Automatic1111 - Installation and Set-up - Getting Started Explainer and Guide - Part 1/4

what's up everyone Michael here with a stable diffusion video today we're going to be talking about automatic 1111 web UI and how it interacts with stable diffusion and how to use it this is going to act as a general guide in discussion video covering pretty much as much as possible it's going to be fairly basic in terms of all of what we cover but we're going to be covering a lot so this video is going to be long and there's going to be a lot of information so if you don't end up catching up on everything don't worry about it too much you will be able to cover that more in depth in the future and just kind of take things one step at a time but this video is going to be all-encompassing or at least try to be or we're going to try and cover as much as possible so that you understand exactly what stable diffusion is how it works with the web UI and how to generate images so first we're going to start off with installation just in case people don't already have it installed and we're going to talk about some caveats with that and then we're going to end up talking about some recommended settings and some stuff about deterministic results that I would like to cover just so you're already set up then we're going to talk about acquiring uh different models such as checkpoints Laura's vaes and the different types of file types and this will get you done with the basics and so that'll be enough enough for you to understand the type of resources that you'll need to actually run stable diffusion and end up getting started and then we'll talk about the web UI so we're going to cover all of the you see here and don't worry if you already have automatic installed and you yours doesn't look like this don't worry about it by the end of this video you'll know how to get set up just like this one so yeah let's go ahead and get straight into it so first off the one thing that I do want to mention is that stable diffusion has a lot of complex elements and in this video we're not going to be able to go super in depth and if you would like to see that type of stuff do make sure to leave a comment below or check us out over on Discord and make sure to like ask questions and stuff and then I'll know what to put in videos and stuff because there's a lot here so if there's anything in this video you're ever confused about you can always ask the community and we'll be able to help you out and there's a bunch of different AI communities out there that have a lot of really helpful people so just wanted to mention that real quick just in case you're watching this video and you get confused by something so let's begin with installation this is where a lot of people end up getting a little bit hung up when they're not very tech savvy for the most part especially if you've never used things like Python and such but there's no actual programming required in setting up automatics web UI or in using stable diffusion in this way so first off what we're going to do is we're going to follow the best installation guide which is actually available over on the automatic GitHub page over here you'll actually be able to see all of the source code and all that such things such and things I'm already watching my words I apologize for that um but you'll be able to actually see all the source codes and stuff like that so if you ever want to dig into it you possibly can um but they have the best installation guide because it's actually really really simple and setting up the UI is pretty easy so we're just going to scroll down on the page here until we see installation and running now they have some more in-depth guides here at the top where it talks about Nvidia AMD Intel CPUs and gpus and then down at the bottom down here it should say apple silicon somewhere there it is installation on Apple silicon so there are some differences between different gpus because stable diffusion uses gpus to end up rendering the image and we won't get too much in depth into that in this video but it is important to note because there are different dependencies depending on which GPU you have and how many resources you have we will talk about that a little bit more when we get into the settings aspect but for setting up I'm going to be talking about setting up a Nvidia install on Windows 10 or Windows 11. however the AMD isn't installed is exactly the same so if you follow this AMD link here it's the same exact steps however you're going to want to use this console command instead of the one we'll use for the Nvidia one and I'll mention that when we get to it but the installation guide is actually quite simple they do have an automatic setup guide here if you wanted to follow that one that'll download all the dependencies but we're going to cover the automatic install on windows so first things first you're going to want to install python 3.10.6 for this version we are using the 1.6 version of the web UI however this dependency may change in the future so do make sure to check the GitHub page but should be this link right here on the GitHub page should always link to the correct one so just follow this link and it'll have an installer python is just the programming language so this is an API slash um redistributable package that allows programs to use python on your computer the only thing to note about here is that although you can go through the entire installation process and just use defaults throughout especially if you don't understand anything just use the default and it'll be fine but make sure that when it says add python to path make sure that box