Дорожная карта развития ИИ-художника | запись от 06.02.26

Skidych. Hi guys. Hello, Hello, hello. I can already see the comments. You will be with me appear, I will see yours comments. Now I have autofocus I'll turn it off, so that it doesn't jump around for me. Moment. Nice. So, well, hello, comrades. Today we have an amazing day because that I will try in an hour and a half get it done, conduct a master class. It's hard to call this a master class. This there will be pain. It will be pain and suffering. Because today you and I will be analyze the roadmap of any and artist in the twenty-sixth year. If you are just starting your journey in and industry, I certainly don't envy you, because you're probably years late for this, well, for about three years, probably Exactly. If you previously did not touch upon any conditional medical duty, if your creative path is not If it started with the medical staff, it was a real pain. If you haven't installed any Coffee UI, it yet automatic 111 if you didn't put no generated images through stable diffusion on my very old and slow video card if you don't upgraded your PC in order to generate nudity, It's sad, but true In relation to you, there is something to envy. I envy you because you don't have it now we need to deal with all these complex issues settings of local neural networks. Of course, if you want to do it dive deeper if you Do you want to study this topic in more detail? on a technical level, then, of course, go ahead. I have my fate turned out this way, that for the sake of neural networks, uh, I built my own PC. I work on a Mac. I am a photographer first and foremost. Now I press the button and get there some pictures. But for the sake of the sake I assembled my first neural network life of a manly PC. Imagine, I don’t know Why do this to play toys. So, I don't use these neural networks, I use cloud neural networks, but about Today we'll talk about everything order. So, guys, I'm putting timer for 45 minutes because I want I need to conduct this master class so that I can then, uh, you didn't write in the comments, like, come on, finish it already. That's why we move quickly around you our bridgehead. Let's start with, uh, analysis of our roadmap. What is this? such, what does it consist of? We will gradually figure it out. If you have some questions will arise, you will answer them You will be able to ask me. It's not a fact that I will I'll have time to read it because there are messages so many. OK. But if something is right this is really, really interesting and something serious will come into your head come, please write capslocks, Call me on my home phone. In general, I will try to answer. OK. Okay, let's go. Development Roadmap and artists in the twenty-sixth year. Today we'll try everything straight structurally put it on the shelves. There won't be any today no tutorials. I won't tell you talk about how to generate images, Video on how to make different effects using Okieframe video cuttings. I won't show you how you need to glue two clips together, because it's all routine. We have a routine We'll leave it to you for later. Routine is this knowledge of tools is a skill use photoshop there, you know, What button can you use to invert? mask, know which button you can press group the new compound clip into Final C. That's all. This is not today, because it's routine. We are not We'll make it in time, but today I'll give you structure. Now if you come now and you don't even know where to start, and in general, in what direction should we move? Which neural network should I choose? look and in general why there are so many of them, today we are talking exactly about this. Ah, here we go. Let's start with Let's get to know each other quickly. If anyone doesn't know, my name is Maxim Gustarev. I photographer. First of all, I photographer. I take pictures of women's portraits. I really like women's portraits. Perhaps, in parallel, you know me from my YouTube channel. Me too I run a YouTube channel about photography, I'll tell you what buttons are on What lenses should I press on the camera? exist, like, uh, in principle conduct photo shoots. Everything like that. The whole industry is tied to me. YouTube channel. I like the business. AND if you think that neural networks have appeared Only now, in fact, neural networks appeared a long time ago. That's exactly mine photography, uh, you could say, is a kind of sublimation of neural network ideas. AND I made neural networks back when they were not mainstream. And even in far away, in the distance, very far away In 2014 I probably made mine the first photo that uh sent me on the path of artistry and some non-standard, serious, interesting, absurd. For some people this may be the stupidest thing ideas, but these ideas are so important to me I liked that I started them develop, I started to expand them, I started come up with various shooting techniques, processing, and in As a result, it all turned into what I have now. Photography is mine love exactly with its use in not some simple variants of execution. Then I called his photograph neohyperrealistic genre portrait. Lisa, wow, this is it. she. And uh, gradually, I uh I realized that it wasn't just one the photo may be in raw form beautiful. I started integrating photo, here are some non-standard ones processing elements, e- various, created loot, actions, presets, Photoshopped, composited, added additional layers and so on. And here it is all this led me to the fact that when neural networks appeared, my brain just exploded. I wouldn't understand, yes, why I used to spend a billion hours on this. TO for example, on this photo, so that I can draw angel wings, I really can spent 6 hours to throw it on textures to add effects glow to throw on all the glitter and glow lights, draw highlights drawing on the body of the model of light and shadow, on chest, on the collarbones, on the neck, so that make her a little more serious emotions, bring her eyebrows together. That's what I'm talking about spent 5 hours of time. Now it is possible make it using the same nanoban or through medical departments. This can be done in seconds. Just imagine, this is how it is for me It might be a shame now that many photographers, many artists can be left without work. About this today Let's talk too. So, now I'm on I use aerial nets. Of course I chase them not to the fullest. Of course I do, ah, I tried once, but it wasn't even that What I tried, in a test format, I launched various projects the use of neural networks in creativity. Eh, I’ll also show you all this a little later, I'll tell you. In general, as you already understood, I First and foremost, a photographer. One project I did it purely on the second account. As a test option I started post neural network portraits with made, generated in the medical journal. They flew in. That's what I'm getting at now. little spoiler that you can thanks to the ultra-low entry threshold, really mega-low entry threshold, one text prompt through Medzhurni already give yourself a very good fast boost in development in this entire neuroenvironment. People for now, let's do it this way, people haven't developed it yet wild disgust of all this and art. We can still get pregnant, If you want, you can fly into the womb. I already have it, I'm already an adult Uncle, I am already a fully formed organism, so to speak. I'm not into the neural network environment. I'm flying so hard. I just don't have it this goal. But there are some moments I test it and immediately give up the whole thing. For example, here is the second account again Max reef, you can come and take a look at him, you will see that there is a huge the number of works that took off. AND if you continue to do the same thing, just churning out this constant routine content creation, then you can go to very good turnover, for views, for popularity and so on. And such There are accounts, I will definitely talk about them I'll tell you today. So, uh, I think, met. I'm a photographer, click on button, I make cute butt waists, people's faces. Today we are talking about neural networks. And first we need to answer a very, very important question. A neural networks? What is this? This tool, is this art? Or is this a fierce enemy? You can now Write directly what you think. This an instrument is an art or it is a truly fierce enemy. I'm sure that many photographers, tell me too You can simply write to us one word: what do you do for a living, that's where you're from you came, this is what you are doing here, Actually, how did you get into this? master class. You didn't just go through it just by and they were like: "Oops, a master class. " Most likely you have seen e somewhere some post, inspired by neural networks art and they're like, "Obana, I want the same thing the most important thing to do. I'm motivated. I want a lot of money right away. And we are like this We start googling and seem to find something AI syntax, there's an ad somewhere, maybe the targeted one came out be recommended somewhere by someone I saw, I don't know, my video popped up there on YouTube, you watched it, and you're like: "Obanki, well cool, that's it, here I am inspired. Now I'll cut down a lot "Bubble. " And we start right away. get motivated. And now here's the next one Our next step will be precisely about motivation. Here he disappeared here a little lower. Yes, well done, guys. Yes, of course, everything right. RAM has become more expensive. RAM have become more expensive. Brutal. I managed to buy it RAM, you won't believe it. 32 GB me bought for 160 euros. Oh, I'm so lucky. Now it costs over 300. Of course, neuronet is elementary, the simplest tool. This is absolutely not enemy. If you are any kind of artist, if you are, for example, a cover illustrator for books, then neural networks are a kind catalyst. They just allow yours ideas to transfer from your brain into a format 2D image in a much shorter time period of time. If earlier from you it was required, but not just the basic ones knowledge, art, understanding of light and shadow, understanding how the brush applies brushstrokes on paper, understanding of color, understanding of composition, then now you much less of the applied kind is needed experience of holding a brush in your hand. To you enough, you still need knowledge composition, color, understanding content, uh, some kind of drawings volume, of all this. you need to know, but you are much easier in this regard information. You just need to know less. Somehow everything has already been sublimated. The neural network is simply your brush. She does motor skills for you. Ah, but that, Whatever you ask her to do, it's up to you. a new profession is emerging, yes, we have Promt Engineering has appeared. Respectively, This is absolutely not an enemy. fierce enemy maybe for those people who immediately they give up, start to be lazy and we really lived there our whole lives, earned their bread by their e routine, this conveyor belt similar works. Now, yes, there is some layers of professions that, most likely, they will gradually fall off because they start be displaced by new professions. In this There is nothing to worry about, it's absolutely Fine. We need to understand that this is just a tool. I didn't come here for nothing. I attached this work. Very interesting case. There in the