LeonardoAI - Complete Tutorial / Guide

hi today I'm going to give you a complete tutorial on how to use every currently available feature of Leonardo AI we've got a lot to cover so let's go ahead and get into it Leonardo AI is an online stable diffusion platform that stands out from the crowd by having several unique features and a very intuitive UI we'll get into that in just a few minutes some people do call it the mid-journey killer I'm not one of those people I think each one of them has their customer base and their own purpose so let's get into it first thing you want to do is go to leonardo.ai now it is in Invitation Only access but it's very easy to get invited if you haven't been there you've put in your email address click on Count Me In and within a week they will send you an invite I believe they send out the invites every every Monday but I could be very wrong on that once you are in the system they will have your email address on file and this pop-up comes up and says if you're asked if you're whitelisted or not and if you are you just go ahead and say yes I'm white listed and it will open up the home page for Leonardo AI all right so when you go to the home page on Leonardo AI you will first see at the top of the screen featured models and these are all if you're unfamiliar with stable diffusion a model is a data set that has been trained on a specific type of image so if you're looking for portraits you might go with deliberate if you're looking for vintage style vintage style photography and onwards they even have a nice little paper art style here that I recommend people playing with and if you're unsure what a model does all you need to do is click on the model and it will pop up and show you some examples that have been created with that particular model so you have a basic idea as to what kind of images you can expect you can also if you want see a model you want and you want to start generating images right away you can go ahead and click on this button that says create with this model but I'll show you a different way to get this a little bit later now below that is the Community Feed here you'll see images that have been created by other users and if you want to find out how anyone did a particular image just click on the image it will come up it will show you what their prompts were if they had any negative prompts they'll show you those as well and it will also show you the sampler that are used the base models they used and the fine tune model below that so you can generate this exact image if you want by using the seed number that's over here off to the left hand side you have the home button which will bring you here you have the Community Feed which Zooms in on the community we saw in the home page and then you have your personal feed these are images that you created you can also go to training and data sets we'll get into that a little bit as well as fine-tuned models these are models that Leonardo AI has created and curated for your use but there are other models and we'll get into that a little bit right underneath that we have ai canvas beta what this does is it allows you to do out outward painting so for example if I do something along the lines of cyberpunk City skyline and I say generate that when the images are returned it actually returns you four images and you can scroll through these and find which one that you like the most so for example I happen to like this nice purple one here I click on accept and then what I can do is I can move this box out to like the right hand side when you do this you want to keep at least half of the Box on the main image so what you're going to do is you're going to widen this picture and it needs some of the original picture to know what it is it's actually doing so once you have that box out there you then click on generate again and it will expand that picture out and again it gives you four options so you can look at the options and see which one fits your style the best and I think number three here Works fairly well click on accept and we now have a new image you can also go in and do masking so for example if you wanted to redraw this building right here and you can increase the brush size with this slider over here I can then brush out this building click on generate and it will redraw that section that I have masked out like so and again you get four different versions that you can toggle through find the one that you like click on accept and you have a new updated image you can then download this image or just cancel out of it completely so we're just going to go back to the main screen here by clicking on exit the editor and then we're going to go down to the all important all image generation button and this is a screen that it brings you up to now we're going to go down the left hand side here and then we'll come over to the image section going from the top you have the number of images these are the every time you generate this is the number of images it will generate for you now one of the things to keep in mind is that you are granted about 250 tokens per day which means that roughly you can convert about 250 images or you can create 250 images but there are other things like the out painting and such like that anytime you click on a generate button it's going to cost you points so keep that in mind when you're picking out how many images you want to generate with every click of the generate button underneath that you have the dimensions and there's three different ways do Dimensions here there is clicking on one of these presets and then underneath that you can adjust the presets manually so if you have a specific size you're looking for you can go that way or underneath that there are three different aspect ratios one to one sixteen to nine and four to three that you can use as well there are some things to consider when you're doing this when you're picking a model that you're going to generate your images from those models were trained at a specific resolution and in Leonardo AI we'll show