Empowering creativity: Generative AI art techniques

1. Hello!: Are you curious about AI art, but don't know where to start? Or perhaps you have tried some of the generators like mid journey, but never got a really good result. Or you have doubled. You have some good results, but now you don't know what to do with the art. If any of these sounds like you, this is your class. Hello, I'm Mandara I'll be your instructor for this exploration of generative AI art. As an artist and a tech enthusiast, I've seen firsthand the powerful impact that AI had in my creative process. Not only for creating the pieces, but also for inspiration. Now I'm very excited to share this expertise with you. In this class, we demystify the process of creating art with AI algorithm. No jargon, practical insights and techniques that will enhance your creative process and expand your artistic horizons. You will not only learn to do illustrations like this one, good ones, but also how to do stickers or wallpapers or coring pages for a book and how to really take advantage to this new technology for your personal life or your business. You will understand the rules to do a good prompt, how to fix the most common AI problems like too many lens, and what tools to use depending on what you are looking for. We are going to use what I think are right now the best generative AR tools, Copilts from Microsoft and NICA. Ready to unlock the potential of generative AI art, joining in empowering creativity, generative At techniques, and less dive 2. Project: Okay. The project for this class is, of course, to create your own image with AI. In this class, we are going to use two generators, Cafe and copilot. You can use either of them for the project, although I highly recommend that you use both and share both with, of course, the prompt. For inspiration and to help you a bit, I've created a list of problems that you can try and it's going to be linked below. Of course, you can use those. You can use some of the problems of the class, or even better, you can create your own problems. That's it. Let's get started. 3. Copilot: Address to go to copilot thatg is there copy microsoft.com and you can go with any Internet browser. To just log in there with your e mail and it can be a Gmail address. It doesn't have to be a hotmail address. Then you can choose a conversation style here. They are pretty explanatory, so the conversation can be both from precise to creative. The only thing you have to do is just ask what you want. In this case, it is important to use the word create. Because if you set only fairies with a play background. What Copiot is going to do is to show you images on web pages where fires are on the web. Because we want to create an image or we have to create several images, it is important to use that word. And then Copy is going to ask you something like that, and then you have to wait. What I'm going to do is I'm going to accelerate all the images, so you don't have to wait with me, but when you are doing it, it takes several seconds, sometimes even up to a minute to create the image. So This is not exactly what I was what I wanted. This is because the precision, B precise is very important when you are using AI. So what I wanted was several images with one fairy on them. But I asked for ferries in plural. And that's what he gave me. He gave me several images with several ferries on it. So we are going to try again, and in this case, we are going to ask for several images with one ferry on them. And the plain background, it is important because he's going to help us to edit later because we want to create clip parts of these images, so we don't want a background right now. Another thing that opio does is he suggests these questions that you can see on the bottom. And you can just ignore them and click on them, and he's going to add flowers to the background or create another image, or you can say, no I'm good and he's going to finish this session. These suggestion questions are usefully good to learn how to ask for things. As you can see, now, what it has give us several images with one fairy or several little different images in the same style of different fairies. This is really good to make collection of stickers or collection of clip parts that you can use on your crafts, so you can sell them. So this are really, really great use for copilot. Okay. Now we're going to do is how these digestion questions work, just click on then, it's like you have asked that question. And you can ask pilot to create the same images once and again, but they're going to do different styles every time if you don't specify an style. That's okay because you can try until you find something that you like. As you can see this style is different. These are really cute. To see them, you just click there and here you can move to the right or to the left with these little arrows. And then we are going to use one of the questions to show you how this works. You just click on it, and what is going to do is add both prompts. This is how they're called those proms together. What Copt understand that you're asking is several images with one ferry. And butterflies in the background. A, you see the style is again different and he added or he added the butterflies. Now, we are going to try to get op to do a collection of images of fries. In this case, adding flowers, So we can have even more assets to use. To create a collection, I just asked for a collection of fairy icons. In this case, I'm going to add a whimsical style. You can say something like a Kawa style or this is important. Here, I'm using a whimsical fairy tale theme, but you can say animals, you can say dinosaurs, you can say car, anything you wanted, everything that I'm explaining, you can adapt to whatever you need. These latest images are great to do a scrap booking, to do stickers. You can do patterns with them, but because you are going to have a lot of images in the same style, and here they are. You click on them severed flowers as well. So as you can see, it's a quick process. In a few minutes, you can have a lot of images of good quality on the style that you want, and you can create collection with them and sell them or use them on your crafts or on your designs. Because you can add little things to a flowers or mushrooms or whatever you fancy, you can have a cohesive collection very quickly. What about if you are tired of fairies and you want to change the subject? The only thing you have to say is do the same with blah, blah. I said unicorns here to keep on theme, but you can say, do the same with mobile phones and Copy is going to understand that you want a collection of icons in a whimsical style of mobile phones. And here we have a collection in fact, several collection of unicorns. They are very quick. They are very cute. But I think if I want to make