twenty-second Jason Allen from Colorado, he won the photo contest. And at a photo competition or not a photo contest, competition art of art, maybe there, I think there were paintings there. In general, he made the image using Midzhurni, another antediluvian version that was glitchy, I drew all this incomprehensible stuff with abstract strokes. and he won this one contest. And one of the artists wrote, that you and I are witnessing death artistry. The Death of Artistry unfolds before our eyes. Creative people, unprotected from machines, jobs are at risk. AND many stood up then for his support. Many people thought so, many people talked about it, thought that Now we will be replaced by wild machines. But, Guys, again, what's going on now? How many professions have disappeared? How many people left without work? The same amount people found new options for themselves -e of its own implementation. That is, press Cool, why not? We can do this too to use with you. That's why I personally don't I think that Naroseti is a fierce enemy. I don't think it's art at all no way. Neural network is exceptional tool. since it can't be art, because art is You. This is what you are doing, it exists. a huge amount components of your work. Your job maybe, sorry, I'm knocking on microphone. Your job, it doesn't consist from one image. And if this some kind of movie clip, advertisement, there is a superposition of video images, music, content, plot, It is clear that art is the final result. But the neural network can, one neural network creates a video, another the neural network can create a sound sequence, the third neural network can cut this video. Eh, you're most likely like promt Engineer, can you give me the correct instructions? give it all time to be cut up and with adequate generation was generated adequately. Then you will need all this. edit, do light correction, then show it all, throw it online, and then it will be art. If you I don't know if you'll come with this art, post it somewhere on Instagram, The question will still be: will they see yours? art other people? If they don't they will see, then it turns out that the Neural Network essentially did a wasted job and no one will ever know about this, and your work will not become art. Art is just what one sees others, what is it like some value. And the value needs to be in First of all, convey. That's why for us The most important thing is not the network bunks - this is tool. It is with these tools We will work with you. Let's move on to the next one. And how will we be? motivated by all these neuroinstruments? Now we are with you we will go through the classics stages, uh, of acceptance of denial, anger, bargaining and so on. For starters, rather in total, again, you've already told me a lot there wrote everything. Someone wrote that postman man. I hope you have bike. salesman 16 years old. I want creativity. Someone is a photographer. Even Evgeny Bogachkin wrote: "Maxim, I'll ask you right now the main question for me. Me too photographer. Tell me which one is best synthesize a photograph of a model. It's called eh avatar, huh? That is, if we want make an avatar, yes, actually it is possible make lore through different neural networks. You can do ENT with your own through Flux character and beat him up. It is possible through Medzhura can't make an avatar one by one photos. It is possible through nanoban. Nanobana redraws well. Now The funny thing is that it's a set Neural networks can solve your problems. If you really push yourself, you can install a local neuron, m, some kind of Ponyrealism on SD Excel and do everything locally. But that's all deepening. We are in these recesses now Let's not interfere, guys. This is all no use to us. it's so important because that's what we're we'll waste so much time with you. OK? And motivation, Where are you from, I'll repeat myself, where are you from? where did you come from? You are on this The master class was attended in order to solve some goal. And the goal appears, when you are inspired. Here you have it some kind of inspiration, you caught it motive, and you wanted to do something. Most likely, you came because We saw these somewhere on the internet values, such numbers. Any cool guy from I know this guy myself. He's on VKontakte, when I was just starting out take pictures in 2010, he has There is a public page on VKontakte Take a photo. And so I go there regularly I uploaded photos through it, you can say, constantly, constantly posted each one Public Photography Day, in the proposal threw it in. They published my publics photos. This is how I grew up audience, created information noise, declared himself. It's the same thing now. You can do it everywhere. It's huge the number of publics that can repost your work wherever you can offer your work. This is all works. So, now this guy makes neural network portraits in midzhurnia. This is purely medical work. Here just pure midzhurni very high quality, cool. And he's going crazy with from morning to evening. And he has huge ones views, and it has a huge audience. There are 50,000 likes on some post. collect. This is all in the future. is monetized. That is, again you, inspired by art, of course, we don't remove the fact that we have motivation not only creativity, we have motivation through monetization. We see the number of likes, we understand that oh, And here we have Hannah Catawba who knows how to sell their neural network prints, and they are bought for real money. Aw, that's cool, huh? Why can't I do this? You can get inspired by this, and it's cool. something Samuel Pogi, who makes awesome arts also from the superposition of neural networks, editing and sound design. You look at this, you see, there are 13 million, oh my god, 15 million views. 15 million He only has one view on odnoms, even if the percentage is interested and 1% from This 1% will buy presets from him, that's all That's several thousand bucks right there You can lift it up on this. And it's real inspires. This is truly inspiring. Then you look like, "Oh my God, Is there any other guy who makes funny videos. " Not just art, but So you saw that it's entertaining. content, it's gaining even more views. Have you heard on the internet: "AI is bad, it's sad, a bunch of brain-diluting content. " In fact, this AI content is so easy it's going so well is moving forward, and in the same way he is monetized. And internal monetization on TikTok, YouTube. All this exists, and for now This still doesn't get blocked. That is, uh, The same TikTok pays out without any problem. money for monetization of AI slontent. Again, the main thing is that the content is there viral. He may be a slop, but he must be viral. And so, I hope I haven't loaded you yet. I'm fine now, Fine. Let's look further. Just for fun, just for fun. I did it myself, uh, guys, I tested different things. I'll tell you more, I even I made AI avatars with girls for onlyfan, but more on that later. A, accordingly, I made a YouTube channel, where I uploaded one video with neural network matrix. I did it cool continuation of the matrix scenes, and He's been on YouTube for a very short time. got 83,000 views there. This is very Good result for the first videos. And again, I'm getting to the point that now This topic is viral, it's very good flies in with cats that shoot from martyrs, with cats that run and they get into people's faces there. Open just, for example, this one Instagram account of this type Valentina, and you will see that it’s a brutal trash, just brutal trash flies well. Accordingly, one of your motivations maybe, maybe one of you came here. You can give me a plus. This is for the sake of the desire to do this fan content with a hint of what's to come monetization. Next up is, uh, AI influencers, AI models. Now they're starting to get tough. easy to use various leather flaws. There's like a model there Svetliga, then she's gaining in general billion views. You saw it 100% These are the kind of advertisements that we have now We will teach you how to create AI influencers. There are already separate courses. All this is here it's already so obvious and jumping, what are you trying to do with this? fight back. And it's very difficult because of this. fight back. Ah, and people make money on this grandmothers. People are making huge profits from this. grandmothers. I just switched from cats, but before I come back again here, quickly, I'll get back to this to the little cat, this cat, With 30,000 likes, he's just sitting there. calmly and looks at the bun. He so cute, he's so pretty. AND there many people write in the comments: "Wow, "What a beautiful cat. " But if you have you ever encountered generation? through a neural network, if you are a neural network artist, let's put it this way, then you will see all the neural network artifacts. You will understand that, well, this was done in a quarrel, yes, there, in the advanced model without any water signs, but it's all the same neural network. But 99. 99% People won't understand that it's a neronka. They watch and it's natural content, even If it's not AI difficulty, if it's not simple there's a mouse running out there, which Shows biceps, it could be realistic content, and people are on it now they do both views and, accordingly, this whole thing monetize. Same thing with these absolutely plastic ladies. 90% of the time I'm just like, "Wow, oh my god, how do you fall for this? men? They are plastic, they are not real men, not just men. Now, If we take three or four months ago, Okay, six months ago, before Vitiligu, and before this maniacal one, just, uh, virality Vtiliga, everyone made girls. Here basic account. You can come in see how much she has now subscribers. I'm interested. I am this one I took the screenshot about a month ago, I guess. three ago. I won't be lazy, I'll open it now and I'll see how much she has now subscribers. This is completely neural network account. 630. Okay. Now she already has 630,000 There were subscribers. 