is checked so that it adds python to path which basically allows anything on your system to call python within its own environment which allows you to actually use automatic and put it anywhere on your computer and run it from any folder which is really helpful and it also avoids a lot of the technical details that you would have to have otherwise so it's really good to have then you're going to need to install git which is a program that allows you to pull data from different repos such as GitHub as this one and so automatic uses it to pull updates pull information pull extensions and so on and so forth now after you do the initial setup the web UI can be run completely offline so you only technically need git at the beginning but there's no reason to uninstall it it's a really useful program and you'll be using it to auto update stable diffusion in the future gather extensions and update stuff so it's really important to have but step three is where we're actually going to get a little bit more in depth and actually cover some stuff in more of a guide type sense it says download the stable diffusion web UI repository for example by running the git clone and then the link to the repository.git and this part confuses a few people basically what it's telling you to do is go to the folder where you want it to install open a command window enter in this command and then you know install it through that way so basically it's it's pretty simple but we're gonna go ahead and cover it here because not everyone's used the command line and there's some cool tricks that I can show you real quick so if we go over to my automatic install you can see these are the files that we're going to be installing using this git pull and everything is open source as well it's the same exact files that you would end up seeing at the top of this page here so it's all these files as well so we're just basically going to be downloading all those files if you're curious but navigate to wherever on your computer you want to install the web UI this can be anywhere but try and make sure it's somewhere that doesn't require admin privileges because on some setups depending on the way you have Administration right set up it can cause some issues so I recommend like the documents folder or a separate hard drive or something along those lines if you have one available and so anyway so we got an SS or SATA SSD here that I have that I use for stable diffusion and we're going to go ahead and add a new folder call it something descriptive so you know what it is so I'm going to call it the auto web UI and we'll just hit enter there and now that we have our folder and we want to install it inside this folder we're just going to go to the folder right here and we're going to hold the left shift and right click if you're on Windows 10 and go to open Powershell window here or if you're on an older version of Windows 10 it should be open CMD window here and that will actually open up a little window but on Windows 11 you can actually just left click it and it may actually say just open in terminal so what this is going to do is it's going to open either a command window or it's going to open Powershell and it's going to automatically put us where we need to be inside this folder so we don't have to manually navigate to it which is super useful if you don't actually know how to navigate through Powershell so now that we have that grabbed we're going to go back to our web page here and we probably could have copied this beforehand but we're going to copy this command they told us to get so the get clone and then the web address we're going to go ahead and copy that make sure not to grab anything extra like Extra Spaces before and after or the period at the end we just want the link along with and clone at the start so we're going to do that then we're going to go over to Powershell on Windows 11 you can just right click or or on Windows 10 or Windows 11 you can hold control and then press V and it'll paste it in there and then we'll go ahead and hit enter and now what it's going to do is it's going to go ahead and download the resources and it's going to put everything that was on the GitHub straight into our folder there so we can go ahead and close this we don't need it anymore go back over to our web install we're going to go to auto web UI like we called it before it's going to have a stable diffusion web UI folder here we can double click on that and then we're inside the folder now if we look back at the instructions the last instruction it has is run webuser.bat from the Windows Explorer as usual not Administration user basically all that's saying is go to the web user dot back here and you'll just double click it and it'll start running and it'll start updating automatically and downloading all the requirements now it's going to take a little bit and it's going to seem kind of confusing but it will just do it very very quickly and you'll be able to see the progress within the command window but give it a good 15 to 20 minutes when you think it's stuck just in case your download speed might not be up to Snuff it'll have to download a lot of files and such and do a lot of driver type nonsense with whatever graphics card you're using now like I said with AMD D you do have to use a different command so how we did the get clone here I probably should have mentioned this beforehand um but you'll want to actually use the one for AMD which is git clone and then you want to use the direct ml link that is right here and then they have another additional thing here which you probably don't need howdy there it's editor Michael again