a collection of stickers with it, I need some rainbows as well. So this is going to be something I could sell later. You can sell this kind of thing like digital collection or you can make yourself stickers and sell them on Es and places like that. And you can do patterns to the five fabrics or wallpaper or give wrapper or clipads for any hobby or that you can have you want to sell to. I use these kind of images to create stickers for kids. They're usually a great gift. Here you have a lot of Unicors with a lot of rainbows. Now that you have opened this image, we can click on customize, and what that's going to do is going to open the designer. This is an easy to use design program, very similar that anything you can have on your phone. And with this allows you to do is to quickly remove backgrounds or quickly select parts of the image. You can copy then and you can then put them on any design program that you usually do. This is affinity designed is the one I use. So you can play with them. You have to know that all these images are 1080 by 1080, so you can do really big prints with pile, but this is not the use that we have for it. But for two stickers, for example, I've talked several times about them, so let's do one. The only thing you have to do is just add a white line and there you have a professional looking sticker that any six year old girl will going to love. This is our first step with copilot. As I did several examples so you can have an idea of how this works. But I'm sure you have a lot of ideas of what to do. So we are going to do something a bit more complicated. I'll see you in the next class. Okay. 4. Backgrounds: This class, we are going to create illustration. I know I said that all the images are 1024 by 1024 that for reference is the size of Instagram post. It's not a lot that you can do with them. You cannot bring them as a whole illustration. But there are things that you can do. One of the things that you can do is back grounds for phones, for example. It is a good way to learn a bit about generative AI with copilot as is very very quick and it's free. We are going to create later in this class, bigger and more complex, really big images. But what we're learning here it will be useful later. Let's do it. In this case, I said, create a foam background. Probably, you don't have to say a foam background, but it adds details on adjectives and precision with AI is very important. Create a foam background on a fantasy inspire cute to. Remember, gives are your friends. Beautiful architecture, calm, full of fos, no people. And let's see what copilot does with this. Here they are. As you can see, the images are more complex and they are full of details. That's something that AI does very well. So we know how to do this. We know that we can ask copilot to add something to an image. We can have this kind of conversation with it, and let's see what happens when we ask for a river to the image. Here is the style is different. We have a river in the center of town, which I like. But you have I probably realize that even if we said no people copilot added people in several of the two in fact of the images. That's because AI is not perfect. Now, other thing that you can do is change colors. Perhaps you want more arthy colors. You just simply don't like the ones that op using. You can directly as make another one with different colors and you can see there and it's going to change the style and it is going to change the colors. This is more bluish colder than the other ones, and you can see the style is completely different in some of them as well. But it's not the style I was looking for. You can always ask opio for what you want. What I want is more medieval houses, a bit more classic fantasy. Houses. Again, AI is so quick that trying things is always a good idea. You can always go back but if you are an artist, you know how long will it take you to create one of those images. I know that every one of my pieces takes around three four weeks of full work. This makes a difference. You don't have to use these images exactly if you don't want to, but you can use them as an inspiration of a composition or just getting ideas of how you can do something. That's more likely is a bit of a Tudor town. I like the towers. What you can do now is choose one of them, click and download that you have it, you can put it whatever you want, and then you can open them in your program of choice. The only thing you have to do here is there are two things. You have to know the size of your foam background, it's going to be 69. I think now all the phones has 16 now as background. But just I'm going to put there several sizes so you can check. But you can take the image, make it a bit bigger, as I'm doing here, and of course, it's going to be bigger than the size of your phone. But then you can choose What side do you like the most. You can use this side of the image for the background. You can use that side of the image for the background, and there are phones that you can use y the image and it's going to move. You can do that as well, but here you have it. In fact, this is the background I have on my phone now. As you can see, you have a very quick way to create images that you can use, but also you can practice for more advanced AI generativelaor. But all these images have a use, and I think the quip. 5. Colour palettes: Okay. You can also ask copilot for do a color pilot inspired in something. You can say something abstract, like in this case, fantasy tiles or summer, but you can also be a specific and ask for flowers or a picture. Let's try it. As you can see, what copiot does is creating an image on the theme and then gets the colors from it. From this image, you can easily get the colors with an image editor or you can ask copilot for the colors in this image. Okay. As always, you can low it, of course. Let's ask copy for the color for the specific colors. As you can see what I did is asked for the number of the colors. What Copious give us is the x h x numbers of the colors he used or he has in this s. Now, you can also ask for palette that we go well with another color palette. For that, it is good to know that there are seven color palette schemes being the principles monochromatic analogous and complementary you can ask copilo for them. And what it does is it creates color palettes based on the theory of a complimentary or analogous color palette. Another thing you can do is to upload an image and ask Copy to create a color palette from that image. Copil is not going to give you all the colors that are present in that image, but he's going to give you the main palette. You can create something inspired in that color palette. It is very powerful. In this case, I've uploaded one of my pieces is thinking and he's going to give me the main colors from this image. Here they are. Those are the hex colors of the image I use. Another useful thing that you can do with Copil and any generative it really is to create is to help you with branding. So you can ask copilot for a color panel for anything. For example, a summer holiday hotel branding. It's going to give you you can just give what is your company about or what is your VNA out. It can be your wedding, it can be birthday, it can be your company, it can be just work for school. But you give copilot or any AI the feeling that you are looking for, and he will give you a branding color like this one. Remember, if you don't like the first one, you can ask again, just ignore the numbers. They are not the real numbers of those colors. But I think it is a very powerful tool, and of course. Of course, you can use those images for your social media. 