601 630, that is plus She gained 30,000 subscribers. Cool. Here again, who he signs, well, it's obvious that this is absolutely built-in plastic lady. OK. You, of course, go here this link, there is a link to Only Fun and there's chiki-piki. All through some local van with full s a complete lack of censorship is what makes it fools money, and it works. That There is a trend, virality. this very news item that is on this you can make money, she's still it works, but gradually, of course, This should be fixed soon and should be regulators will appear that will do all this will be cleaned out. I think it is will appear. Well, because it's 2025, 2025 year, previous year, this was the last the year when you and I can distinguish real content from neural network. All. twenty-sixth year. That's it, right now. they do things that are hard to tell apart impossible. They connect again, Here's a new trend there on Vitiliga. And no Only catfishers are picking up these trends, but these trends are immediately picked up And representatives of neural networks, those who give working tools. Same Hicksville immediately grabbed this topic and offers a set of different skin tricks for every taste. You can do anything. Accordingly, in addition to girls now the guys who make these girls, We bought out the topic of what can be done and the guys to another part of the audience. After all, no only with us Men pay for female content, but also women for male content. And this is it really, actually, interesting. Personally I'm interested in this because it's, well, something new. Accordingly, motivation. You might have been inspired by this. Well, perhaps if you have already encountered design, if you are a designer there, are you engaged in printing, pre-press preparation, if you are a photographer, then, Perhaps you are interested in both production. And now you came, uh, for just to understand what, in principle, it can be integrated into yours profession. Actually, a lot. AND the first thing we start with, with what, in principle, our creative path begins in the industry - this is our first experience, which is a huge trap. You were inspired, you saw cool things there cases of some guys and they're like: "Let's go, What are we doing? We are opening medical journals. Medzhurni. Why medical duty? Yes because medical staff are a force. From this everyone must begin. You had a master class with Nastya, she told you told a lot about Medzhurni. You must admit, it's cool. That's how much There are cool things to do in Midzhurna. AND We were like, "Yeah, cool, let's go. " They took it photograph, took promt, just him copied, just copied, like this they pressed on it and boom, they did their thing generation. and they're like, "Wow, damn, yeah. Cool, we're with you super professionals. " And we get into this point. I think everyone knows the effect Dunning Kruger, when we are with the hugest the meaning of confidence that you and I super professionals, but without any experience, no knowledge, nothing at all, we think that That's it, here it is, a gold mine. Now we are like We'll start making some really cool stuff together things, and, in principle, this can be done end. Oh, nothing of the sort. Here this decline that you will feel in the near future, he will be catastrophic, because you find, again, some projects, you see, that's it Yes, I don't know, it's in the same place Medzhurne is some interesting protein. You designer, you want to think: "Hey, Now I'll suggest some nearby cafes make a cool neighboring hotel, which is engaged in production jackets, I'll make them like this business card". You find some instance, you find the text prompt that does such a business card. You copy it, change, put some key phrases. Let it be written: "Reds "The jackets are cooler than the green ones. " I've been wearing them for 4 years. one jacket survived 20 plus washes, 30 I glued the guys together in this jacket. You start generating. And what it turns out? But it turns out to be crap. And the first one the stumbling block is getting it complete [ __ ]. And so we were like, "Okay, come on. We'll probably generate it one more time. " Generate again, get more worse. The third time is even worse. And we are like: "It's all medical stuff, it's all rigged. These are the very bastards who rigged it up. there so that we don't get what we need result. " And the joke here is that, unfortunately, the instrument besides the fact that it exists, they also need to use. That's right, they need to be able to use it, and then it begins more fun. Let's watch another one this is an option. Oh, also on the internet. found. Oh, and Kondratyev made a video, cool video that went viral, I hit 40,000 likes. Cool. Everyone likes him. and A and Kondratiev is on site. A and Kondratieva, What would I do now, again, I need to understand where we should go. Every time when we get this feedback from the audience when we, when we were praised, when we were told, that you're a good guy, and Kondratyev, you made a cool video, we, we have a reflex is triggered that we want more of us we also want to receive this praise, we want to make more of these. rollers. and if I were A. I. Kondratiev, I would now I just started churning out these videos, different videos on different topics. I would just start to delve deeper. Why this is great? Well, because that's the topic there. this one uh Pixel Art, this one is eight-bit. graphics, let's not call it that eight-bit graphics, this is pixel Art. So, we're like, "Wow, "Cool, we're going to start doing this now. " And maybe some of you, is engaged in AI design, thinks that: "Well, this is Pixel Art. Cool, you can do it with this in general Pixel art go into Game Def. In terms of content resulting, it's cool. Well, looks quite interesting. Beautiful video clip. uh, somewhere it's even nostalgic pixels are visible. And we think that with this with content we can, in principle, go into GameВ. You can't even imagine how much GameDave requires drawing assets, to make computer games. Uh Many indie developers love Pixel graphics. And so in order to I need to draw some kind of car really spend a huge amount time to pixelate it all draw. If anyone ever he was doing pixel art, he understands, In this case, the neural network does all of this. not for real. If you go to channel to Adam Si Yunis, Yonis Yunis, he from Australia, yes, he is from Australia. You just watch a couple of his videos and you will understand that Pixel Art created by a neural network this is the gap between pixel art, which makes the neural network real pixel art that used in Gamev, which used in games. If you go to echo website. This is a website some kind of aggregator of games that are made community makes community, puts it out into the open, or there sells for money, you will play, just open several games with real Pixel Art, you will understand that the neural network - this is absolutely not what it is it doesn't even come close to that What a real Pixel should look like Art. The neural network is artificial. Naironet he doesn't know how yet, it's very difficult for now Important note, the neural network is not yet knows how to make real pixel art. She, Yes, she's getting closer, she's trying to get him. pretend, but it will still be uh not a real product yet it is still very difficult to integrate into real game. I got high on something one moment in the summer, this summer, I decided make your own game. I'm through I watched some tutorials on the internet for a while. Uh, this is just, you know, the past summer must have been so very swift the growth of neural networks, uh, from the point of view applied visual. What you can to do in fact, there on request do almost anything. So So, I opened it and started watching videos, I started studying this topic, I started studying there is an elementary programming language, on which all this is done. I decided make a game. I started throwing it in and found it there are some sets on the internet, I manually edited it in Photoshop so that it Everything was tiled normally. And then I'm like I think: "Damn, I'm in '25 I live, we are with you modern people, why not use neural networks? " I started try to do all this in a neural network view. And it turned out to be complete crap. Here then came the realization, the pain, the decay, negation. I tried to comprehend it all, What doesn't work is this. You can't just, you can't just take it and open it neural network, any neural network, uh, do good tileset and animate character. In any case, it will be from you learning the basics is required, learning bases, studying not only the most neural networks, but also understanding everything that grows around this neural networks, wok around what you want to create. If you want to create tileset, you need to understand, uh, rigging, you need to understand animation, the character's anatomy, you need to understand coloristics so that all this doesn't stuck together into one purple pancake so that everything is together worked well. It won't help you neural network. Not to sow is just a small tool, and any area you will take it. Editing, creating short videos viral videos for rssiks, for some office that bakes pies. All together it requires from you learning the basics. And it is to these foundations We are slowly coming together now. A It's been 12 minutes already. Oh, we have 12 minutes. left. Let's do it quickly, let's do it quickly. Where are you? Here, here you are. Listen, I've already dragged it out. To make it cool result, you need three components. Just remember, write it down, do it screenshots, keep in mind that three are important to you, extremely important components. This is artistic, technical and talent. Let's start with the first one. Artistic. The artistic component is what you create. This is the idea, the content, motive. What do you want to show? Uh, what? the idea you want to convey to the viewer. If You can write now, you can print letters, just think about it, What do I need to write to you now? analogue of Harry Potter. War and Peace, yes, even a poem which. To do this, you can easily, you have a tool, you have keyboard, you can type letters, but to fit the idea into your text is extremely difficult. And this is not learned, this is not developed. in a couple of hours. This is all, all yours background, this is your whole life, your observation. So the idea, it depends on yours from your experience, from your background. You need to grow, you need this erudition to be able to create cool idea. This post has made it to the top for good views, because here I have a cool idea. Yes, that's the idea, exploiting the nostalgia of the nineties, all these images. If only I were here just, I don't know, some cat, who plays with a ball, maybe it is would be great for those who love cats, which playing with a ball. But here it is audience that viewed this video, uh, and she put likes, she's strict from the Soviet era, which grew up in the nineties. I myself grew up in nineties, and I saw it all, like carpet after carpet was beaten with a carpet beater street, I did it myself. Accordingly, we also add here basic fine arts in artistic component. This understanding of the composition, how everything should be to compose, coloristics, these basic foundations, basic bases, butter oil in any in this case, try to at least somehow look on the Internet. You can write on the internet there is what you need to know for start in drawing. Already this enough. there, coloristics, uh, setting the light. Why am I doing this now? I say, uh, in photographic words? Because that's my background. I'm in principle, I grew up on this. Respectively, This is the artistic component. Next, the technical component. The first thing we'll start with and in general Today, in principle, we will continue - this is and tools. That is, of course, Neural networks are important for us. We need understand what neural networks do for what answer and how to use them. Here too we attribute promting, skill va skill express your thoughts in the form of a prompt. Now, of course, many neural network models, the same Midzhurni, he's quite demanding when it comes to writing. promta. He understands the register, he understands various special characters, he understands the spelling sequence words. But such networks as, for example, Nanobana, you may write with errors spelling. Uh, no network digest your text, she will be able to understand, but in any case it is correct to form your thoughts - this is also very important. That's why you open the internet there. somewhere and look at how to write prompts for different neural networks. Then again following. Classic software, guys. Without him, no matter how much you want him