Generate THE BEST AI Anime Images For FREE! (MeinaMix, CetusMix & More - Stable Diffusion)

this image this image and this image were all AI generated in this video I'm going to show you how you can set this up completely for free including things you can do if your PC is very slow if you're a complete beginner watch this entire video and I'll give you everything you need to know to make amazing images however if you have seen my previous video tutorial and already got yourself set up with anything V3 then I have good news because all you need to do is to Simply download these new improved models and that's it and you're done go to this timestamp to skip over the basic setup so first let's begin with the requirements we'll be downloading a program called stable diffusion that can create images from text also images to images but I won't be covering that the amount of vram your graphics card has will limit how high of a resolution you can render however if you have 4 or less gigabytes I have a workaround which I'll show towards the end of the video If you have a very slow graphics card like really slow I would not recommend doing this because even if inferior yes you would be able to make images you'd need to wait minutes to even load a single image and in this case I would recommend you check out yodeo.com which I've recently found which has essentially all of the models I'll be mentioning in this video and you can generate high quality AI images all without needing a good PC yourself another small note this program takes up 20 gigabytes of space it's not tiny but not really the worst either so now that you know that let's get into the setup first off we'll need to download two depend is that this AI generator named stable diffusion runs on Python and git all of the links are in the description so first off go on to python.org so when you're on this page you go to the bottom and right here you want to download Windows installer 64 bits once it's open you'll see this and it'll say install now but before you click that go to the bottom here and check add python 3.10 to path this is very important without this it just won't work so make sure that this is checked now you can click install now and that's it once this has downloaded you can close this out and go on to the next website or send the description kit Dash SCM and then go on to here and click the 64-bit git for Windows setup download this and open it up once this has opened up just click next and here we want to make sure that everything is selected this top one it doesn't really matter you don't need this but up to you make sure that all of these are selected very important and then go next I mean I've already have this and then when this comes up up you want to make sure that it says use notepad as gits default editor most likely will be something else but you want to make sure it's notepad not notepad plus plus not anything else use notepad and then next next just click on thing it should all be correct by default next up you want to make a folder wherever you want your AI to be it shouldn't matter where it is but make a dedicated folder I'm gonna make mine on the desktop and I'm going to call it stable diffusion now go into the folder open it up and right here at the top you want to hover over the navigation panel click it's going to select like this in blue and then you write c m d enter a terminal will now open and as long as your git installation was correct you can now write in git clone and copy paste this link that I have in the description in this GitHub link and click enter now it's going to install stable diffusion onto your PC automatically so just let it load but in the meantime we'll be checking out the amazing models for this AI generator now here you actually have a bit of a choice go into the description and go on to civitai.com here you will see loads of images all of these are models or how this page likes to call them checkpoints make note though if it says Laura like it does here then these are a different thing entire early so ignore them for now so what is a model basically a large collection of images of some type that make images in a specific style there's more to it than that but that's all you really need to know I'll be picking an anime one I personally use but as you can see there are also real life models and models of all types whatever model you pick the next process is the exact same but in order to help you out I've actually went and downloaded 12 of the top anime models from this website and gave them all the same prompts and description and then use the same seed so they generate in a similar way so you can see the difference between the model you can see the names of the model so if there's any of these you specifically like then remember that name I'm gonna pick Mana mix which in my opinion is very nice and on top of that is easy to work with and isn't overly stylistic but don't be worried to try a different model yourself so whichever model you picked just click onto it and then click download this will download as either ckpt or a safe tensors file format next up is the VA file basically the color correction now technically you could use any VA file here in fact if you check the description of most of these models they'll give a suggestion but I will recommend one myself which almost no matter what model you're using will look really good it's also the most used one so it's a safe bet also the one I used to create this image from before so just go with it why not go on to the hugging face website which is in the description and click download this will download ACL f8.animate2file a very catchy name I know so now all you need to do is wait up and let these two things load and then come back now that you have both of the files the safe tensors file and the VA file you can open up your folder going to stay up the fusion web UI go into models then go into style to the fusion and in this one you want to drag your first one your safe tensors one the actual model second off you can go back to models click VA and here you want to put in your VA file and now you're almost done all you do now is you go back to step the fusion web UI this folder right here and right at the bottom you'll see webuser dot pads click on this