so I botched the AMD mention in this video and so I'm doing a bit of a quick post edit thing I'm just going to throw up some screenshots where basically when you install an AMD you actually just want to open the command terminal or Powershell from say um the windows bar or start menu or whatever just use the search bar and type in Powershell and it'll pop right up go ahead and open that and then run the entirety of the AMD command here and so basically you want to run that it'll go into the program files inside of your system and I don't know exactly where it installs because I don't have an AMD system and so I'm not installing and such but it should go to the program files and then you can just open it and use it from there or you can just move the folder wherever you want now the reason you have to to do it that way is because this command here actually runs some updates for the sub modules and that needs to run in order for it to grab the appropriate drivers and set things up for the AMD Branch so if you're using an AMD GPU you do need to do it that way for the current guy that they have now in the feature that could change so make sure to check their AMD guide I unfortunately don't have an AMD system and so I can't speak to how it works and stuff like that so if you struggle with that a little bit make sure to ask like some AI communities and stuff like that for help because otherwise I unfortunately can't really cover that so hopefully that's not too confusing hopefully that makes sense uh anyways back to the video just ignore everything I say about AMD here I apologize stable diffusion web UI direct ml this is a separate branch of stable diffusion's web UI for automatic and it supports amd's direct ml instead of the Cuda cores that Nvidia uses if you are using another type of graphics card I unfortunately don't know a whole lot about the Intel CPU and GPU setups or the Apple silicon so you'll have to follow their guides on here but they are quite extensive if you go ahead and click on these links here you can see It'll guide you through everything that you need to do for Mac OS specifically there are some Rights Management type stuff you'll need to do so keep that in mind but if you're on Windows super simple same thing on Linux super simple and basically now that we have this window or this sorry not window now that we have this folder we can just move it wherever we want on Windows and run web UI from that folder it runs inside its own contained environment so as long as this folder stays intact you can even rename it or whatever as long as all the files inside here remain intact it'll run perfectly fine now I already have a stable diffusion setup and it takes quite a while to do the initial up like I was saying like 15 to 20 minutes and we ain't got time for that so instead of going through and actually doing the whole setup process on a clean installation I'm going to go ahead and close my automatic that's open right now and we're going to reset the config and then we're going to go through a couple of options that I like to change in order to set it up in the way that I would like to do so so let's go ahead and find the config file in here so if you want to ever manually edit things you can all these files are openable so for example on the webuser.bat if you right click it and you can go ahead and oh sorry go to more information and you should be able to click edit and that'll open it up into a notepad file so as we see here we have our notepad file this allows us to put in command lines redirects for python git or VNV directory which is like a cache folder temporary directories so on and so forth now one line that I'm going to recommend putting in here is called the auto launch so I believe it's Auto launch let me actually double check that because I have my customized one so we'll go ahead and open up my customized one as well you can also copy this folder and run it from within this folder wherever you want so you don't have to run it exactly like that oh it has no uh Dash in the middle there we go so Auto launch is really useful um what Auto launch will do is when you click on the webuser.bat it'll automatically open it up so after you've done that initial setup of opening the webuser.bat you'll just close the command window after it's done setting it up and then enter in this command hit Ctrl and S to save the file and then when you double click the web user.bat it'll automatically open your web browser with the web UI which is super nice to have another uh command I like to have is called update check as you can see down in my bottom one and update check basically just does a quick update check some people like to do auto update which also is quite useful basically it updates if there's another update to the main branch but I don't like doing that because sometimes updates to the main branch will end up causing issues or have bugs and So to avoid that I just do an update check that way I can know when there there is an update but I don't have to force myself to update and the last two commands we're going to cover here is going to be low vram and medium vram now I even though I have a good graphics card with a lot of vram I do use mid vram on sdxl now this is a specific command when you're using an sdxl model which we'll talk about a little bit later it makes it so it uses the mid vram command which basically is a memory saving tool that makes Generations take longer but saves you a lot of memory now I like to generate large images and so this is helpful for sdxl however if you're on a lower end graphics card with less vram or even like a mid-range graphics card