You wanted to, but you won't be able to do it. everything is purely on one neural network. Neural network - it's a tool, but for now this neural network tool, it's crooked, it unfinished. It still needs some improvement many, many years, decades. AND make sure you write and immediately got the perfect result, you can. But if we are talking about serious ones controlled results, again, write a prompt, get the perfect one the result is possible, but it will be a poke a shot in the dark. That is the efficiency percentage here this ideal and perfectionism on output, it is very low. If you want get strictly controlled result, then you most likely you will have to refine what you have created. That is, to do post-production, this is again, without classic software, Sorry, nowhere. Let's talk about this just today. Well, and applied experience for that outside of AI is also welcome. If You are a photographer if you understand how camera angles work better than Wй angle differs from narrowly from any narrow angle, why is the aperture F1 will it blur the background more than F8? This Fine. Just like with coloristics. If you understand why warm colors go well with cold and create a temperature there the volume of the separation of plans is also good. Applied experience outside is very useful thing. If you have never encountered the visual art in my life, but suddenly decided to create and influencer, you do this you can do it. But most likely, uh The AI ​​will take over some of the work. That There is AI that already does a good job of setting up light, in composition, but again get the perfect result, just the very understanding of what you did Okay, he won't come to you without a background. So it would be uh good for you again at least some experience outside of AI. Well, not decades, a little faster. Again However, we are not talking about the end point. That There is a neural network that will be available in 3 years, it will be worse than the neural network, which will be in 10 years. And years, it will be worse than the neural network that will be through 50 years old. That's why Here we are not all relative. OK, let's move on. What neural networks do we have? do they happen? Let's quickly run through this component. Everything depends on us entry threshold. Neural networks can be simple, can be complex. Complex ones are local networks. There is absolutely complete freedom and absolutely complete control. If you come on Haging faces, if you come to Navitai website, look at models, checkpoints, You will be simply amazed by the things that are lying there. And if you also register and Turn off censorship filters, then you have the ears will simply curl into a tube, the eyes they will crawl out from the forehead from what a piece of trash, how simple, How shameful is it that people are doing this? Here. And then it all ends up on onlyfans of all sorts of shame. So, what can be done here? This is absolutely complete control. You can do whatever you want. You Okay, I won't go into details. You can do anything, whatever comes to your mind. E are the bravest fantasies, here you go, go there. There is absolutely no censorship here. Uh, there you can create one character and due to its very good consistency, due to the good consistency of the character, you can create completely different images in different actions, in different movements with the same character. The price of the issue is here too. Interesting. Or you rent some cloud hardware, or you work on your server, you can buy, no I know, computer, put it there several very powerful video cards, which each cost $3,000, and then you will be, well, somewhere on average level of neurogeneration. Then you will be able to generate some there 15-second videos in Full HD resolution. Ah, that's it. That is all that more seriously, the same blade, for example, Blade 3. 0, which came out and is coming soon they'll add it to the syntax, but you won't find anything like that locally you won't be able to do it because it's completely different volumes of data. But to do there's a picture of a lady with nice You can find all the forms here easily do. It's complicated. Why is this difficult? First of all the queue is difficult because from you It requires incredible, vast knowledge. which will simply destroy you. I say this from my own experience. I tried to study, uh, let's not just base, and basics of neural network question through Ko. I cursed everything in the world. from installation models, from LOR settings, from installation these downloads of checkpoints, from various workflow. There are so many of them. And if you don't understand all this, don't worry about it not one day, not 2 days, not a week, you need to spend a lot of money on this time. And every time new ones appear models, you need to keep your hand on pulse. This is something you really need to keep an eye on. If you're strong in technical, uh, system administration if you If you're a good programmer, then you're in this. You'll figure it out. If you are far from programming system administrator administration if you do not technical specialty, then for you it will be an absolute pain because You can't just look it up on the internet tutorials. No, you need to understand that happens, and such tutorials themselves make tutorials. This is extremely difficult. I uh can't recommend you to go there. to enter this sphere because now, and, again, in my opinion, again, this is exclusively mine subjective opinion, but cloudy neural networks, they start a little bit get ahead of local ones. If upon appearance, eh, accordingly, when it appears neural networks, remember, everyone was hyping it up Stable Diffusion, because Stable diffusion just makes pictures, then Jurni appeared, everyone crossed over to us Jurni, they went to the cloud. Ae, local neural networks stopped, they are also constantly evolving. They appear new models, retrained models, more on large volumes of data with better results. There are enthusiasts. Again, if you go to the same Civitai, you You'll see how much content people are creating. There are a lot of people making content, and these People understand this. But it is necessary right now, my whole life dedicate to this because it is extremely difficult. I'm scaring you for a reason because it's really difficult. This is this It's not easy, but you can try for the sake of it experiment. Sit for a couple of days and say, "Oh, come on. I don't care about them, I'm better off as a simple promchit I'll write. " It's complicated, guys. It's simple. ease and accessibility. This is all not the neural networks you see now the Internet, which are accessible through some website. You go to CLink, you go to Runway, Google Veo, etc. further. On GPT chat, you can write texts, he sends a request to the server processes it for you too returns something. We are already working with you, well, I personally I work, I don't know about you, I work in syntax, because for me it became mega convenient. I already, uh, I don't even I remember how long I sat in it, 2 years, I guess that's when the syntax appeared, I said: "local networks, go to "Go for a walk because it makes me feel better. " I wrote a test and got the result. This It's very easy and it's very accessible. This very simple. You can make text requests, you can work with photos, video and sound. Here is a fast and high-quality result, because you delegate all the, uh, mechanical work for large servers, who do it for you. Price the question is the cost of the subscription or tokens you spend on this or that. And now if again the flag into your hands if you try to go along on this path, I will say even more, try this way, test it. But I'm sure if we're talking about creation content, then you will return to this, because these guys, these guys and these Guys, everyone who makes content, they waste their time on not studying technical component, not for study administration of neural networks and their setting, fine-tuning, you can do it there customize anything. They do exactly that content. Here they are doing all this, uh online via cloud services. But if you will take some serious production that is sitting somewhere in Toronto and bangs out Vfix graphics for cartoons, movies, and so on, right there already, and not only do they make do with cloud servers, they sometimes use them local networks. Well, not sometimes, but they constantly use local networks for a little more control. Uh, OK. Where do we begin? I won't tell you because We're just back on the same page now a billion hours of time will pass. We already have 50 minutes have passed. Let me do it quickly I'll tell you this, and then we'll move on. What can we do? in the cloud? What is important for us to do? We must understand that all neural networks, they are created for something. They all are divided into something. They are trying to solve specific tasks. The question arose: what does cloud mean? What does local mean? Cloudy is, Marina, when do you write? in Telegram request: "Generate for me kitty. " and he sends you the finished one video of a generated cat. This it means that when you sent a request to cloud, in a cloud somewhere in California, uh, or the Chinese have huge ones computer rooms where computers are located several million bucks. And all these computers work. at the same time, to make you a cat, who plays with a ball. Brutal. Just imagine, really. the servers are really expensive, they are cost a shitload of money. SSD RAM, video cards have gone up in price now, because that's all Manufacturers make video cards for so that you make a cat that plays with a ball. Well, me too, who shoots from a mortar. This is [ __ ] awful. Here. And these are cloudy power grids. you can't throw it there naked girl because there is there are certain rules, there is a guide, which will not allow you to do nudity. And local networks are those networks that work on your computer. That is you disconnect from the internet, you install a local neural network on your computer, turn off the Internet and you start generating there. Robert Pattison is playing with his Batman. Let's say he plays Batman, and he generates this for you, and there already without censorship. So, uh, you need it for yourself again, understand what is better, What problems, what kind of Nero will solve yours? tasks. If you need Robert Pattison, who plays with Pattison, please, These are only local networks. If you I need a quick, final result, which may not be censored, then for high-quality results and fast The result is a cloud solution. What's in can this be done in cloud solutions? Let's start with text. You have a huge amount different language models. You can create scripts, format, conduct fact-checking. For example, when I I'm writing a script for YouTube, I can take it the whole scenario, throw it into the GPT chat and say: "Please check the facts, everything that I am, my speculations, maybe I am I'm wrong somewhere. " That's it, he's will study it, scan it and maybe tell me to provide specific analytics. You can create different schedules, take care of your body, use a neural network as a translator. That is, all this can be part of yours. work, your final result, content you create. You can create a content plan, do all this through language models. I used sometimes even for writing code. I don't I'm a