6. Logos: In this case, what we are going to do is something a bit more abstract. A with AI and in this case with op you can be more abstract and ask for a logo or a design Po inspiration without being precisely what you want, but giving the information of why who you want it for if that makes sense. In this example, where I created what I said the prompt is generate generate a logo for a female artist, especially in fairy tales art. That could be me, for example. As you see here, we are not asking for ferry or a medieval town like previously, we are saying we are an artist and we need a logo, and this is what we do, please create it. This is great, not only for to generate the logo, but to get inspiration. Now, as you can see very well in this first image, copilot and any generative will will make more mistakes than a human will. For example, in this case, the brush is upside down. But this can be only an inspiration and second, these mistakes is easy to fix on later. Keep trying and don't be discouraged just because something happens, that is part of the generative AI and it's always fixable. As you can see here, C that only create logos is also create more caps of how to show this logo. This is very normal because how AI works is that they create images based on previously created photos or illustrations. It's the case, it's very normal to have on Internet, for example, logos that are on a more cap. So I'm going to do again, what I've done before, and I'm going to add some extra things. In this case, I'm going to add stars and medical creatures because it goes well with my branding and you can add whatever you want in your image. I'm doing this so you see and you understand what the process is. You something simple and you get it more and more precise with these kind of questions. Now we are going to open one of these. As we talked before, copilot makes mistakes. The generative AI has a special difficulty with lettering. Fairy tales is it's okay, but especially artist, a specialist is not well spelled because he doesn't understand the spelling, it's only shapes for them. What we can do is go to Microsoft designer. We click customizer always and we select the test to edit it out. So don't say remove object because then it's going to be this white shape. What we need to do is use this erase object option. You sell it carefully. Yes click on the letter, once you have the selected, click on Erase object and copilot is going to the designer and in this case, is going to delete it and fill it with the background. There is. You can in this case, what I would probably do is delete all the letters and then get this letter and as inspiration to add later my own font and my own text. As you can see all the images are different. I like this as well. In this case, he created a black and white image that it also worth because I didn't say colorful. You can say colorful and try to avoid this black and white because but because you weren't precise, you will have these different options, and it's very normal to see a black and white logo, isn't it? Now, I want to add and this is again, the process, you create something, you see the results and you add or change something. I want a bit of a more whimsical style because personally, my brand is very whimsical and you can do this for woodworking workshop, for your coffee shop, for your birthday, for a special day for Christmas. You can do this for anything. You say, Christmas logo for my family and he's going to create it. It's very useful. You can use the logo in web page, Christmas cards, as packets, little stickers for your gift wrapping, and of course, as your professional logo as well. Then we have another four designs that are definitely more munical you can use whatever you like. 7. Midjourney vs Nightcafe: Okay. Let's start talking a bit about me journey. To get to me journey, which is a very famous generator of AI art, go to that web page, you're going to find that animation, and then you have to click here on the right, J veta. That gets you to me journey. You just upset the invite. This is this coded channel. This is where me journey resides. As you can see, you have a lot of channel here just pick one of the newest one. I'm only going to show you the free side of me journey. Very quickly, I'm going to explain why I don't use me journey that much, and I prefer to use NICA f for all my AI work or almost all my AI work. The main reason is that it has no privacy. I know that you can have a private if you pay for it. You are in private channels, but still, I find it a jarring to have to work in the discard channel to create your art. I don't really like it. I feel bit inclusive. And the other reason is, I think that NCA f has more options and it's easier to use. So If you want to learn med journey, let me know and I can do another class about me journey. Talk to me, and I will do it. But I really think that cafe is a better option for almost everyone. If you love me journey, please don't help me. Let's start with cafe. You go to that address there and you find this is the first thing that you see is an art generator and it is a free one. It works with credits, but I didn't have I've never have to buy credits to create my art, and in fact, I've created this whole class without how to buy credits, just to show you that you can do a lot just with the credits that Cafe gives you every day. By the way, I don't work for NCA f NCA f hasn't sponsor anything. This is my honest opinion. Twist start just clicking, looking to create Here you have. This is the start screen. This is where you create every scene. Right now, it may looks like it has a bit of a lot of action and it's very complicated, but don't worry, you can start right away. Like in compit only thing you have to do is go to the test prompt and write your prompt. What I want to do here is talk about what does, what makes a good prompt. To have a good prompt, you need four steps, five steps. You need to say exactly what do you want. You have to describe your