you up here a warning sign if you pick a resolution that is outside of what was of what the model was trained on underneath that you have the guidance scale and step count basically the guidance scale is how closely you want Leonardo AI to stick to your prompts higher the number the more strict it will be in looking at your prompts the lower the number the more freedom you give Leonardo AI or stable diffusion to generate things that it thinks you'll like but if you go too far it can mess up your your image and if you go too low it can also mess up your image I find the general sweet spot to be between 7 and 10. underneath that you have the step count the step count is how many times you want stable diffusion to look at the image as it's creating it the more times it looks at the more time the more detail it can add to it however the larger the number the more time it takes to create it and like the guidance scale if you have the number two high it can mess up the image or there's diminishing returns where you're saying it to 100 may not improve the quality much over say 40. so you want to find the The Sweet Spot for you I find my sweet spot is between 20 to 50. under underneath this you have tiling now tiling let's say you want to create a wallpaper you want an image that can be tiled seamlessly so you can stack the images on top of each other to the sides of each other and it will look like one continuous image that's what tiling is for it's basically stable diffusion we'll look at the image and say okay this is a you know we're going to arrange the the design so it could be repeated endlessly underneath that you have imaged image now what this does is you can upload an image you like and then use that image as a base for the new images that you're creating so it will take that image plus your prompts into consideration when it's creating the new image for you and it gives more weight to the image than your props so they basically build on each other underneath that you have the show advanced settings and basically what that does is it turns on the used use fixed seed and scheduler using the fixed seed like I had shown you before if you click on an image it will tell you what seed was used for that so if you're looking at the Community Feed and you click on an image it'll say that this seed here is what was used for that so if you said everything else exactly the same it will go ahead and create you that exact same image if you want but typically you leave this off because it will generate a random seed every time so you get random images every time underneath that is a scheduler which if you're used to something like playground AI it's also called the sampler this is where you pick your models or your this is where you pick the I guess you would call them guidance files for your image basically if you have an image you can switch between these and it will change how the image looks based off of that particular image enhancer the one I normally use is Euler ancestral but I highly encourage you to play with all of them so you can get a better feel as to what they will do to your particular image and of course right below that is a reset to defaults button which as you guessed it will reset everything to its default settings getting over here to the main part of the screen the top part here is your prompts basically these are the things you you want in the image that you're creating underneath that are your negative prompts these are the things you don't want in the system now you'll notice that some of these have multiple parentheses on either side of them the way that stable diffusion works is it uses these as a general guideline it may not take all of them into consideration or may not even understand what it is that you're asking it not to include but if there's something that you definitely want in there then you want to put some brackets around it because that tells stable diffusion pay attention to this particular prompt more than the other prompts now it's not a guarantee that we'll do it but it at least increases your chances that stable diffusion will look at your prompt and say okay this time I'll do it underneath that this is your model this is the the main thing that you want to look at when you're creating an image different models have different purposes so for example if I want to create a female character there's a female character model if I want to create create a portrait there's a portrait model if I want to create anime there's anime models and if I want to create something that's more artistic the Leonardo models work just fine for that but if you want to look at different models you can click on this spline down here saying select custom model and what will happen with that is it will do this pop-up here and if you created any models in the past which we'll get to in a few minutes they will appear here on the favorites if you favorited any models they'll appear here the platform models are models that Leonardo AI has looked at themselves they vetted them or in some cases they completely created some of these models and you can feel free to go through these and you can pretty much assure yourself you're going to get some decent quality images from this because the team at Leonardo AI has gone through here and you know made sure it these will generate good images and then you have Community models these models here were generated by the Leonardo AI community and when you generate a model you have the option of sharing it with the rest of the community and there are a lot here that you can go through and play with and I highly recommend doing that at the top you have two different ways to sort this actually three different ways you can search the gallery if you if you know what the name is or you're looking for a certain keyword you can sort it from newest to oldest or oldest to newest alphabetically the number of images that was generated it by that particular model which gives you hints as to how popular it is or you can sort by the category that the model