags/v3.12.3:f6650f9, Apr 9 2024, 14:05:25) [MSC v.1938 64 bit (AMD64)]
Version: 1.9.3
Commit hash:
Installing torch and torchvision
Looking in indexes: pypi.org/simple, download.pytorch.org/whl/cu121
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)
ERROR: No matching distribution found for torch==2.1.2
Traceback (most recent call last):
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 48, in
main()
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 39, in main
prepare_environment()
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\modules\launch_utils.py", line 380, in prepare_environment
run(f'"{python}" -m {torch_command}', "Installing torch and torchvision", "Couldn't install torch", live=True)
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\modules\launch_utils.py", line 115, in run
raise RuntimeError("
".join(error_bits))
RuntimeError: Couldn't install torch.
Command: "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv\Scripts\python.exe" -m pip install torch==2.1.2 torchvision==0.16.2 --extra-index-url download.pytorch.org/whl/cu121
Error code: 1
Press any key to continue . . .
Now said that I don't have Python 3.10.6 installed:
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 . . .
I get the same error , don't know how to fix it EDIT: I just followed the github's install process instead and it all works now
I guess I'm not an absolute beginner! LOL I'm trying to come up with the more appropriate adjectives but somehow my choices are probably not considered acceptable language on RUclips.
I think knowing how to use these tools to get great output is the real magic. But I'm trying to make up for my lack of artist's skills, rather than using it for an ideas system. Very interested in any follow ups you do
Old 'stable-diffusion-webui-master' not working, Use New Version 'sd.webui' according to your Graphics Card
Have they introduced new scripts? I haven’t used SD in a while :)
this doesn't work cant get passed the torch error
wont launch the ui, says couldnt clone assets
What says couldn’t clone assets and where?
@@SurfacedStudio this is the most recent reply i could find, mine seems to download fine but becomes stuck once its 100% downloaded, just ssaying theyres 2.5gb/2.5gb ? eta: 0:00:00
Sounds like an issue with your internet or the connection?
Generating images guide 8:53
Thank you!!!
Welcome!
Is 3 mins alot to generate 1 image?
Depends on the image. And what you qualify as ‘a lot’. And your computer.
useful video
It says couldn’t launch python
i followed steps but now its saying python was not found
ok never mind i fixed that problem now i have "no module named pip"
Nice
Thanks
this Ai really detail and beautiful
Thank you. You made this less intimidating.
it gives the torch error every solution I searched didn't help
Can you be more specific? “My car doesn’t start” is hard to diagnose via internet. Where is the error, what does it look like, what have you tried, what happened when you tried those solutions, what are you trying next?
do i need a graphics card?
An NVIDIA GPU is recommended, not sure you can run it without a graphics card altogether
I'm not at all impressed by the results shown in the video. Is this worth the hustle? I've used several paid AI image generators that produce truly amazing results and take your feedback to improve. Does Stable Diffusion take any form of feedback from the user to regenerate an image? Would I be able to train a model?
Shows the need for RUclips to have a x3 speed
At x3 you might almost cut my average video length down to 20m 😉
@@SurfacedStudio Lol I can imagine! With SD specifically it seems like every single process has a billion steps to get through but you're right, I guess that's everything
well took me a few hrs of troubleshooting but I finally got it work. the only downside is I am having to use my CPU instead of my GPU and renders take forever. Anyone know how to get it to work with AMD GPUs? I have A RX 6600.
did you find a way to make it work? I have the very same GPU
This do not work with intel processor just AMD processor.
venv "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\venv\Scripts\Python.exe"
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Version: 1.9.3
Commit hash:
Traceback (most recent call last):
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 48, in
main()
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\launch.py", line 39, in main
prepare_environment()
File "C:\Users\Kedem Dray\Desktop\New Stable Diffusion\stable-diffusion-webui-master\modules\launch_utils.py", line 386, 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 . . .
I'm running this on an Intel processor. Never used an AMD one so I'm not sure what your problem is?
Sounds more like your graphics card isn't strong enough? What GPU do you have?
@@SurfacedStudio I'm using: NVIDIA GeForce RTX3060 Laptop GPU
Please make a complete tutorial video on ComfyUI without dedicated GPU
Everything already change on the links. Its like online Linux Tutorials.
Interesting! However, with all of the issues I've read below, I think I'll stick with traditional art, like a certain Jack 'King' Kirby. All he needed was a pencil, paper and his cigar. 'nuff said!
doesn't work. just get missing torch 2.1.2
Double check you have the exact version of Python installed that is required by the version of SD you downloaded, that’s usually the issue
Does anyone sell new PC's with Stable Diffusion installed...?
I don’t think that’s a thing. But I might be wrong, there seems to be a market for anything these days. It’s pretty easy to install yourself so not sure if people would pay for that
@@SurfacedStudio Thanks. May be easy for some but I see comments of many people having difficulties with installing.
doesent work
Cause you've done something wrong
Your explanation that neural networks don't store images isn't correct. They absolutely do! They are essentially compression algorithms. It's like saying a JPEG image isn't the same as the original RAW file. If you give a neural network the right prompt you can get a pretty faithfully recreated copy of the original (depending on the network and the particular image of course).