content. Do you want a photograph? Do you want a three render three rendering? You want a sketch, an illustration, an embroidery. That is how you should start your problem. Here you said as you can see a three D rendering. The next one is what do you want? Is the subject, it can be an owl. It can be a person, it can be a tree, a castle or in this case, a pastel color star or a star. Then as relevant details. In this case, passel colors, I could have saying the number of points. Although number is not something is very good at, but I could do a stone castle or I can do where I'm going. Then you should describe the style. Of what you're looking for. I'm looking for something real or erialistic or abstract or artful, or use an artist's name. Be careful, and don't use artists that are still alive and working today. But because that is a bit, complicated. We're going to talk about this a bit more later. The last step for a really good prompt is defined the composition. You can say the m, the camera view, the experato, less solution. You want a pixelted image, you want a hk image. In this case, what I said was a sallow deep of focus and the style, it was a whimsical pastel color, whimsical look. I want to something simple and I have this style that is a bit of a cake. I like the style. That's how you start working with Cafe. That's it. I know it's very simple. I didn't want to do anything complicated. We'll do a lot more complicated and beautiful things on the next class. The only thing I wanted to show you is how you can immediately start to work with nicaf 8. Nightcafe: This is the main page of nap. This is what you see when you enter the first time. NaF is an R generator that uses different models like Di and stable diffusion. I prefer to make journey for three reasons, the privacy. You don't have to share your work if you don't want to and if you are going to use AR generator for work. This is very important. You can decide to make your work public or not. In the journey, everything is public. The second reason is you can use different algorithms, different what they call models. You can use D, you can use stable difficient and they have several model and you can try and the results are going to be different. And to be honest, it's easier to use a lot easier to use that med journey. To start, the first thing you have to do is just log in. You can go there and log in and you can use your Gmail account. No problem. Then you have here text menu that we are going to see later, and here you have work that other artists decided to do public to start, just click read. We are going to in this page, don't worry if it looks complicated. It has a lot more options than NCA but you have this little interrogations where you can have information and don't worry about this too much right now, but you can always go and read and have information and he for everything. The main things I wanted to see now is that you have a space to put your test pront and you have these what is called advanced pront editors, where you can choose what effects, I think, under calling. You are going to use, but don't worry too much right now. You have it there. We're going to use it a later. Right now, just to see that you have that oceans and you can change model here. But for a star, we're going to use the by default ones. That is this stable diffusion dx L one and our text pront T star, we're going to use a basic basic text pronts going to be a stone castle. That's the only thing we're going to say here. We click and create, as always, I'm going to accelerate the process, and this is the stalks. As you can see, is very basic. Let's see what this advanced pro editor does. We are going to sell it, for example, striking. As you can see there, what it does is it adds word to your prompt. Just with that, look at the difference between the two castles. You have a lot image. Than without them. And the fun parts that you can try different one, like for example, the Oigami one. And you have a castle in stone color, but in paper. So it is good that you try this because you can learn a lot. You don't have to only use them. You can see what they say to learn and try to get your creation, your illustrations to the next level. And as you can see this looks like impression is castle, I think works very well. And then we are going to see a bit more advanced options. If you don't see, just click there. Let's change the advanced prompt editor, so we will have something different and you just select what you put delete and click on the one, and we're going to try with di this time. Here a number of images. What you are telling the editor, the generator is that you want more than one image each time. As you remember we compiled, you cannot say this. You can say several images, but he's going to decide which one, how many You have a lot more control, and this is very important. You have a lot more control with CAF with this genitive art that you have with copilot. The images are bigger as well, of course. Here we're going to choose four as number of images. You can go up to 16. Then here in ast aspect ratio, we have by fit one by one, but we are going to do nine by 16 because it's great to use on social media, and of course, it is good as well to do as a background for your form. With these changes, we just click Create. And here you have the first image is always a combination of the other images, but you can click and look at that. I see each image separately, and each of these images are one creation. As you can see, it has nothing to do with the previous castles. It is really fun. I really like this one, and it has the nine by 16 expiration. Now, once you like one of them, you can click here on vale and use that image, that one that you like as a start image. From here, every change you do, every time you click Create until you delete this start image, the next creation is going to be based on this first image, so this is great. You can you can evolve literally, evolve your image, fix mistakes in a quick way and tweak your image because with this, what you have is an AI that has memory. Guess your star image and you can evolve it. In this case, instead of asking copilot, add a bridge. Here what we do is we change the prompt, so we add a bridge and a moot. Remember as always. AI generation is not perfect. Even if you say I want a mode, it doesn't mean you are going to have a mode. And here you have, this is a star meg where you're starting with and it's going to modify it to have a bridge and hopefully a mode. As you can see, you have a bridge. You don't exactly have a mode but I have to say they're great. Make me think of super Mario Bros and that kind of a static. You can think that the mode is made of sweet of camel. In this case, you have a really caramel mode. Here you have another like caramel liquid but they're not of