programmer, I'm not a programmer programming, but GPT chat, or rather cloud, uh, Cloud, I don't remember, in the bathroom, I'm in the bathroom, I think I did, and he helped me write A small script for Photoshop. And in Now I can do it in Photoshop with just one button open open photo, save to multi-format in different formats, in JPEG, PSD, etc. This is all language models. We won't be on them. stop. From the photo, it looks like Nastya to you. told about the art combine. Just the most incredible artillery combine, eh, which allows you to generate completely different images. He has all sorts of outpaints and inpaints. You can draw any object, you can copy the style, you can make some non-standard ones there. aspect ratio, you can do video, you can just open it the same Midzhurni, open the tab styles and see what styles people have created, uploaded to open access. AND you can just take, I don't know, any Prompt, where a cat plays with a ball, or e- that this Robert Pattison has become attached to me for the infection. So, you can open and try in the same style, just by copying the small link, again, Nastya has already told you all this, copy a small link, create in the same style, so that it’s just Here's a photo, give it to me directly I'll make it bigger like this. To do such a thing in order local neural network, You need to go through 10 circles of hell. To you need to configure Laura. You need set up a specific model that was trained on such images. A It's incredibly difficult, just search for it yourself administration. And in Midzhurna you do everything in one Promtom. You just copied it style, this S, RF, tykatyka-tyk, they threw in the numbers and received. You can make avatars by your photo and much more. Medzhurni cool tool. You already know, that he's cool. You can besides to make images, in addition to artistic ones images, medical journals are one of neural networks are probably the leader set up the squash. That's right. Medzhurni is the leader in generation artistic images because she was not trained in the first place simple photographs taken there your dad on his cell phone. They were trained on images -e valuable s point of view of fine art. These are paintings, these are paintings of famous people artists, this is a world heritage fine arts, this is photographs by popular photographers. All it was done without their consent, Of course. But we will skip the legal component. question now. That's why, uh, Midzhur does not give a very cool, an incredibly beautiful picture. Here it is exit There is, let's say if you, well, don't need it, if Do you understand that because medical journals Now it is clear that these are medical units. You can understand that this is exactly it the style of this particular neural network. And at many appear immunity to this. We scroll through the feed and like, "Oh yeah, damn, well what's there, where? I'm sorry, I hope you don't have that in your eyes. "flickers. " And you open it and: "Well, damn, well this is medical journals, come on Let's skip it, we won't like it. " Here it is I already have this feeling of understanding that these are medical journals. This is what the medical department looks like, Yes, it's cool, it's incredible, but gradually it gets boring because there are medical journals your own style that makes you talk about this Midzhurna when you see this situation on the Internet. Eat other neural networks, for example, SDM, which makes photorealistic images, much more photorealistic. again it depends on Promta, depends on the situation, depends on Sida. Sid is not the same seed that you enter it, but just from time to time. You make several generations, some They make absolutely amazing stuff. AND if you watch these, Mizhurni is making some kind of resolution there 2048 on the long side squared, generates a square 2048 along the longest side, and Cdryam was like, "Bam! " 3. 456 at 4. 608 4K3k4 and you just look at this detailing. A Photographers, why the hell do we even need them? are needed if now any brand, any The brand can take on the Dream brand. and generate details like this for a billboard. If anyone doesn't know, the billboards that you see on the street when you drive car, there's a billboard hanging there, he has permission there are 4,500 points along the length, along the long side. 4 reps, in super rare cases, six. That is six is ​​already a lot. six is ​​already He has a lot. Uh, near the billboards, It turns out there are no large billboards there is no point in setting high detail, because the printers that print These billboards, they're not needed. high detail because we never look at billboards with distances of 10 cm. We always look from afar. And when we look from afar, for us angular resolution this billboard, it's not that important. Exactly that's why on big billboards layouts are used. I'm in the reprint I worked in preparation. I did all this. Uh, My God, what a fun life I have. Small resolutions. This one permission, this kind of permission, this is way beyond the pale for that Print out a huge billboard. Accordingly, photographers. A The question is, why do we need them now? if it's possible to take a girl, uh, generate which with which there will be no need to conclude contract, model release, with which not you need to bother cleaning her pimples, which can be viewed from any angle, in any generate clothes immediately and which you don't even have to pay a fee. This is, of course, scary. A little bit Everyone tensed up, photographers. Accordingly, Cdrim is very cool neural network, which is also in the cloud, which can be used. Further nanobana. Nanobana is just trash. Trash, this is complete trash, total crap, holy [ __ ]. Mind blowing, there are no other words, because when you open it, there write something, and the end comes, because It seems that it's already at its limit there is no perfection. But the limit of perfection There is. I'll quickly have a couple of them I'll show you some examples. I have a lot of them here. I somehow just knocked together a lot of stuff I did everything. Here's just about the spriteset. Remember, I told you I was talking about spritesets for Game Dla, when I said that they can't on eroset make a normal spriteset. TO Unfortunately, they can't yet. Same uh spriteset uh will allow you to animate the character frame by frame. Well, how does it work there? classic animation. Here we have it first frame, there is a second frame, leg moves, the third frame is still a leg moves. Now if all this animate, uh, overlay all these pictures on top of each other and then one by one, uh, it’s too bad to show them, then it will turn out to be crap. That is, for now he is this he doesn't know how to do it yet, but he's trying, that one The same Nana banana is trying this to do, but not quite there yet. Again same, the limit of perfection, uh, perfection there is no limit. I'll show you a couple of examples. Here you have it photography, girls. You came to the photo shoot, did one frame, your flash drive is dead, you only have There is one photo. You prescribe promtik, tu-ku-tu-ku-tu-ku-tu, and you get in addition to this photo here's another layout of four frames with different frames. Cool. From a distance, no one can tell the difference from reality. After posting on Instagram, no one even will understand that something is cool here. You can change clothes. You can throw it on additional objects such as wings. You can upload two frames. Here are also two frame, and with neural network wings. Here I have a photo. I uploaded, Promtik wrote: "Make a layout from five different angles. " Bubuhsh. Look, it's cool. You can on Instagram add some too. By the way, it's prohibited neural network. Oh, I'm sorry. Instagram - This is a product of the Mme company, banned in territory of the Russian Federation. Wind it up to the university. You can combine photos, you can upload a bunch of pictures and at the exit get the finished composition. It is possible from limited number of pieces, you can from a large number of pieces. All this works. You can replace the background, you can make some local color corrections. Let's say you took a picture, you have an overexposed background, you can darken it. You can change the back plan, you have a model and bam, it became right in the jungle. You can steal it stylization. If in case of a magazine we copied the stylization of the learned model, you were required to provide a keyword sрф, there's something written there, model, which is trained by other people, then NanoBana can just go and steal it the style of the picture. Here you have the original, second original. And you say: "Make me this one a photograph in the style of the first" and you get this situation. Well, That's cool. A Okay, guys. All the promts, uh, If you want, I'll just take this one the presentation I made. This is Mira, just in case. The application is called Mi. Miro. And I'm in it I'll do all this now. I'll just do it I'll upload a large PDF. You will be there watch because there is a lot here. You can have black and white images make it colorful. You can create covers. Personally, I'm all in the tail and in I use nanobanana for my mane, to make covers for your YouTube. You can restore photos, make them from black and white to color. You can make some comics. This is also, in fact, a fun situation. You just write the equation, write the solution this equation and write the solution on black blackboard in a Soviet school. Crap, he even understands this. That is, nanobana, She not only takes your promt directly in the form of, uh, visualizes, but she also understands what you wrote. She is his digests, she solves it. Not just 5 + 5 / 5 to 3. And he decided correctly, but it is also a triple integral over volume took. Well, that's cool. That's not cool. Cool, Certainly. I double-checked, guys. I double-checked, everything works here. Right. It was actually for me Cool. Also like this from Bashrga cool topic. Will uranium scrap sink? in mercury? He understands this. That is, I told him I don’t write there, let the scrap, the scrap float, and I'm writing to him specifically, "Show him what will happen. " Here this is very cool. This is all you can do stop there for many hours get stuck. It was also cool with the cube. Show me what will happen to this ice cube. in an hour and in a day. Here he did it two pictures after an hour and after a day. Here. That's all too. In general, it is necessary banana is just multifunctional a combine that does incredible things things. AND localizers, designers. That's it, that's it is it really possible for us to do this? Here's what to do localizer? Is there any design? You I made the design, you just throw it in, you say: "Translate into Turkish, translate into Russian. " Bang, translated. Well, this is brutal. Eh, horror. Neuronetworks with video. A quick word about the video let's run through it. Katya, there is no point. I'll answer the question quickly. On which one? Is it better to write in Prompt language? In Russian or in English? This is actually important. If you are Russian I wanted to say, if you don't know Russian You know, write in English. If you don't know English, write in Russian. But if you are at least there plus or minus Do you have any knowledge of English? You can write promt in Russian. language, but translate into Russian English via Google