Creating (free) AI Art with Stable Diffusion: Beginner’s Guide

we're gonna learn how to install stable diffusion with custom models to create images like this this and this let's get right into it we're going to start with some prerequisites first off we need to install git to get started we need to head over to the official website of git which you can find Linked In the description below you can download it by clicking the link at the top or one of the links below corresponding to your operating system once it's downloaded just open the file and the installation process will begin you can stick with all the default options it's really easy [Music] now that we have get installed let's move on to our next prerequisite python 3.10 to get python up and running on your machine we need to download it from their website which you can also find linked below once you're on their website simply scroll down to the bottom of the page and click on the version that says recommended during the installation process make sure to select the option that says add to path and that's it we're ready to get into installing stable diffusion to start we're going to clone the web UI Reaper trust me it's not as scary as it sounds here's what we need to do create a folder of where you want your stable diffusion to reside I personally like to keep mine on my desktop but feel free to choose any location once you've decided right click in that folder and select gear bash here a git bash terminal will pop up and in that terminal type in the following command that corresponds with your GPU manufacturer sorry you can't copy paste it but it's on screen now as well as in the description once you've entered the command it should only take a few seconds to load and once that's done that's it it's cloned you should find a new folder called stable diffusion web UI alright step three we need to download some models you have two options here you can cover the official stable diffusion models or try alternative bundles made by users which I'd highly recommend and is what I'm going to show you the best website to get models is citify.com as if thousands of custom models to choose from all suit in different needs for starters I'd highly recommend trying out deliberate it's a powerful all-purpose model that can generate some great stuff once you download a deliberate or any other model of your choice simply move it to the stable diffusion web UI slash models slash stable diffusion folder now set for is optional but I think it's a good idea especially if you have a slightly weaker computer in your stable diffusion folder find the web UI user.bat file and give it a right click now choose the edit option and look for a line that says command line underscore arcs and change it to command line underscore args equals dash dash Med vram by doing this you can generate art at larger scales don't mean sacrificing a small percentage of generation speed but it's definitely worth it for older computers alright now it's time to launch the word diffusion simply double-click on the web UI user.bat file to launch the program remember to launch it as a normal user not as an administrator during the first run stable diffusion will install some dependencies so it could take around 10 to 20 minutes once it's done you'll see the command window shown address copy that address and paste it into your browser [Music] and now we're in the stable diffusion UI at the top of the UI you'll notice the option to select your model or checkpoint as it's called but since we've only got one model we don't need to mess with that as it's already pre-selected for us beneath you'll see two text Fields one for positive prompts and one for negative in the positive field you can type in all the things you want to see in your image separate it by commas for example I'll type in blue hair Barista well look coffee shop serving viewer coffee and realistic in the negative field you can mention what you don't want to see in your image for me that would be anime bad quality and watermark and underneath we can see a variety of options we have the sampling method which I'm just going to leave as the default but depending on what you choose we'll have a different method for creating the image will therefore produce a different result but Euler is fine for my needs we have the sampling steps which is how many times the image will improve itself higher values would take a long time so I'm just going to leave it as 20 for now which is good enough for most use cases we have the resolution which I'm going to slightly increase we have batch count which is how many images are produced I want three images so I'm going to set it to three batch size is how many images are produced at once which means generation speeds faster but costs more vram so I'm just going to leave it as one CFG scale determines how true to your prompt the image will be the lower the value the more creative the image output now that we've configured our preferred settings we can do it generate and see what it produces wonderful even with a simple prompt it produces images like that well I hope this guide was helpful for you by the way if you're new to stable diffusion or struggling to generate the images you want I have guides hosted on my website jamesbeltman.com let's go over prompting techniques as well as some of the best extensions for upping your stable diffusion game anyways thanks for watching if you have any questions or need any further assistance feel free to leave a comment down below

Channel: James Beltman

Views: 4,140

Keywords: ai art, stable diffusion, stable diffusion tutorial, ai art generator, ai, midjourney, ai art tools

Id: WyOE2RreN6Y

Length: 5min 20sec (320 seconds)

Published: Fri Jun 02 2023