or a high-end graphics card with less vram say less than eight gigabytes of vram then even running on SD 1.5 models which is uses a lot less memory than sdxl can actually run out of memory and then you can't generate images so to actually fix that if you're on a lower end graphics card with less memory you can either do low vram which will slow down things a lot or you can do mid vram and try that out um I recommend only doing mid vram if you possibly can um try it out without these two commands first but you can always come back in here and add these commands later there's also optimization options within the web UI but it's easier to do it in this command line argument so there you go but even on my RTX 4090 this is the commands that I normally use last one that you see here is just a redirect to an extra folder that can load checkpoints I use for like merging uh models and such like that so we won't cover that in this video but there you go and then again once you said like once you're done go ahead and hold Ctrl and S or go up to file and Save and then this will save it so that when you launch the web UI it will be launching and the next thing we're going to look at is the config.json file I am just going to take mine I'm going to make a copy of it and I am going to make a DOT back because I already have a backup of my older file so there we go and we're just going to delete that file and then we're going to double click on the webuser.bat and we're going to go ahead and open it now you don't have to do the config thing just going to open up a window here we'll be looking at that model later for use there we go so I reset my config so it should look exactly like uh yours will when you end up first launching it so here we go we got the blue UI there we go so now I'm back to what it normally looks like I'll just do a quick comparison on what we're going to set up and what it initially looks like which there's only a couple of changes like at the top bar we're going to add those quick list things we're going to change the color of it and the theme to something that we like and then we're also going to change the defaults which for some reason it saved the default so I guess the defaults are saved in another file I was unaware of so that I didn't know I apologize for that but generally speaking these um The Prompt spaces will be empty and set up differently if I go ahead and click these back to where they would be by default real quick just to try and make this thing easy for everyone to understand so typically speaking I believe your UI should look like this except for the model up in the top left you won't have the abyss linked in there because you won't have any models there but when stable diffusion launches it will actually load up a default model the One release by stable diffusion or a stable AI sorry and it will be the base SD 1.5 model it's okay you can use it it's good for like getting used to things and like trying to test it out but its quality isn't very good so the first thing we're going to cover is we're going to go over some settings within the web UI to try and set us up for deterministic results get some convenience type stuff set up and then I'm going to relaunch my web UI go back into my defaults just because I have some custom stuff set up and then we're going to go and acquire models talk about installing them and then we're going to go over the web UI and how to actually use it as well as some extensions so let's go ahead and start off with our settings so let's go ahead and so go over to the settings tab here apologies if I'm going too quick by the way you can always rewind or slow down the video or something like that and the first thing we're going to cover here is actually the saving uh images and grids option here so you've got a couple of buttons here don't get too worried about there being too much text here um and we're going to go ahead and slim down our windows so we can see all the settings in one little convenient spot here so you don't have to look across the screen so up at the top we got apply settings and reload UI so anytime we change any settings we do have to click apply settings every time I do a settings reset or any update I always go to apply settings up at the top and you'll see it clicked a bunch of stuff and automatically set that up there what that does is it automatically puts everything that has not been configurated already to defaults within the config file and this means that it will keep those settings even between updates that way if you update the web UI you don't get surprised by some setting that randomly changed and your web UI is suddenly broken in some way that you think like you don't understand or whatever so that's a nice little trick to do so after every single update I always do that just to make sure it sets them defaults so it does doesn't randomly change between updates which is quite nice I'm gonna go ahead and click onto another Tab and then click back oh it just stayed there oh man that's going to be annoying let's go ahead and click reload UI so that it'll reset the UI and reset our settings menu because having that large bit of text there is going to be a little bit annoying that must be a bug in the current version because normally it didn't do that I apologize for that all right so let's go ahead and start off with what we're going to end up doing first um most of these settings I'm going to recommend keeping on default but over here in the saving images we want to enable the saving before the face restoration and settings such as that so if we scroll down here we say save a copy of image before doing