falls under so if I'm looking for fashion models click on fashion and these are the models that come up for that if I see a model that I like I can click on it I can say I want to generate with this model now one of the things to pay attention to when you're looking at this is the resolution of this model this basically says that this model is trained on 768 by 768 so anything that's outside of that may give you differing results so when you go ahead and click on generate with this model it will come over here and it will automatically have that model selected you can then click on different styles now Leonardo's style is basically a modifier on top of things that you can use you can go back and forth between these ones I haven't really noticed that much of a difference but I'm sure it does something what it is I'm not entirely sure I think it basically refines the curves a little bit next to that you have negative prompts and what that does is it turns on and off the negative problem box and then you have prompt magic what prompt magic does is it looks at your prompts and it tries to refine them a little bit into something that stable diffusion will understand a little bit more then over next to that you have the generate button the generate button is just what it says you click on that it generates the images that you asked it to generate down below that you have the image generation Tab and this tab shows you basically all the images that you created one of the things that makes Leonardo AI unique aside from their custom models and the ability to have Community created models is the prompt Generation Now what the prompt generation does is I can type in acute Doc and I can say I want six more prompts you have the option of adding 2 4 6 8 10 15 20 or 25 different prompts to this and I tell it I want six extra pumps added to it I click on id8 and it will generate six different sets of prompts and in each one of these it adds extra prompts and it basically allows you to look get different things and it's nice because it can give you ideas you didn't generally have and if you like one of these you can then just click on the generate button and it will automatically generate an image with that or if you want to add something to it or delete something you click on this down arrow and you click on edit and you can then come in here and edit the button or edit the prompts so that is this main screen all done and dusted now let's get into the other main feature one of the things I want to touch on real quickly is when you choose a model there is a chance that when you click on something it will change the resolution that you set up before so you just want to keep an eye on that if you had a specific resolution in mind you want to make sure it didn't get changed if you change your model type all right so let's go back to the home screen here and we're going to go into training and data sets what training data sets are is if you want to create your own model from images that you have it's very easy to do here you come in here you click on create new data set you give your data set a name and a description and you want to make sure that this description is as precise as possible because if you decide to release this to the general population others will be able to know better what type of images they can expect from your particular model so once you have your name and your description typed out click on create data set and it comes to this screen and this is where you would upload the images that you want to train a data set on when you do that you want to make sure that all the images are basically of the same type if you're going to do animate upload anime figures you can do portraits upload portraits you know if you're going to do architecture upload architecture you also want to make sure that they are all the same size so 712 by 712 512 by 512 Etc and that way stable diffusion will go okay we know exactly how to fill out the size of canvas with this type of image and you want to upload between I'd say 8 to 15 different images to create a decent data set the more images the better if they're closely aligned now what Leonardo AI does to help you with this is it actually gives you access to the Community Feed where you can add images from here to your data set or if you have images in your personal feed you can then drag images from there as well or you can upload your own image once you have an image uploaded let's say I want to create a data set with this I just click on that it pops up here and then I can click on the train model button I'm not going to do that here because I'm not creating a model once you have about I said 8 15 images here you click on train model and it will go back to Leonardo ai's servers and it will start to train that data set it does take a little bit to do that and you can check on that by clicking over here on the job status and it will show you where it is within the process of being created once it's created it'll be added to your data set sets so you can easily use them for images going forward so this is my complete guide to Leonardo AI as you can see it's fairly simple to get through however if you do use Leonardo Ai and you have tips that you'd like to share with others please do leave them down in the comments also if you have questions leave them down there in either I or maybe one of the other viewers will be happy to give you an answer to your particular questions so I hope you enjoyed that if you do please feel free to leave a like button and as always I wish you a pleasant day have a good one

Channel: AI John

Views: 180,477

Keywords: AI, art, LeonardoAI, Stable Diffusion, deep learning, algorithms, image generation, web application, customization, text prompts, user feedback, user-friendly, real-time, visual art, creativity, sci-fi, landscapes, portraits, abstract art, professional artists, art enthusiasts, technology, machine learning, computer vision, neural networks

Id: PXcgAzwcDX8

Length: 18min 25sec (1105 seconds)

Published: Thu Feb 23 2023