Fooocus - Stable Diffusion SDXL Install for Beginners!

hello and welcome to More nerdy rodent geekery why are the apps for generating images using AI also complicated to use they've got weird settings with words like sampler and whatnot I don't want that I want something which is the opposite of comfy UI I just want to focus at least I think that's how you pronounce a triple o sound anyway I just want to focus on the prompt and that's it nothing complicated like control net or other techno jargon getting in my way well brilliant what sort of Hardware will I need to get this thing running well ideally you'd have at least 32 gig of RAM and an Nvidia GPU with at least four gig of vram AMD gpus may work on Linux doing the usual AMD things for pytorch and whatnot but as I don't have an AMD card I haven't been able to test that the requirements on the GitHub page say just 8 gig of system memory is fine however that does mean setting up an absolutely massive page file meaning things will generate a lot slower during my testing I found it used 28 gig of RAM it does however appear to work in Google collab as well there is the link for that so if you don't have your own Hardware have a go on the collab and that should do you nicely brilliant so let's Jump Right In and get this thing running if you're a Microsoft Windows beginner you'll likely want to download the portable application if you're unsure where to download that from it's where it says click here to download however it has been compressed with 7-Zip meaning you'll need something which can decompress such files such as these 7-Zip application itself that you can download and install from 7zip.org just like it says there once you've got the file downloaded and unzipped use that run dot bat thing to get you started this will automatically download the models for you as well so no need to do that and that's it you've downloaded it you've installed it well done on the other hand should you prefer to just do a normal installation using Anaconda which I'm sure you'll already have installed if not pop over to anaconda.com download and install that then you can crack open your anaconda terminal and copy and paste the commands from the GitHub page there in the order given if like me you've already got it installed then you only need a couple of those commands you can just CD Lucas and then conda activate Focus let's copy and paste that one in as well there we go and now I've CD into the directory and activated it ready to to start remember if you haven't already got it installed you'll need to copy and paste those in one by one in the order shown then you can just start the app with pythonlaunch.py and there it is it will automatically pop open a new browser Tab and you've got your application up and running as a side note for either installation option should you already have the sdxl models downloaded you can simply link them into the focus directory there it is so Focus models checkpoints and I've just created links for the base and the refiner in there okay let's take a look at this UI then nice and simple just to prompt and a generate button there's also something scary down there which says advance but don't worry we'll take a look at that in just a moment well now all you need to do is type your prompt and then press generate so there is my prompt and I click generate it says there processing text encoder so it's going to go through through this and it will give me an image which is a little bit blurry to start with but eventually becomes clear whoa you've done it you're now a professional AI artist and can sell your images for Millions but do you want to turbo boost your Creations right up into the stratosphere well it's time to click that advanced box I know this is meant to be a simple application and it's got an advanced box and that always seems a bit confusing but never mind Advanced there we go so we've got a number of tabs here we've got setting which has loads of options in there style which has even more options in there and an advanced advanced box for the Giga Chad brains but I'm getting ahead of myself here so let's go back to these settings and see what we've got in here we've got performance speed and quality we've got some aspect ratios number of images to generate a random seed and a negative prompt if you want one as well changing the quality will basically up the number of steps from 30 to 60 in total so quality will take twice as long to render as speed all the other options in here should seem fairly obvious as with the style tab again you can just read and see what the style is so there we've got a photographic style or a pixel art style or a real estate advert because that's a brilliant style to choose the advanced advanced section has bits where you can change the model there you've got a base model a different refiner if you want also Laura's and you've got an advanced advanced advanced section which also has subling sharpness so then you know we've got three depth levels of advanced here this is this is too complex now so let's go back to the settings and we'll just change some of these and and see what happens all right so the default is 1024 by 1024 I like 1152 by 896 because of reasons I also quite like the performance one that seems absolutely fine to me two images because why not and I'm going to leave the negative prompt empty meaning all we need to do then is pick the style now cinematic default is the sort of default one so if we pick another one of this let's go uh art style graffiti why not let's have a look at that so I've got exactly the same prompt as before all I've done here really is just change the style to Graffiti and as you can see there I've got two images and it has sort of Applied that style excellent so maybe you didn't like graffiti style you can scroll down as you can see there's absolutely loads of them how about this one a paper craft flat paper cut so just applying a different style here and we'll get two completely different images awesome there you go so completely different style just from picking one of the styles that you've got available there that's quite easy isn't it we'll just quickly go over the advanced section there so say You're really really big brain and you've downloaded a model from Civic AI well you can go over to the advanced type you might have to click refresh all files base model then you can just select that that'll be a different base model and this time when we generate you'll get a couple of images from that model instead obviously what would be really great would be to just have a generate air Now button somewhere big down here which automatically does a lot of that for you so it just picks the prompt pixel style and a size and there you go it generates an AI image for you but of course this software has only just been released still if you found that fun then maybe you'll like this next nerdy rodent video too

Channel: Nerdy Rodent

Views: 8,276

Keywords: nerdy rodent, ai, ai art, fooocus, easy ai art tutorial, SDXL, stable diffusion, SDXL fooocus, SDXL easy, simple SDXL, How to download Sdxl, Sdxl for noobs, Beginner Sdxl, Sdxl basics, Easy stable diffusion, Easy Sdxl, Simple SDXL, Stable diffusion SDXL for beginners, Professional AI tutorial

Id: NfNwmKM3sxc

Length: 7min 47sec (467 seconds)

Published: Tue Aug 15 2023