Translate, via anything and double check because very often it happens that this is the very best, You know, this is the most common one. example. And you write, look here attention, I wrote: "Take four shots of this woman. " Why a woman? This is a young girl, beautiful girl. Why I wrote woman? I should just fly from this girl for such words. And the thing is that if you are Russian, in Russian The girl who wrote the message "Please throw it in" neural network, then it will translate the girl as Girl. A Girl is a little girl, and You are immediately censored like this. That's why It is best to translate Russian into English, double-check. Because sometimes they come out not quite eh, maybe, double interpretation of the translation of the word, and you just won't get what you want. That's why promt engineering is a disaster What an important thing. You don't just need, Well, it's advisable to understand that you write, but control what you write. You wrote in Russian, please check it. maybe you have extra phrases, you need to optimize the text, you need to compress it it, you need to make it more capacious. Eh, then everything will be fine. Okay, well let me also just have a video sound, yes, and here further we have there further trash will become even more fun. All It will be recorded, of course, everything will be in records. What model is it, too? Fashion. What is a model? If you open Google Translate and write a model, translate it into English, Well, later, I won’t do it now, but then the module can be interpreted as, for example, there computer model, just any model. Model boat, assembled from paper. This is a model. That's why Of course, a lot depends on the context It depends, but it's better not to do that. That is Always give more e clearer instructions. Text Neuronka, yes, you can ask for the GPT chat help make promt in English, but double check again because There is always a certain percentage of glitches, always. there is this uncertainty. Besides the photo, here you are they slapped a bunch of photos on there, You've done everything, now you want this bring everything to life. Why do you need to bring all this to life? You want to make a video. Let's say you so they took some girl there, no I know, there's a girl holding it in her hands bubble. You did a good job in Sidrim image quality, now you want make her take this bottle She opened it, drank it and sat there contentedly. How to do this? This can be done by cyphreys. You can make several cyphraims. Let's say the first frame, she holds a bottle in his hand. Then you this you upload the photo to the nanobanana, please let her be there drink from the same bottle. He takes a second frame. You are the first and second you throw the frame into the neural network, which video generates and intermediate frames She makes the video straight from start to finish. end. The first frame will look like this. In the second frame she is already drinking. And there write that she should turn away the lid, brings the bottle to his face. Accordingly, for this we already need video models. Now I'll quickly do this I'll tell you straight away and we'll do it short break. Accordingly, and for you are important first and foremost, Guys, first of all, what's important to you is Models are extremely important, Sor, Veo In principle, this is a must-have for Runway Alif. have. These are the very first models, with which I would recommend you start with work. Everything else is in quality additions. What can you do there? Well, Of course, the simplest thing is this various image to video. You took it any photo on the internet, you threw it into the nanobanana, through the nanobanana you added a computer to a person's hands and The magician wrote a request and let him speak. phrase, and in Russian at that, that's nonsense. Horror moment. That's how it happens. I talk so much, I'm so sore got sick, accordingly. What's wrong? What the hell is this? Both of these are [ __ ]. ABOUT, That's it, I've lost my voice. I'm sorry, Guys, it happens. Accordingly, that's all is done through video models. There is inpaint. You can do it in pain video. What is inpaint video? Look, I have a hat on my head. Do you know what you can get from a nanobanana? hat and replace it with blue hair. Seems easy. But what if you want to video replace blue hair? You don't you can just take it and, uh, through Nanoban replace the hair and then animate the video, because the movements will be unnatural and everything will break for you. There is a reason for this additional neural networks, for example, Runway Alif. There is something in the Kling now motion copying, motion capture You can do the same. Accordingly, you also need to understand additional neurons that which will help you with such things implement. Okay, I wanted to show you, but I won't show you. U I got the hard drive, so I'll show it here. Now quickly right here, like this, right at the very end throw. At, look, not here, right here. Crap, and my hair turned blue. That is, all this is already being done through neural networks for working with video. Well, this one I'll show you a specific example a little later to you. The same thing happens with LipSync. Let me show you first and then I'll explain. They will say that a magician is a complete, but I'm sure you're wrong Question: how to do it? And to go to the isikat, that the magician is complete [ __ ]. The man holds a magician in his hand, that is, he it doesn't move anywhere, there are no gluing. And the next frame. It's exactly the same in your hands spins this Max Studio cube. This It's made into a hodgepodge of neurons. That there is a certain pipeline here, which allows you to do this. And for in order to it was a little clearer, I was straight I'll quickly sort out this situation. We have the original. Here There's a guy holding something in his hands. a red rag. I take it and throw it in Everything is in nanoban, I'm taking Max Studio, please he needs to uh make the man hold computer from the second frame. Here, Please, I have a frame. Next through kling. There was no motion capture back then. Now you can take the original frame, how he holds, how he speaks, with his lips pronounces. You can take the capture movements, throw it on this picture and It's much easier to get it directly, uh, less hemorrhoidal situation. But then I did just as I did. That is, I did it through the blade just came to life, as the man stands and looking at the computer. Here he is, accordingly, he simply looks at computer. I have a three-second video it worked. Next is this three-second video. Next step. We take the sound we throw on the path separately and make it lipsin. For example, through the wedge model Same. I did lip syncing through the clink lasting 2 seconds. That is Now I got this video which it came out of the frame, but he also has lips are moving. And the result was finished video. The same thing, absolutely the same thing happened here. That is The first step is to hook it through the nanoban computer in hand. The second is through the blade. bring this video to life. And the third is add audio track. That's how things are. That's why, eh, that's why these are the pipelines, I don't like this word, write it down, love it the word pipeline. Doesn't it make you cringe when Someone says pipeline. Payplay. instruction, sequence, algorithm. Let's get technical. These algorithms are completely different. for different tasks. Therefore, in order to in order for you to study these algorithms, you understand, you need Where to get them? They're twisting. They do distort, yes. Lipsing is a separate tool. Payplay. Something like that, you know, immediately struck me pipeline association with lineup in surfing. This is the place where the wave appears. And here it is Yeah, it's in the kling, guys. If someone It works in the syntax, but you have it separately tab. If you are using a Telegram bot you do, you don't go into the clinch, you do it through lip synchronization. There is synchronization. You click on synchronization, and you already have it separately is selected. Either you do it through the blade, or through runway. It's all very simple. So, before that Let's move on, let's move on to these like once the pipeline is hated by everyone and algorithms. Where can I get them? So that you don’t freeze for long, I’ll quickly I'm showing you this situation. We have in syntax Coming soon, very soon, 16. 021 February starts a stunning, incredibly the coolest, awesome course, which consists of 21 lessons. Lesson 21, guys. two top speakers and all sorts of goodies, trinkets that will allow you not to just to do something there, because we said, but first of all we need to understand and explore this kitchen from the inside. A, accordingly, I invite you to our A course on Syntax with Nastya Romanova together with this by the amazing Maxim Gustarev. You will study on Eroset. You can read who we are, but I think you already know already, if you are in our master class you are present, you already know that we have some knowledge. We have I have something to share. And with this knowledge we will share with you, we will give you stable result so that you already generated various interesting ones normally things. Why do you need this? First of all, this whole situation passes in order to you've figured out all the neural networks, which are located in one place in Telegram bot or on the SK website. AI. Here. Well, you can read it yourself here. these situations. You don't have to spend hours googling, collect a piece of information. All of these beautiful words, they are, here, honestly Speaking of which, they don't appeal to me. For me the most important thing is what we convey I will share with you my experience, all the accumulated experience, all the important things that will allow you not to only generate from simple text Promtama, but first of all, understand anyway, what are you writing there, what are you there you write down what you want to get from its neural network. That's exactly it will be on our course. Not just simple Here is a crude theory, this is first of all practice because you are here even It is written that you will form your portfolio, you will have a normal, adequate a portfolio that you can use start. Why is this cool? Yes because you will at least feel it out soil, choose certain ones directions that interest you. Having made a portfolio and received feedback connection, you will be able to understand, and where you develop further. That is exactly it the beginning is such a good kick in the ass, which will let you go further open navigation and further study of everything and everyone. Guys, this course, I'll tell you right away that this is not a course, but this Not finished, completely collected within herself store up knowledge. This is a huge layer information, but that's not all. It is very important that you continue to learned to educate themselves. For us, I think for Nastya too, for me too first of all, it is always important, any training, I always talk about this. For It is very important for me to give you the right one vector of development direction, and then you we'll give you a bunch of stuff ourselves information that you will digest, which you will understand, and then you will set off on an independent voyage. You will no longer have any further