course, made of water or what you perhaps think. Another thing that you can do to try and change is to change your model. In this case, as I said, we are using the SDX L one that is free. But You can use Dali or you can use others. The thing is, some of them they say P and you can only use pro if you are a pro user. To be honest, you can do a lot without paying. I do all of my work in NICA f, and I don't pay for it. And I have very good results most of the time. So let's try with Juggernaut. Juggernaut is one of my favorite models. I really like the effects. It has a fantastic point that I really enjoyed. This is the difference. We have the mode, we have a great bridge. We have floating candies. I think these are great castles. And grade images generated. Again, imagine how many hours you will have to spend to create something like that. Now that we know the basics, let's explore a bit. You have that menu there of my creation which you can see is, of course, your creations. In my models, we are going to talk about this later are models that you can create My collections collections that you can do with different of your art, and this is a great part and explore, you can see what other people is doing and they have decided to make this public. You can choose if made this or not public, but this is public, and you can see all the data of how they made this image. You can get a lot of inspiration. Then you can go to the challenge and cafe Cafe is not exactly free. It works with credits. But you can have five credits for free daily just for visiting the page. If you vote on the challenges, you have two extra credits for free, it's very easy to have seven credits daily and you can always participate in this daily challenge. I do. I've earned credits this way before. It's a great way to learn how to use this program. And here challenge. Desk click enter and choose one of your creations and enter the challenge that way, is free to do people vote so you know where you are. And here you have the credits that you have well, a lot of options that we are going to see F that's the introduction of Calas 9. What makes a good prompt: Let's talk a bit more about prompts and what it does what it makes a good prompt. Here we have the description of what I want. I want an illustration and what I want a red fence. Well, this is a red fence. But it's not a good image. It has some errors on the body. Is ternar errors. It's a bit weird and I think we can do it better. Remember what I told you about what Max could prompt. We are going to add a few things. We have what do we want? That is a illustration. I said of what that in this case is a red finace but we can add more details. For example, I'm going to add golden wings or golden feathers. I think golden feathers, I like to be more to give it more depth to the illustration. I want it flying on the sky. L, this is already better. When you give you don't want a really long prompt, but you need a prompt that has more details than the problems that we use in copilot. Now we want the style of the illustration. I'm going to be a bit risky here and I'm going to use per realistic. So it's not bad, but as you can see the star has having a bit of a problem on the leg, has it has like a third win, but I'm going to keep going. And I'm going to say, I'm going to add a bit of a composition or camera angle. And this case, I'm going to add bolmetric lighting. Yes, I know it has we are leg. We can fix this later. This problems come because I've said it realistic and it's an illustration. There is a bit of a battle there and this kind of no coherent description or problems can create problems. But what I wanted to show you here is a quick way to do a good problem. What do you want? Of what do you want it. Add details. Describe it like the style or the colors and give it a bit of a photo definition. Well, if you will do this from a camera perspective, where will the camera be or lighting, all these details matter. I think it's a very cool finish. Although I promised I'm going to show you how to fix that combination of leg wing. Let's keep going. 10. Stock images: One of the advantages that you have when you learn to use generative AI is that you can create your own stock images and your own mocaps as you probably know, those are difficult to find if you want something very precise and many times very expensive. Let's see how I do this. I'm going to start with this text prompt. I'm going to specify that I want a photograph because I want something realistic. As always, description and adjectives are your friends. I'm saying that I want a feel of dewy snow drops and a sunny winter day. Then I'm adding realistic whimsical and these last two words are important. I'm saying ground level shot. It is good that you learn the cinematic and photographic technical words that you know what is ground level shot or a bird size shot or a close up because this gives the AI precise information of exactly what you do, and the results are going to be better for that. Let's try and see what we have we are going to use the advanced options in this case, and I'm going to change for an ratio aspect ratio of four fires because I want a stock photo a long as one. It's not bad, I have to say, It is exactly where I ask for. They are perhaps a bit plasticy. I can do several things here. But what I'm going to try first is to increase the run time. Once I like an image, especially on a photo. If you like it, just give the machine a longer run time and here. The difference is astonishing. The light is better, the shot is more beautiful. The only thing you have to do is give the machine a bit more time. Of course, you know that you can use the advanced prompt editor and you can use words that's like DSLR. But sometimes just giving a bit of a longer run time is going to be enough. Now we are going to try to do stop photos but that we can use for MCPs. So the prompt that I use here is, again, I'm saying a photograph of Boho and Co master bedroom. Here are the description and the jectives again. I'm adding the words with one blank frame in the middle of the room. That blank frame is what I'm going to use later to put my own art so I can use these photos as a mocap. It's not only that you can use them, you can sell them online, good photos and usually very on demand. So let's see what the generator does. Oh, it's not bad. It has clearly two mistakes. The first one is the image is not blank. I'm not worried about that because we have white on the outline, we have in the frame, it's going to be an editor, you just put a image on top, so I'm not worried about the images. I'm a bit more worried about the bench on the feet of the bed that is missing half the bench. Of course you can fix that with an editor, but we can probably do better. One thing that sometimes it is a silly thing, but If you are