face restoration before applying high-res fix you can enable this if you want I typically don't but I do when I'm testing so you can play around with this but basically what these settings do like save copy image before applying image correction um what these three settings will do is it'll create a copy of the image that you're generating before it does these extra processes so face restoration high-res fix and color correction are post processes that will run after the image is already generated so doing these settings will make it so that you already have a reference because sometimes when you do face restoration or high-res fix or something like that it might destroy the image or change it in a way that you're not certain about so having these enabled will allow you to look back at how that generation went and be able to customize it later this will make more sense once we actually get into using the web UI next we're going to go down to the lossless compression so over here we have use losses compression for webp images I recommend enabling this one and I also recommend disabling the save a copy of a large images as JPEG just because this takes up a lot of extra space because it'll generate extra images generally speaking when you go through these settings I try not to have extra stuff saved I also end up disabling the grid saving system as well but we won't cover that in this video because it's not not it's not necessarily something that most people end up doing and that should cover it for this page another thing that I would like to do is the directory for temporary images actually down at the bottom if you're running stable diffusion off an SSD like I do setting up a temporary directory I'll install on like a hard drive or something along those lines is typically a good idea so you don't end up burning out your SSD by like writing a bunch of image files to it all the time so I use an extra hard drive so if we go over to my browser here I have a hard drive e and if I go to here I have one called AI tooling and I have an output folder and so I'll end up saving things to here and so I have a temp folder here that will save all of the temporary files and automatic web UI and stable diffusion in general creates a lot of temporary files so this can help save your ssd's lifespan quite a bit so that is quite important you just copy the path there so for example if I go back to my e Drive I go to AI tooling and I go to my output I can just go to this top bar here right click copy address as text go back in here Ctrl V paste it in there or right click and paste and it'll end up going there and then all the temporary files will go into there now technically I would do the temporary folder but you you get the point you just copy the folder where you want to go and then it'll go in there I recommend saving the temporary folder as in a separate folder than the image generation folder which we'll talk about right now actually so if we go to the left side on the next PATH it says pass for saving so when you click on here it says directory for images this is again if you're running on a SSD you can end up having it go to a hard drive instead so if we go back to my output folder here I have one for grids I have one for images I have one for logs and I have one for temporary folders and you can just copy those so if you go into images for example here's that one image that I had here and so so anytime I generate an image the final image will be pasted into this folder so yeah it's quite nice to have it so you can set it wherever you want and I recommend if you're on an SSD like I said set it on a hard drive if you have one if you don't just kind of deal with it I guess and try to limit how many images you're generating in unnecessary quantities such as disabling grids and such anyways so there you go you could set those individually now I've noticed some extensions will actually just use the output directory for like the specific ones so if you do create one default directory and you want it to be for all of them do copy that into each of these so for example if I wanted to do it for images I would copy it into all of these and then that would cover all of the images to input into the same folder otherwise some extensions will use the wrong folder and that can be kind of confusing when you're first using extensions I know for example the super merger extension that I use saves it into the default directory for text to image for whatever reason and it ignores the defaults which is annoying so if you have this empty it'll save it in the outputs folder in side stable diffusion just to quickly cover that as well if we go back to my web UI in automatic if you go to your automatic install here the output will actually by default go to a folder called output outputs right here and if you click on this you'll see all these folders pre-made and then if you go to images they will paste into here from your text to image or you know text to Grid or whatever you end up doing so just to cover that real quick anyways let's actually get back to the rest of the settings we want to cover next we want to go to the API this is more of a sanity check so there's a lot of dubious guides out there that tell you to enable a bunch of different settings or have like setup scripts in here make sure that forbid URLs to local resources is enabled so basically what this does is it means that say a URL that is put into say a malicious checkpoint or malicious code that somebody had ran on your automatic web UI you might have other problems but this basically prevents automatic from changing local resources such as say