Can anyone recommend a much dumber tutorial?
Did ya ever find one, i need it T-T
The older models where much better
My generations all come out looking terrible. Just tons of weird stuff. Additional hands, legs, extra heads, weird psychedelic glitchiness all around, I can't get it working right at all to produce anything halfway useful.
Python not found.. press key to exit.
So before you waste your time like I did, this doesn't seem to work anymore, the right version of python doesn't seem supported anymore so I can't download it, so all this is for nothing thanks.
Hey! Thank you for saving our time
Did you found any other way?
stop blabbing and just TELL US!!!!!!!!!
Install Stable Diffusion. Enter prompt. Press Generate. Profit!
oh...the cat images were kinda disappointing ngl...
I don't think the results were very representative of the capabilities - the video was very much focused on how to get started rather than about 'how to get amazing images' :)
Doesn't work. the newest python isn't compatible, path isn't correct. waste of time. find a better tutoral than this.
So basically anyone with a Mac laptop won't be able to use it.
Why not? Mac can run python
@@SurfacedStudio Cool, just thought it would require a decent graphics card.
Mine is a lowly hardwired Intel UHD Graphics 630 1536 MB.
You do need a decent graphics card as it runs on your GPU. That’s separate from whether you have a Mac or a windows or a Linux machine though :)
@@SurfacedStudio Thanks, I did a little research into connecting an external GPU rig into my Mac Mini but feel it will just be bottlenecking the GPU. So I'm going to upgrade my PC build instead, I've been spoiled by the glamorous life of low spec Macs. 😂
I’m new to Ai. What is program is suitable to make a rock music video with yourself part of it? I haven’t seen many
lol image if god did a copyright claim on humans.
So all of that just for some sub-par art. I guess i'll stick with BingAI. Great video though!
iT DOESN'T SEEM TO WORK ON WINDOWS 11 DOES IT...You couldn't make it any more difficult could you? De Quoi est-ce que tu parles? HUH?
Working fine on Windows 11 - it's the only OS I've been running it on so far
If Ai is so smart, why doesn't it know how many fingers a human being has?
AI isn’t ‘smart’. It’s good at identifying and applying patterns very very efficiently. It does not ‘understand’ how many fingers humans have because it doesn’t understand what humans are ;)
oh sweet, I though sdxl is paid
As free as the SD base model :)
very disappointed
Omg I'm so excited, a right-wing AI that's not censored and controlled by the disgusting left wing monsters, and it's totally decentralized, I love this, I'm gonna make so much money creating funny AI images haha. These guys that are developing these open-source tools are the real heroes.
delete this video, no lesson learned here, just frustration, i been wanting to install this stable diffusion since 2022, but still no youtuber helped me,
Unsubbing.
Hello brother. I'm new to AI. I'm curious about why some people are claiming this to be unethical. Is it Stable Diffusion or AI in general? I'm new to this, can you give some insight? I want to know if I'm missing something before getting involved.
Ugh AI? instant unfollow. get some morals.
Right
why did you submit your message online? get some morals. Mailmen also have to feed their families. Hypocrite
Why is this unethical? Do you mean this program? or AI in general?
You do realize, it's not really ai like in the movies, right ? It's another computer program, like all the other programs that your using, where programmers sat down and hard coded like 20% to 50%, and the rest of the code was completed by libraries of code the program has in its database, developed from humans in the past. Programmers have been doing this for decades, although it's not quite as easy for the coder as these programs are making it for general user... still, it's JUST a program that interprets a user's simplified code words and then references actions it should take based on those words.
These "ai" programs aren't exactly intelligent and are more like really decent at decoding adjectives and nouns, and then referencing a database of what to do with those words.
So, really, your "moral" opposition to this, is really towards computers in total, and you should go back to the 1940's before there were computers.
@tyrzxv lol lmao imagine being such a dork and defending this shit, yes I know it's not actual Ai like "the movies" I know it was programmed by unethical losers who don't have an imagination and train their loser program on other people's work to generate this shit.
Go touch grass.
Why do all of you think everyone is living under a rock because they don't know ai? Maybe people actually have skill and prefer to draw by hand or actually do photography, being able to put words in a prompt literally requires no talent.
There's a difference between 'not knowing AI' and 'not having heard of AI'. I think if you fall into the latter category and you've somehow come to this video then some news might have passed you by.
Also, do note that I deliberately never call it AI 'art' anywhere in this video. It generates AI images. Whether it's 'art' is a separate discussion
The words in the prompt is only the last little bit of making ai images if that was all I took every ai programmer wouldn't have a job
Also writing promt for ai takes a baseline understanding of the keywords the ai can understand kinda like people who edit photos in photoshop
@@kyleschultz3130 go be mad somewhere else friend, this isn't a constructive use of your time
Too much chit-chat - 4mins in and 'Install java' is all I've got!
Not sure what you’re watching but none of this video includes installing Java
Unfollow
cant trust anyone now a days. off to another video, we all the same problem. nice troll idk by who tho