courses in general are needed because now on the Internet, in principle, there is information. She is she just scattered across the vast number of different resources. If you even If you visit me on YouTube, you will find videos on neural networks, you've already got a bunch of them from there You will find all sorts of interesting things. But to don't be on the lookout, so as not to wander from one place to another resource to resource, here we collect everything together, and it will all be here. This course is suitable for beginners because that this is a confident command of the basics neural networks, this is an understanding of how it is worth The task for neural networks is what to tell them. This the first works in the portfolio, I already talked about this said, the skill of creating content, which previously seemed difficult. Here Was it difficult for you to find something in the nanobanana? do? In fact, if you know how write, to ask, then It's very easy. Of course, we will give it to you diploma. For advanced users this will be a bit more structured systematization. We will collect all the knowledge into one and allow you will see a clear picture. Of course, There are new tools, I assure you. what to show. Even in editing, even in Photoshop, because my blog will be mainly about editing, about Photoshop. Photoshop, Final Cut, CapCut, Inshot, that's all. Everything you poke with fingers, simple routine, not connected with neuroses. This will be my main emphasis. Accordingly, it will be interesting, What is the structure of the course in brief? Poke-poke-poke-poke-poke. There will be 21 lessons. There will be many lessons. If you take the tariff Creator, then you will have six blocks 16 lessons. If you take Creator Pro, then you will have five more bonus lessons. Click-click-click. What's inside you, accordingly, introduction? Let's figure out what Syks AI is, how to use the site at all, because that there is a GPT agent and all sorts of tabs, there are a huge number of them. In this too We need to figure everything out. First lesson, I think he will fundamental to figure out where you'll be in continue to work. The second will be analysis basic language models. How linguistic models think as they understand yours requests, again, what to write. All This will be in the second lesson. Third block - this is a fairly large block design and visualization with the help of you, Anastasia will tell you what idiogra is, cleans, runway, sore image. Of course, there will also be a huge lesson on journalism here, because it's not like others neural networks are again a combine harvester. If nada banana This is a model that communicates with you. text, then medical journals are a model, which requires you to know uh key phrases, sliders, their sequences. Respectively, The fourth lesson will be very long. it is precisely according to the setting of Midzhurnia, according to its total control. Please pay attention attention, there is homework, and homework too there will be no homework at all. Horrible. And you probably lost me now, Yes? And I'll come back to you now. So, moment. Do you know what happened? I have The camera overheated. I didn't overheat, but the camera did. Oh well. Look, here I am again. Do you know why? Because Maxim now he's a streamer and he has webcam. It's normal that and the quality is generally normal. I don't know. So, OK. I'll finish it quickly now. I don't finished speaking. Write in the comments, guys, what, how You have things to do. Are you there? aren't you tired? Is everything still normal with you? Mood? How are you? Everything works fine. There will be STL. So. Oh, so many questions. Yes, it's the same course that announced in the twenties of January. Accordingly, this is it. Let's do it this way I'll do it so you can see that I I'm reading. What is par for the course? A little further I'll tell you. How long does the course last? Classes are appearing twice a week, also check it out. Want You can pay extra, I think. Valeria no problems at all. Can. Uh, it's on. Is a Syntax subscription included in the price of the course? Uh, sorry, I have paws, I don't have control this information, but I don't think so. That is Most likely not. Although the guys need it ask. Well no, I don't think so. So, What is the difference from a knowledge base? I didn't understand the question. Like a course from the base knowledge. What is the difference? Knowledge Base - This is basically all the information that I know. So, access forever or on some period? I also have paws. The administrator needs to respond. Eat Does it make sense for beginners to buy Proversion? Of course, because, guys, kids, that's all we have now we say, this course is not This is not quantum physics, it's a course, which one you will look at, understand, and which one you might need will be rewatched a couple of times if you will forget something, but having this course before yourself, you will review it there rewatch it a couple of times. completed task, completed homework and We watched it again. Now, once you do this, you remember all the information. This is not quantum physics. You won't need to constantly return to this course. No, of course not. Therefore, the basic one, which is the crate, the tariff, the first tariff, it gives you information, and you understand it, you are studying it, and Creator Pro is also additional information that is not will eclipse the first one, it will simply be addition. This is a little further away, bigger functionality that you need in any case will have to be studied further in the Internet, but in this case you just You will receive it on an additional tariff. So, uh, of course, It's better to take Creator Pro, uh, because There is simply more information there. So, It's suitable for beginners and for everyone. Of course, guys. There is no stable diffusion, it won't happen because it has to be done separate course on stable diffusion. This the situation is so broad precisely with from the point of view of system settings administration. It depends on what you have video card, and all that stuff. Well that's it tin. So. 21 lessons or six by 16? 6 + 16, I guess she wanted to say. No. 16 lessons for the basic tariff and 21 lesson. That is 16 + 6 for Creator Pro. Uh, I probably made a typo, guys. I'm sorry. Yes, even free classes on YouTube to me. What are the free ones? classes? What are you guys doing? You, here I am now I'll show you a life hack. Look, Free lessons, write to me. Here you are you take it like this Me too. You open Telegram, And There is a community on Telegram called Syntax Family. And you just do it like that scroll and look, that there is a huge amount here educational information that you can integrate into your creativity. There are just some here and not just some there promty, but instructions on how to generate video, because what to do there again various gluing, how to control generation. If you go to The official website of Syntax, that's where it is knowledge base. And through this knowledge base You can even start with this. Here in front those, uh, how to uh think, in general uh look for some resources online, you can start with the Syntax base. There is a lot of useful information. Like this. There is no speaker in the chat. Don't know. So, Let me quickly continue. Moment. Okay, okay, okay. Hope, I hope you haven't missed me yet. Come on, we have something else to talk about. talk. Respectively, the fourth thing we'll have is, uh, The fourth block is the video future. There They will tell you about video generation, about everything, everything, including from the runway to Hixlda, about unique directions cameras. About presets and how to mix them and so on. You will have a wLlin, animation by text, by image, function elements. In general, if again I I'm saying all this now, but this and you see it anyway. Мижурини, sedнс, writing prompts, it's clear who would I doubted it. Here Jurney, Sedns, Klink, Runway, Hicks, VO3, Sora 2, that's it. This is what is needed now to learn. These neural networks will be in the future shoveled across and for you provided. If you just want, again, I don't understand now what here inside, because it doesn't matter. What matters is that these will provide you with The neural network and understanding of the work will give invaluable experience working with these neural networks. This is important. Awesome. Ae audio, self with yourself, you will run through the audio pro voice cloning, about Shuno, about sound generation. Suna is actually mine Love. I'm currently using sun in my tail and in mane. I'll show you some examples a little later. even. Ah, well, here is the sixth block - this is purely my block. Uh, this is straight to me. AND there will no longer be a neural network here, but here there will be more will be -e work with the technical component and with classic software. Why with technical component? From a technical point of view component of neural networks. I'll tell you, How to get high-quality generations. AND it's such an abstract phrase like: "How receive high-quality generations. " Write high-quality industrial goods. Guys, no way, no It's all that simple. Not in one promt everything important. Let's say if you need You can't just change a model's clothes take and upload a photo of yourself models. You need to understand in what aspect ratio, in what color space, in what resolution is this Upload a photo, in which car scale. What the hell is going on? Horrible. I choked again. Again Sorry, guys. ABOUT. I haven't held master classes for a long time, I haven't chatted for so long in a long time, I have a voice breaks down sometimes. Respectively, Here we will talk about mood boards, references, ideas. We cross out the pipeline, write the algorithm, operating algorithm. I'll show you how to handle all of this. The case on the phone, how to do retouching, how move color, how to prepare, so that not to get a loss, a wild loss in quality. This will be the thirteenth lesson. Lesson fourteen is Photshop. Guys, Unfortunately, you can't get anywhere without Photoshop. nowhere. If you thought that now anyone AI creator can easily do without Photoshop, well, It is possible, but the quality will be poor. Photoshop is basic, uh, basic the structure on which any visual is based. If you take any person, related to visuals, photographer, there, web designer, e printing, a person associated with printing, with anything that has a visual element art, even that of an artist, he understands, What is Photoshop and how important is it? Accordingly, I will tell you about operations in Photoshop, how to create masks, how to apply these masks, how work with cropping and so on. Of course, there will be applied information. I won't tell you straight away chew base, like how the buttons work, like create and invert a mask. I won't get your attention stuck on this. I will be more use applied methods processing. That is, everything is really fast. do with detailed commenting. Here. Accordingly, it will be fifteenth lesson. It's already the sixteenth lesson. We will move from photos to videos. We have there will be editing. Mounting on the phone via CapCat, Edits and Shot and editing on computer. If you are working on a Mac, This will be the Final Cut. If you work On Windows, it will be CapC. Everything that concerns cutting, layers, effects, exports