you commit a grammatical error error when you are doing your prom, that is going to affect the quality. I swear to you, this is going to affect the quality of the limit. It's going to fix that and try again. Of course, you can try again and again and again there. This is a lot better. Still we don't have this blank image, but still I'm not worried about it, and this is a beautiful photo, very stylish. And you can use this on any web page. The beauty of this is that you can do it again and again, or you can change the style. We try again this time, I like this one as well. This time it has changed, as you can see, the focus where the camera is and I'm trying to get here it is. If you keep going and perhaps eliminate some phrases here and there, you can have the blank frame. Remember that what you put at the beginning of the prompt is going to be more important than you put on the back of the prompt. As you can see, the Bojan is less prominent here than it was on the first one because here is on the end of the prot, and the other was on the beginning of the prompt. And once you have something that you're happy with, you can say, now do a living room. Now do a kitchen. Now I'm going to change the style. You can repeat this process once and again and you will have one photo and another another that you can use or you can sell. Again, looking in less than 4 minutes, we have six images that are really, really good to use. The only thing you have to do to change the style is delete Boho and put, for example, classic a temporal and here you have. Here, the blank and the white dominates. Everything is this white color, and you can say, I prefer something more colorful, you can try until you have the img that you want. 11. Unique images: It is very interesting that you can do a lot of different things when you're using generative AI. Sometimes we only think of illustrations or photographs. But the quantity of original things that you can do is really big. We are going to have a bit of fun and we're going to create a bit of different things this time. Let's start with a folded paper illustration of a beautiful sunset over a mountain range on a lake. This is very simple. I'm not getting a lot of information there. Folder paper illustration. But let's see what Cafe, I'm using as the Excel one as default. Here you have. It is beautiful one, but it looks like an illustration in not really paper craft. Let's see how we can improve this. We are going to use folded and punched paper background. Then vibrant colors. Remember that it is very important to repeat and evolve your images. That's better. That has a bit of a paper feeling, doesn't it. But I'm wondering if there is another model that will work better with this. Let's try Dali. I think we haven't tried Dali before. I would like to show you different models and Dali of course, the open AI one. Not very impressed. I have to say that since I recorded this class, Dali is available and it is a lot better. In this case, I think that perhaps something like crystal clear that it is specialized in photo realism, could be a good option. Remember that you have community models as well. There are pro for people who pays, but you can use a few of them, you don't have the model tokens, but you can use some of them once or twice. To try them. There it is. I think that as you can see, it really looks, especially the sky, it really looks made on paper. Now we're going to try something new. In this case, we are going to try to do a model of a BC Victorian three D city. Volumetric light is another photographic concept. It has to do with this a bit of magical light. You can read about it. I like the model, but I think it's a bit dull. I would like to have more color. I just added colorful This is good. The volumetric light is the one that creates a bit of a blur sometimes. Let's add ground level. Let's add ground level because I think I want to feel like I'm in the city. It's not bad. But I don't think it's getting the ground level that I was looking for. That's because I just said ground level. I didn't say ground level shot. Remember, you have to be precise and there it is. You have a colorful, it's clearly a model and it is ground level or as close as you can get probably in a model. Now, we're going to try another thing. In this case, we're going to try this kind of embroidery. The Sardou I'm sorry, I'm probably killing the pronunciation of that word, but is this kind of three the embroidery. That is very beautiful, and you can see that I just said embrosry of birds with flower beds, and gold thread. But I want it closer. I just a close up v on the same prompt. And he is. As you can see what you can create with AI is not only photos, not only iustrations. You can create almost anything you can imagine. The last one, just a fell one. Let's create a low polymeric ommeri medieval stone castle. Getting back to the stone castles, because it's ommetrics like the old video games. I really like this one. As you can see, you can have a lot of fun with generative AI, and I really recommend that you try and try new things and have fun. 12. Advance techniques: Colouring page: This case, we are going to do something different that is create a coloring page. I'm going to stop here so you can see the prompt with me. The first thing, if you want to create a coloring page, you have to use that first quote that I'm using here. Clean page from a coloring book for adults or for kids. That's your decision depending on when you do. But that combination of words for the starting of the prom is what works best for me after several tries. Then what you want and you have to say as always of what do you want it? We're going to with the fantasy theme. We're going to say of a medieval castle, you can put here whatever you want. It can be a skull, it can be a dinos dinosaur, it can be anything. Fantasy inspire, we are starting to talk about the style here, and then this block of words after fantasy inspire, you have to use as well. This combination of words between clean page from a coloring book and then the rest of the prompt is what you have to keep. The words are black and white, high resolution, clear image, white background, ink outlines. I think all of them makes sense, and the combination of clean page from a coloring book, of blah, blah, blah. With black and white high resolution, clear image, white background in lives, outlines. It works very well. Let's give some time and there you have it. This is the first page that we have. I like it. I think I would like something a bit more ertalpire. But as you can see, it's perfectly usable coloring page. Of course, you can go to editor and make the whites and the blacks. Sometimes I do it with affinity designer. Playing with the levels, let me know if your interest in learning how to do this and I can add a small extra class