system 32 and like just crashing your system because that would be unfortunate we don't want that to happen um yeah and I just have that as a sanity check in here then we go over to stable diffusion this is when we talk about the sort of setup for deterministic results now not everyone cares about deterministic results but as we'll get into a little bit later deterministic results are very useful and really important for actually generating high quality images and experimentation not everyone cares but let's go ahead and set this up so that you know we'll just go top to bottom so we can get deterministic results starting off with the number one doesn't matter most of these don't really matter so we're actually going to skip over this one this is just how many checkpoints are loaded at one time keep one model on the device checkpoints are cached unet go ahead and leave this on automatic that way it'll use whatever the model is designed to use this means that other people that are using the model will have the same results as you or you'll have the same results as them if you set it to none then it will try to um use uh whatever you did from the checkpoint but you can also use for example if you leave it on automatic it will use one the same wait use one with oh with the same with same file name as checkpoint that's worded a little bit strangely um but what automatically allow you to do is you can set a unit model next to the checkpoint and some models will actually have a unet model next to them so this is why we end up keeping it up on default now you could just set it to none and use whatever is default for the model but using automatic is probably the way that you want to use it just to keep setup more consistent between uh different models enable quantization in case sampler we're going to keep this disabled this will change the existing seeds so anybody else who doesn't have this enabled will have a different generation and it's not necessary so we're going to ignore that enable emphasis which is just the UI thing make K diffusion Samplers produce same images in a batch as when making a single image this is very important to have enabled so so basically when you're using case diffusion Samplers it could carry over the sampling rate between different Generations will slightly cover what that means later but not too much in this video but that's important because if somebody else is doing batches and you're doing single images you want to be able to reproduce that and it'll say in the metadata as well so you can end up generating whatever you need oh by the way enable emphasis is a UI thing but it does check or it does make results deterministic in the sense that if somebody else is using that in their prompt you'll want this enabled so that you can copy it in your prompt all right anyway so anyways prop word wrap length limit we're going to keep this on default which should be 20 and then the clip skip we're going to leave this default to one but we're actually going to add it up to the top bar so we can customize it so generally speaking for a lot of models on SD 1.5 especially for new to models they're trained on clip skip two but it will depend on which model you're using it will say in the description which clip skip you're supposed to use but you can also experiment with it we're not going to cover too much of like what clip skip is in in this video upcast cross attention layer to float32 we're going to leave this disabled but if you're using a graphics card that doesn't support float 16 this is a nice option to have enabled however I'm not certain whether it changes the seed results or not it doesn't state it does and it shouldn't but I'm not certain so I generally leave this unticked obviously I have a graphics card that supports float16 so we don't have to worry about this but yeah that gets into technical details we won't be covering this video but just keep that in mind that if your images aren't generating or you get some sort of error it might tell you to turn on Up cast Cross attention layer to float32 and if you do so it shouldn't change results but it is going to be slower so there you go this one's really important random number generator resource this is going to be set to CPU instead of GPU basically what this means is that it's going to generate a seed to generate noise for stable diffusion to interpret into an image we won't cover that in this video but if you run it on GPU then it's going to be specific to your GPU and Driver version which means that if you update your driver or you update your system or you change to a different computer or somebody else on a different computer tries to run it it will generate a different image if they're not running the same driver version and the same GPU so we actually set it to CPU that way we can ensure that even if our system updates or a driver changes we can always generate the same exact image each time because it's using the CPU which is deterministic deterministic meaning it'll be the same every time which means that we can generate the same image if we need to which is good tiling this is optimization type stuff we're going to leave it disabled because it does change result quality it doesn't necessarily change results in in its own right but it can change the quality and then high risk fix we're going to leave it on second pass this is what most people do and it's pretty much the best way to do it but you can change it to first pass if you want like um Legacy support type results which gets into technical stuff we're not going to cover here anyways next we're going to look at stable diffusion just leave all this