right click on it click edit and this will open up a notepad and here you want to add one line after set command RX you want to add git pull what this does is that every single time you open this up which you'll need to do every time you're using this AI it will automatically update itself so that you're always on the most up-to-date version this makes sure that you have less problems with the program in the future and whatever not so very good do it second of all if you have a weaky graphics card or using an AMD graphics card you'll need to do an extra step here with a weak graphics card 4 gigabytes of vram or less or if you later on notice that you're still having problems go to this line right here set command line RX click space and then do dash dash Double Dash Med vram or alternatively low vram this will use up less vram and allow you to make higher quality images at the cost of time it'll take a little bit longer to generate images than normal but it's not the absolute worst especially if it's your only choice if you're using an AMD system you want to write the Double Dash Precision full space Double Dash no dash off space Double Dash skip torch Dash Cuda test this is all in the description so you can just copy this in this is required if you're using an AMD system because normally this is only made for NVIDIA but this completely works for AMD as well you just need to add this in and guess what if you have a slow AMD system also make sure to add dash dash Med vram or low vram now save this and go out of it and go back to this folder now you want to double click web UI user bats this will open up a command prompt and it will download everything to make sure everything is working as it should this can take quite a while like really long sometimes for some people it just takes 10 minutes for some people 30 to 40 minutes but after this you're completely done which is very nice so let this load in the background go get a coffee or something and come back now that it's done it might right away give you an IP just like this in that case write this IP into your browser and Bam you're done sometimes you might need to restart the terminal then just turn it off again and then go into your browser and write in the IP and here it is here's the GUI one last thing at the top you want to click like this and make sure that it has this selected second of all you want to go to settings then go to stable diffusion and then you want to go to sdba click on this and you want to click whatever VA file you downloaded in my case klf8 anime 2 and then you want to go right to the bottom you see clip skip you want to change this from one to two I've already changed some of the settings that's why it's already done but it should be from one and you drag this to two after that click apply settings and now you're done you can go back to text to image and make whatever you want I won't go super deep in this video on how to use it but I have made a completely separate video on where I go in depth but it's also not super complex to do this so first of all you write in tags here for things you want included like short hair pink hair one girl bunny outfits either that where you could write a sentence instead so in this case that would be girl with short pink hair wearing a bunny outfit whatever suits you best and now you can just click generate and it'll create a very nice looking image and there we go that's pretty nice but of course we could make this a lot better by checking out these settings on the left here so let's have a quick look at what all of these settings mean sampling method is what method is used to make these images some are better than others and I would recommend anything with DPM plus plus at the beginning of it so these three sampling steps is how detailed you want the image to be 20 or 30 is pretty good but you can go higher but it'll take a lot longer the higher you put this the longer it'll take width and height is the resolution often like putting width on 512 and increasing height to around 800 to 900 but 768 is good as well CFG scale is how much the AI will conform to your prompts the higher you put it the less free it will be but it will create something more similar to your prompt but in my opinion you can keep this around 7 maybe 8 9 is good but six could also be good so just do a bit of trial and error and you'll find something that you like if you have a pretty good graphics card you want to go to this Iris fix and click on it and turn it on and then you want to make sure that if it's set to upscale too you want to drag this to one and with this you have all of the optimal settings now generate an image and you get this this looks very nice but if you zoom in you might notice that it's kind of blurry like what's up with that why is it blurry well no problem we can actually fix this because on the bottom here it says send to extras click on this and what this does it sends it to a place to upscale what upscale so you click here on the upscale one thing and you write a click either Ester gun for x or estergan 4X anime and I like the anime one because we're doing anime ones then you click generate and look at that that's way cleaner it's not very blurry anymore and yes it's not perfect but this is where you'll need to try it out a little bit you'll need to get used to how the bot works and I've actually made a completely different video on how you can do this much better if you have any additional issues check the comments for what others have said and alternatively you can DM me on Discord and I'll do my best to help but if you truly want to make the best possible images you can then watch this video right here to see how you can make better prompts and make better custom characters [Music] thank you

Channel: Preston Ch.

Views: 71,801

Keywords: Generate AMAZING AI Anime Images For FREE! (Anything V3 - Stable Diffusion), AI Images, how to generate ai images, stable diffusion, anything v3, how to use anything v3, how to use meinamix, how to use stable diffusion, the best anime ai models, the best ai models, the best ai generators, the best loras, the best checkpoints, the best models for ai generation, ai art, ai art generator

Id: yuUfiX5oYFM

Length: 11min 21sec (681 seconds)

Published: Sun Apr 09 2023