here. But I think it's outside of theide of this class teaching you how to do this, but it's very easy. But to be honest, most of the time I use the pages like this because if you are using pencils, it's okay. You won't see any of the shadows, but if you are using a pain that is a bit transparent, you will have the shadows in your final drying and it works very well. So Now, let's make sure that you are in advance mode here. Let's go to Advance prompt editor because I want to add different style. But I don't want to use the present here, I want to go to modifiers. You can spend ten, 15 minutes reading through all the modifiers you have here. They are very useful, I'm going to use the story book illustration because it matches the style I'm going for. If you're doing coloring book, I always be conscious of what you use as your modifiers. Here we have what I was looking for. A really cool fairy tale inspire drawing. You can use any of those as your coloring page. Now that we know that these words work, what I'm going to do is teach you how to say them. You go to Adams prompt editor, and as you can see here, You have the prompt on top of everything. You can clear safe and this prompt is going to be one of your prompt editors, and you're going to find it in favorites. We reset, we go to advanced prom editor his on my favorites. If you click on it, you will have all those walls and you don't have to do a medial cast. You can do a fire house, but the, the technical part of the prompt, you can save. And here you have, you have another coloring page, even has a castle on the background, you have a nuclein page and of course, I kept the fantasy inspire, but you can delete that, you can delete the part of a fire house fantasy inspire, remove that change for anything and you will have how to do a coloring page. When you are working with copilot, you know that the maximum size is 1024 by 1024. But NCapht doesn't have that limitation. When you are doing tries trying to get to the image an image you like. I recommend that you use the smallest size possible because it's cheaper in credits. You just go there and use as many images have you need. When you like one, if you want to print it or you want to use it in a bigger size, you go there to enhance your creation. You can choose all those sizes. You have the credit price there. And just click click, you won't see any difference because the image image is the same. But if you put roll scrolling front of the image and go to download here you will have it. You download the image and you opening in your computer, you will see that it is a lot bigger, it's 8,000 by th with 8,000 by you can pret anything. You can do a poster, you can do fabric, you can do a lot of things with it. Of course, you can do a book of coloring pages. I hope you enjoy this 13. Fixing mistakes: So, in this case, I'm going to show you how to fix some of the classic mistakes that AI makes. Those are extra limbs, extra holes in this case, extra fingers, badly or we are anatomy. For that, what we use is a negative prompt. So let's start what I did here. I just created with a model of d x L one, the default one. I used the pressed vibrant because I wanted something full of color, and the pro was a simple one, as you can see, it was an illustration. I didn't want a photo, I wanted an illustration of a blue unicorn with a bright blue man in the style of extract expressionism. You have this pretty cool images of unicorns. But as you can see, several of them have more than one horn, one of them have five legs. The only one is anatomically correct. It is two blue is the one that I like the least. What we're going to do for six d? For that, we are going to use an advanced prom, that is a negative prom. Here is five legs, quite funny. Here you have that you have three horns. First, we're going to choose the one that I like the most and for that, we'll see the images. This is the one that is too blue. It's not about one, but it's not the colorful one that I was looking for. Let's work with this one. We click on evolve. With that, as you know, is going to use this image as a starter point. We go to the text prong, and we add extra horns -1.5. 1205 is the maximum that you can do. We are going to the maximum, don't do it more than that is not going to work. Is that simple? You say what you don't want and you do a negative prompt. This is the inline way to do it. There are other ways a bit more complicated. We're going to use it in the future. But with that, You will have as you can see, again, images, four images because we have number of images in four four images of four unicorns and each of them has only one horn. As you can see, some of them have some mistakes. This one has an extra leg. But what you can do is do exactly the same with extra legs, extra fingers out of focus. These kind of mistakes that you can see, this is the way to fix them. 14. Masks and more: In this class, we are going to talk about working a prompt mask. For that, we are going to start with a very simple prompt is the illustration of a fairy with wings, forest background, golden hour natural light. What I'm looking for is classical fairy with a key fairy. This is a beautiful one. I like it, but it's not what I'm looking for. In this case, we have to evolve our prompt. I'm going to add a few things here. I want Scala wings. There are very specific transparent wings, and let's try with mushroom hat. I want that whimsical ferry. That's closer of what I want. We have the mushroom hat, which I really love. We have the wings, perhaps a bit small that I was looking for. Let's work a bit more because what I'm looking for is a keyed kind ferry. Let's try that. As you can see what I'm doing here is a that looks a lot better. That's really close of what I was looking for. But look, we have one, two, three, four, five, six fingers, that's a problem. That's a problem where we are going to find a lot. I'm going to show you, I'm not going to totally fix it because it's going to take a lot of time, but I'm going to show you the steps that you have to try to fix it. The first one, this is with the dragons extra fingers in a negative one and ban hands -1.5, and we create. Yes, we have the correct number of fingers now, but there are these long bony hands a bit, I don't like them. Those are called three fingers. We are going to do a negative prom for three hands or three fingers and you can look for mistakes of AI on Internet to know how to call these kind of things. And look, no more hands, there are three parts, but they are not perfect either. So how can we fix this? The first thing that I would do is to get that let's start with Happy. Sometimes that will fix things. But we're going to do here is to get these negative problems and put it higher because remember the things that are higher are more important when the