at default but one thing I did want to mention is that there is a low aesthetic score and a high aesthetic score which if you're playing around with sdxl you get more into the technical details you might want to customize these but generally speaking you want to leave them on default as most people do because you'll get the same results as them you know deterministic and whatnot like that next for Vie just go ahead and leave these bottom two on full um some guides and such will tell you to put it on taesd for whatever reason and then people get errors and they're confused why um just useful I don't know why uh people recommend that but I've seen it a couple of times so make sure that that is there and they explain what a vae is up here as well next we're going to go over to optimizations this is where the next important bit of our deterministic results ends up coming into play up here on the cross attention optimization so there are various different ways to optimize it including xformers which won't be listed here because you have to install those separately and run them in the command line but any type of optimization that affects memory handling will cause some sort of non-deterministic result and the reason why is really complicated we won't cover in this video but basically what I recommend here is the sdp no mem skill dot production without memory efficient attention and what this does is it basically runs memory or runs a no mem optimizations on the sdp t or scaled dot production which will speed up your generation but it won't affect the memory aspects of the generation I think it saves a little bit of Iran but not a whole lot so if you are on a memory limited system you might need to use say xformers which we won't cover in this video or sdpscale.production which will affect your deterministic results and change your results so keep that in mind but generally speaking I recommend sdp no mem it makes it a lot faster it makes a lot easier to generate and it works pretty well however in the future there may be other cross-attention optimizations that we might recommend so make sure to you know if we're you know five months from now make sure to talk to people and figure out what the best optimization is but at the time of recording sdp Nomad seems to be a pretty safe bet but if you're running out of memory and stuff like that you can try the other ones should you please to do so the only other one on here is the insurer the bad prompt and negative prompt is disabled over here so padding the prompt to negative prompt to the same length this is just a performance optimization but it does change the results so you don't necessarily they want to enable it because most people are not going to change this from default but if you are having problems with performance or you want it to generate faster you can try this just know that it'll change the results that you get so if you try to generate an image with or without this option they will be different depending on the setting that you have enabled so keep that in mind um next we're just going to quickly talk about the extra networks real quick we're not going to be covering these too much um but the one thing I did want to mention is the show hidden file directories so basically if you add a period to the beginning of any folder inside windows or Linux I believe it will hide the folder from programs or by default anyways and it hides it in the file browser but you can actually enable it so you can see these this is just really helpful to have enabled just because you can use it for organization and sorting files and stuff so it's really nice but some automatic setups will actually disable this option or it might be disabled on older versions of the web UI so if you're updating the web UI you might run into some issues here and then over on the user interface this is where we're going to add that top bar over to our screen so down here if we scroll down we actually see the Quick List setting this is really important in my opinion so if you click on here you've got a bunch of different stuff and anything that you add will actually add it to the top bar up at the top so if we go back and look at the way my web UI was set up I had the vae I had the clip Skip and I had ETA noise we won't be covering what most of those are but we will be covering the AE so let's go ahead and add in sdvae here so you'll find it on the list you can just type in to end up you know have it eliminate your results so you can just type it in and hit enter or you can go and find it oh sorry you can't hit enter Because it selects whatever selected that's a new feature I haven't noticed that Vie and then we'll go ahead and click the SD underscore vae so we'll click that and then we're going to go ahead and add in clip I'll go ahead and show up like on screen what I already had here in my settings just to make it easier for us but clip stop at last layers and then we're also going to add the ETA noise so the ETA noise seed Delta and basically you know once you have that set up you'll have to apply it and then reload our web UI and it should show up at the top go ahead and load this in there we go and now you can see we have it up at our top bar the last two things we're going to cover real quick is uh two optimization settings I like to have it like that but we're just going to keep it small like that there we go all right so if we go ahead and go on here and we go back to our user interface now you notice how mine was different colors that's just a radio theme so if we go to the second option over here it's currently default but we can set it to whatever we want