How to Install AUTOMATIC1111 + SDXL1.0 - Easy and Fast!

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
in this video I'm going to show you how to download and install automatic 1111 web UI with sdxl so you can start making incredible air ready let's get started so there are a few ways that you can install automatic 1111 but we're going to stick with the recommended methods listed on their GitHub page to avoid any sort of confusion okay so the first thing we want to do is download and install python 3.10.6 this is the version required for automatic 1111 and anything else probably won't work very well if you have any other versions of python installed it is recommended to uninstall them first to install python use the link in the description to head over to python.org then scroll down and click Windows 64-bit installer to kick off the download once it's ready launch the EXE the absolute most important thing is to take this box down here which says add python 3.10 to path I cannot stress this enough if you skip this box nothing will work and you'll need to start from scratch so make sure that you click that box unless you know what you're doing keep all of the settings at default and finish the installation okay so next we need get get is a software versioning system and it's going to help us download and update automatic 1111. you can use the link in the description to download the Standalone installer of get for Windows and just use all of the default settings okay once that's done we're ready to install automatic 1111. first create a folder and call it a11 you're probably going to need a decent chunk of disk space when you start downloading models so choose where you put the folder wisely next head over to the a11 GitHub page using the link in the description look for the green code button near the top of the page click it and then copy the URL under where you see https you can just click the button beside the URL to copy it next you want to go back to where you created the automatic 1111 folder click into it and then click into the address bar from here type in CMD hit enter and you will see a command prompt box appear when this appears just type in git clone and then paste in the URL you just copied and by doing that you're copying the a11 software from the GitHub repository onto your computer you can also use git to update automatic 1111 every time they release a new version you can do that by going into the stabledevusion web.web UI folder typing in CMD in the address bar above there and then typing in get pull and that'll just pull the newest version we just installed it so it's up to date but that will install the new version if it's available you can also use get to restore back to a previous version or a previous build of automatic 1111 if you need to you can do that by typing git checkout and then the key of the commit that you want to revert to and hit enter I'm not going to do it here we've just downloaded to it but you can get all the commits from the GitHub webpage and I'll leave a link directly to that in the description okay so we're almost there next to generate sdxl images we're going to need the sdxl model head over to hugging face using the links in the description and download sdxl base 1.0.save tensors you can just click on the download file icon beside that and you'll also want to pick up the sdxl refiner which is another six gig download you can find links both of these into the description that's all we need for now for this tutorial but if you're interested you can also download the Laura which is a 49 Meg file once they are downloaded head back to your automatic 1111 folder then stable diffusion web UI then models and then stable diffusion and copy those two big files in there you'll probably want to use SD 1.5 models at some stage so I recommend making an sdxl folder and saving the files in there for now to keep things organized you can add any model this way and you can find more sdxl based models over at civetai.com this site can be super not safe for work so change your settings according to your preferences okay so there's only one last thing we need to do before we launch automatic 11 11. head back to the stable diffusion web UI folder and look for a Windows batch file called Web UI Dash user right click on that and choose edit this will open up notepad and you'll see this we'll only be editing the command line ARG section so be careful not to make any other inputs click the side command line args and type in dash dash Auto launch and no half vae I won't get into vaes here just know this setting helps to reduce distorted images and black outputs next if you have an Nvidia card you can enter dash dash xformers for better performance and if you have an Nvidia card that says 1600 series or better you can also type in dash dash opt channels last which should give you another speed boost if you have a graphics card with 8 gigs or less of vram type in dash dash mapv wrap once you entered all of the settings hit save and you're finally ready to launch double-click the web UI user batch file that we just edited another command box will pop up and start loading automatic 1111. it'll take a few minutes the first time you launch this so sit tight once it's finished loading the web UI should open up at a tab in your browser when it launches there's one last thing we need to do to get the most out of sdxl click on the extensions tab then available then click load from to get a list of all available extensions there is plenty of time later to check these all out but for now we just need the refiner extension so type refiner in the search and hit enter once you've found it hit install and now click the install tab click check for updates and then apply and restart which will install the extension and restart the web UI Okay so we've made it we have automatic 1111 with sdxl installed so let's do a quick rundown of the most important settings and create our first image so starting from the top the first thing we see here is our prompt box and this is where we're going to put the prompts the text prompt in that we want our image to be so I'm going to put in this prompt a cyberpunk girl in a futuristic city at Sunset so a fairly simple prompt but a good one to get us started below that you will see the negative prompt box and you can put words in here that you will want stable diffusion to de-emphasize in your generation so a good example would be NSFW if you want to keep your Generations safe from work now sdxl is meant to not take negative prompts as much into consideration so you can play around with this but your mileage may vary below that we can see sampling methods under here you'll find a bunch of different options sampling is a pretty complicated topic but just know that each of these options will produce a different result and some are better than others and some are better than others at different scenarios some of the more popular ones are Euler a which is a good all-rounder or DPM plus plus 2m Karis which is a good one particularly for photorealism next there's sampling steps so this is the number of iterations that stable diffusion will run to get the image based on your prompt note that bigger is not always better here and working in a range between 20 and 30 is probably the Sweet Spot next if you're getting weird faces in your Generations you can take the box for ReStore faces which will focus on getting more realistic proportions and then under that you'll see the width and height settings now it's important to note that for SD 1.5 the standard was 512 by 512 but for sdxl this is 1024 by 1024 or variations on those sizes the reason for this is that's the resolution that sdxl was trained on using anything smaller is just going to produce garbage there are a variety of sizes you can use to achieve different aspect ratios for this resolution and I'll leave a link in the description for a list of sizes you can try below beside that you're going to see batch count and batch size so batch size is how many images your graphics card will generate at the same time so to say vram I always stick this on one and then batch count is how many times table diffusion will repeat a generation with a consecutive seed to give you some different options I usually stick this on for the CFG scale controls how much stable diffusion listens to your prompt for the final image the lower this number is the more creative it gets for a setting of one stable diffusion probably won't listen to your prompt at all where a higher number will be super strict so I keep this between five and seven the seed is the number used to create noise that stable diffusion will then generate your image from every single number will create different noise patterns and so the image created will change based on this number and finally there's the refiner extension we just installed this is going to use the refiner model we downloaded earlier so go ahead and select that from the drop down menu this extension will use the refiner model on the last number of steps in the generation that we set here so if we set this to five it's going to use the refiner model for the last five steps of the generation the refiner model is meant to give you more detailed images but your mileage may vary so go ahead and play around with this okay we can finally make an image so now we can just hit generate and if a few moments we'll have a masterpiece now I've sped this generation up because sdxl is pretty resource intensive and it was pretty slow while screen recording but I'm usually getting about 30 second Generations on an 8 gig graphics card if you're finding sdxl is a bit slow check out this video for some tips to speed things up let me know in the comments how you've been finding sdxl I'd love to know and if there were any installation tips that I missed if you found this useful hit the like button and feel free to subscribe for more content like this thanks for watching all the way to the end catch you later
Info
Channel: Incite AI
Views: 12,516
Rating: undefined out of 5
Keywords: stable diffusion, SDXL, SDXL1.0, MIDJOURNEY, automatic1111, a1111, stable diffusion install, AI art, sdxl easy, stable diffusion tutorial, stable diffusion ai, stable diffusion xl, how to install stable diffusion, install stable diffusion, sdxl 1.0, sdxl automatic1111, automatic 1111, sdxl 0.9, how to install automaic 1111, install a1111, a111, a11111, stable diffusion easy install
Id: gePzepJBqD4
Channel Id: undefined
Length: 10min 23sec (623 seconds)
Published: Sun Aug 06 2023
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.