AI is deciding what to do. I got this and I put it in higher than the background because I don't care if the background that much. Look, those has a lot. So if I want to improve it because there are still a bit wear. What I'm going to do is to evolve this image, clicking on that button there. What it does is this uses this image as a star image. Now we are going to make modifications and this image. They are going to be few changes of this image, and we are going to go there and make a mask. Let me show you. To make the mask, we go to star image settings. Here, you have the settings and you have as you can see, several options. You have this control net that helps you to make postures or the shape or something. But this is not what we want now. What we want is to use the non option and use a mask. We click on create mask. What a mask is What would you do with a mask you cover or you select one part of your image, and AI is going to change only that part of this image. In this case, NCafth a red mask, other programs use other colors, black, white is usually the colors that you use of mask, but it's very visible, Nica uses red. You don't have to be incredibly precise in this case, and you just select here the hands. You can select the whole arms, vehicle only to select the hands here, and then you create a pro only for those hands. You don't see illustration, you have to click and done, if not ph you have done anything. Here, you have the black area. That is the mask. And now instead of using the prom that jewelry, have the little that save it just control C to copy it and put nice key hands. Do a prompt only for your hands. That's always try to be precise. There are key hands. But they're not in a great posture. They're like mix like they both hands are together. Let's try the gesture and put in a amaster gesture. Look very soful What happened here? You are forcing AI to create something that is we are or it is impossible. I tried to see if there was a problem or how everything was written. But because the hands were like twisted to be intertwined and I force it now to be upright. The image, the AI can't create something that makes sense. You have this awful thing. You can change things, but you have to be coherent with what you have before. Let's try with intertwined to check how to write it, intertwined there and see if we improve. I'm not going to do the whole process. Look, this is a lot better. You have a nice pair of hands that are correct. It's not exactly the gesture I was looking for, but you can keep going until you have. I don't want to bore you with another five tries, but you can try as many times as you need until you have exactly what you want. I like this one. I think it was very well. It is a bit of a nervous ski gesture, I think it was very much. We're going to let it here. 15. Bonus: Patterns: An interesting thing that you can do, but it's not perfect on generative AI are patterns. To create a pattern, you have to ask You generator, your pro has to say something like fairy and mushroom, cottage core, seamless pattern. The trick here is to add seamless pattern. But as you can see now or as you will be able to see, AI creates a very good pattern, but it's not seamless. What I did here and this very quick is to put the pattern created this bigger with a repetition of two by two, and you can see on the seams that the pattern is not perfect. But in reality, it is very easy with an addition program to fix this problem. For that, what we are going to do is just delete the saps that are on the sam all of them, and then copy some of them and paste them. So you don't have the empty seams. Let's accelerate the progress. Although the image or the way to create patterns is not perfect. It has first, it is a green inspiration point and second with really not a lot of job, you can do a very complex pattern. Here, if you you see what I was talking about, we have removed everything that was touching or over the seams you have the green space there white space, but it is very easy to just fill this space and I have a whole class about about this in skill share and you can check it out. But I'm going to do something quick, you have a bit of knowledge about how these programs work is very easy just copy and paste and move around. But once you have this that it is created with AI, creating a new pattern is very easy. What I'm going to do is I'm going to rasterize, so I can very easily select one of the mushrooms and I'm going to just select this big mushroom. Don't have to be really precise, select the mushroom. Then what you can do is just copy the selection and paste it. What we want to do is just put it on top is put it you can of course the size. If you put it like there is going to use that space that is empty right now. And you can keep doing this and you can move it over the corner or on the top, and then you just copy it on the opposite side. There it looks good. Because it doesn't go over the space, I don't have to do anything special. But here, Now, what I'm going to do is if you see it it goes over the top or over the image on the bottom. I copy it and I just move it the length of the image, in this case is 21204, and I just move it up and down the longititude and width of this image. You keep doing this until you have, and I'm going to accelerate again, I just selate copy copy again and move it until you have a full pattern like this. This is just a bonus class, but I wanted to show you very quickly how you can use generative AI to less than 10 minutes. It took me 10 minutes to do this. You can have a very complex and perfect printing fabrics or wallpaper or graphic designs. You can do endless things with this kind of patterns, and it is a good inspiration also. Okay. 16. Thank you!: Okay. And that's it. I hope you like this class. In this class, you've learned how to use two different AA generators, copy from Microsoft and NICA f. You've also learned when to choose one or the other depending on your needs. You've learned how to make stickers, how to go beyond the simple illustrations, to be honest. You've also learned, and this is the more important thing, most important thing you've learned. You've learned what makes a good prompt. You can use this not only with Naly with copilot, but with any AI generator that exists not or will exist in the future. Finally, you've learned how to fix the mistakes that sometimes AI makes like estra limbs or too many fingers. If you like this class, please leave some feedback. It will help me greatly, and it will help other people to find this class. If you like how I teach, please consider follow me here and why not watch any other of my classes. I have two great classes about Mandalas and I have classes about affinity check me out. That's it. Please remember to leave your project below and I hope to see you in the next one.