it will require resetting I don't remember which one I have set so in the video editing process I'll go ahead and show up whichever one I had enabled um but yeah so you can go ahead and select these and they will change it you just apply it go ahead and reload the web UI and then it will change the way that the web UI looks and then we're going to go into the info text setting as well so we go over to settings go down to info text which should be right here underneath the web user or the user interface and generally I'd prefer to have all of these enabled except for the add username to generation information when authenticated this is by default you pretty much want these enabled and this allows you to actually load up the image that you generate into the web UI and get the parameters that it used so it basically bakes the information that was used to generate the image into the image itself inside the metadata and allows you to pull that data from wherever you want or wherever you Source the image so you can also share your prompts with other people by just sharing the image that you had generated and be like hey I generated this image can you try it on your machine or maybe like they're how asking for prompt help you could just like make an image and then send it to them and then they can load it in their web UI load up the results that's super awesome that is currently in the pre-eng info so if we go back to my images oops wrong button oh I accidentally closed it so let's go ahead and open my images folder here so if we go back to my images here go to images and we just go to PNG info tab here we click our image just drag it into there or you can click it and upload it that way and you can see it shows all of the parameters that were used to generate it and then we can say send to text image and that will actually just send it over here and then um if we generate it it should apply it as we had seen oh it actually generated something else because I have the wrong checkpoint selected so we do have to have the right checkpoint selected there is an option to make it switch to checkpoint but you don't have to do that um I don't but that's what it would look like on uh aom 3 I guess uh but yeah so there we go and it gave us the same exact image that we had before so there we go so that's how you can set up deterministic results get it from a PNG file that somebody else had generated or any other type of file or whatever or you can just take up the generation data that they post say online or whatever and then throw it in your web UI and you should get the same results as long as their results are also deterministic now one thing to note before we move on is that a lot of the stuff will change depending on the user there's a lot of things that you can customize and change that change the way things generate and you won't necessarily have access to all of that and this also includes things like extensions which we won't be talking about too much here but you will see that there are various extensions that you can enable and install and these will change the results drastically depending on what they are all right so I'm going to go ahead and go and get some water so that I can have some nice breath for the next bit of this but we're going to talk about next acquiring models how to install models how to use them and then how to use the web UI as well and then do a short explainer on extensions and such like that so I will be right back let's go ahead and take a hydration break final edit from editor Michael here so we forgot to add in the default option so in the settings panel at the very bottom there's an option called defaults you just click on that and then you can view the changes that are going to happen and then you can just hit apply and it applies the changes that you have in the text to image or image to image or any of the other panels and it will just save a default configuration that you have currently set which is quite nice to use and is actually super awesome and also because I originally recorded this video as a single long part I didn't record an outro actually I didn't even record an outro for the last video either like the fast fourth part of it or whatever so yeah so thank you guys so much for watching make sure to like comment and subscribe and if you want to see more content like this definitely let me know there are definitely three more parts to this as well and it has a lot of useful information I apologize if you watch through this expecting it to be literally everything uh that is literally impossible but I hope that this has become very very useful and this first part is definitely some thing that I think a lot of people need to watch because a lot of these deterministic settings bring up questions from people a lot so there's a lot of people that try to generate images and they don't get the same thing or something doesn't line up properly or they complain that they weren't able to generate the same exact image before or whatever so that's probably the most important part of this entire video actually so yeah thank you guys so much again for watching and I'll see you guys next time I almost forgot Discord Link in the description go ahead and check it out also you can donate on Kofi to support my work

Channel: Michael "Spirit Shard" Stanich

Views: 1,938

Keywords: Automatic1111, Stable Diffusion, AI Images Generation, AI Image Gen, Tutorial, Guide, Installation, Set-up, Explained, Explanation, michaelpstanich, Michael Stanich, Spirit Shard, Michael P Stanich, auto1111, auto

Id: sFvGdKHc3zo

Length: 37min 31sec (2251 seconds)

Published: Sun Sep 17 2023