and, of course, also masks. That there is something that I told you before showed with blue hair. I am this I did it using a mask. I'll show you again, so that you understand that this is also not true just do it. Here's some extra info For the Creator Pro plan you will get more additional six lessons. This training of lore models, keyframes, avatars, askcalingng, creation own mini-cartoon. You can read everything that is written here. There are five lessons here - this is in addition to basic program. Let's quickly talk about the price. Finally got to the price. Basic tariff for the crate. The cost is 270 euros. But if you again don't drag it out, don't drag it out tail, now this tariff costs 220 Euro. 229 euros is the optimal starting price. If you want a little more information, then run to Creator Pro. 299 euros. Now at a discount for the next 24 hours you have There is an opportunity, I'm not pulling, don't pull. Uh, either do it or talk like They say, don't do anything, just rest. Here I am I would recommend you to do it already figure out all this neural network stuff situation and already take some measures actions. At the end of the course, a very important remark. IN the best final works at the end of the course will be published on Instagram Syntax Creators. And the authors will also take part in the Syntax Ambassador status drawing AI and there will be a competition with prizes. And here it is Here I want to make a special emphasis on this situation. To be an ambassador Cool. I myself am an ambassador in the former. Now no longer an ambassador. I am the ambassador of Sigma Rsha and Israel. Sigma Israel two companies. This is if anyone has a photo. I know, there is a Japanese company that produces lenses, objectives, Sigma. They also make cameras. I lived in Russia, was the ambassador of this company Japanese. while I lived in Israel, I was there too ambassador of this company. Right now moved to Slovakia. For now we stopped working because because, but there is still everything ahead. Here. AND being an ambassador is a really cool thing topic. It allows you to use in First of all, let's do it this way I'll do it right away. She constantly allows you will receive it on a permanent basis Ultra Elлиit subscription. This is the top one subscription to Syntaxbot. This is a monthly reward of 300 official green bucks. This Synntax official ambassador status. You can write in your header: conducted, that you are so cool. And your marks Instagram profile. Early access to neural networks and functions that are just appearing in the bot itself. Very cool situation. That's why Study this slide now. Here Everything is described in detail. Here, by the way, Someone had a question about access to the course at 6 months. Five additional lessons about block. I'm there a little bit in the presentation I messed up, but don't scold me, I I'm a creative person, I can. Eh, bye that the discount works, don't be stupid, sign up. For everyone who signed up for these tariffs, applies twenty-five percent discount on any subscriptions Synнx AI. So, get yourself one too. subscribe and generally enjoy life, because this thing is cool. February 16, Guys, we're starting. Now we have the fifth number, in 11 days, well, a little less, than in 2 weeks. Well, we don't have to wait long, so don't be stupid. Uh, if you want to get a better price, then choose one. of these two tariffs on the official website. You can find this course, look, read and sign up now. This is something I can strongly insist on you. recommend. This thing is not bad, this one This thing is incredibly good. Here So. Uh, what do I want to tell you? Well, we We've been playing around with you for an hour and a half already, but I I've almost told you the most important thing, Well, he told the course, but didn't finish telling about pipelines. We haven't reached the end yet. Write how Are you even alive and well? The lessons will be recorded, guys. No no, recorded. And the material will be fresh. so fresh, I'll tell you right away that I haven't recorded it yet. That is me now I'm recording material just for you. I'll definitely send you the PDF file, no Don't worry. Okay, well, Max, say: "Eugene". Evgeny said Evgeny. Let's move on, kids. I hope you not tired yet, because there is something say. Okay, let's move quickly. Synthesis speeches. Speech synthesis. And in the AI ​​syntax you can just imagine if you you're making a video, you need to somehow synthesize voice. Besides the fact that 3 and the blade and Sora synthesizes the voice, let's say you can replace this voice. You can take and record your voice and his reproduce. That means you can reproduce their voices. Again in There is a huge number of votes in the database. You can make custom voices. This is all is done through the same speech synthesis. Nastya will give you more details again. will tell you how it all works. Can You can clone your voice and make music. I want to tell you directly about music, that this is the funniest thing that only exists in sound generation. I'll show you a short video, so that now I have totally recorded a gigantic number of musical compositions, generated through Suna. They are all without copyright, they can be used everywhere. I often use it on my YouTube generations that I do in Sun. So if you do any advertising company and you need a cool one music, you can do it with words or without words, you can generate it through Suna. Suna is an amazing neuron, which makes it so realistic recreating music that's straight to the brain explodes. It's you maybe the weather. Be with me and friendship. Let's do it You have gathered an army. army Poke Pokemon. How awesome is this, how awesome is this Cool. I love post-hardcore, emacore, alternative and other nonsense. And I just took music from cartoons of the nineties years, there's like Darkwing Duck, Chpdale and Pokemon, of course. And so I made these compositions in the style of Wcals, post hardcore, Emma Core, Emotions, RapCore. And this It looks so cool, It sounds so cool. And I do this I use it in YouTube videos. More moment, purely theoretically, just don't tell anyone, But there is such a topic, there is Epidemic Sounds. Do you know what this is? This is an aggregator music subscription that you can use on your own in your own YouTube videos. And it's protected by YouTube's Content ID. AND if you take it from Epidemic Sounds, If you find a track, upload it to YouTube, then you will get a strike, and you can take this recording, some short piece, throw in the sun, make a cover. Legally it is illegal, but if you have a small project, then, maybe you won't get it, but I won't arrives. But don't do it that way. That's it, I'll tell you. said. Overall, the sun is a cool thing, which allows you not only to do music, not just generate it with zero, but also to do covers of famous ones songs. This is so cool. Well, here too There is a sound generator. Here. Fine. Here We've gone over the main points neural networks and it seems to be clear for now, well, They do things like generate images there. and so on. Where should we look next? In addition to what you generated, you also need to know some software classic that will allow you to do this arrange everything, comb it. There are two types of software: complex and simple. Let's start with something simple. If you want to do something here and now with there basic adjustment, you can use Cany, you can use Adobe Express. Adob Express, this is a funny situation for you. You don't need to install anything, that's all works on the web. Here I am now on Adobe Express I am here. You can delete it here background from the picture. Let's say you It is necessary not only to replace the background there, but make some kind of collage out of what you generated. Here you go, through remove background, via resеiz images. And, for example, I often I use the convert to give function. To convert this video into a GIF and insert it there in the same one presentation, I use Adobe Express. In fact The software is cool, there are a lot of functions, some web multitool. Study, See what you can do here. Further, you can work mobile in enshot. If you are working with video, cut video, there is no background to remove, but there is color correction, resize. This can be done in enshot. If we speak about computer solutions, uh, PC solutions, then it could be Final Cut Pro X on Mac. I edit everything in it. rollers, and it is enough to even do some plus or minus non-standard, a little more complex graphics. Let's say, insert using a mask hair. This can be done in Final Kate. A very easy to learn program. ABOUT Most of the people on the course will be there my information, because this is the very thing the most useful thing that can be on Mac solution. If you work for MacBook, on an Apple computer, then this is exactly 100%. If not, work in Capcate. I'll be talking about the cap, that is, how to do it and finalalkade, how do in captivity. The information will be intersect, but I can't talk about that either I can't say because not everyone has Macs. KSAT is also multifunctional assembly tool. There is also layers, and background removal, all sorts of chormakei and other. All this is also done here. to you I'll tell you how to edit a video, how to cut it, how to combine it, how add audio tracks. If you are you working? also with sound so that there put on all the music, then copy and cut, you also need it needs to be cut. All this can be done as раз в файнал кукте, в кепкаате. This is all simple software that requires minimum entry threshold. Even if you never worked with programs there video editing, it will be easy for you to do it through this software. If we are talking about more complex things that you will then in the future you will move on, then here already a more serious one opens up bridgehead. You work with Photoshop. Photoshop it can be attributed to simple, of course, but in fact e Any kind of multifunctional one program, but Photoshop is real a very functional resource that allows you to work with raster graphics, it's still better for complex ones take away. This dad is just a pixel work. And by the way, we will also talk about him on the course. speak. And this will also need come gradually. I have it soon My boys have been waiting for me for 4 years now. U I will also have a Photoshop course here perspective. If you need it right here in full If you want to study only Photoshop, then go there. And so, uh, basic things and basic working applied methods I will show my work in Photoshop on this course. Next Afintifoto, analogue Photoshop. If someone doesn't want to get involved with Photoshop, If you make a video, then you have three programs, uh, well, actually there are two of them. Nm, still. Adobe After Effects, it will allow you make more sophisticated masks, work with camera, but that's already a separate thing profession, it's a difficult situation. But if we talk about VFX computer graphics, if we are talking about, eh, the synthesis of neural networks with computer graphics, here, here's all that, and to you incredibly more serious ones will be needed programs like Adobe After Effects, or Davinch Resolve. Of course, Davinch Resolve too again multifunctional program, it is not It's just tied to the color core, there is very complex internal structures, uh, that allow you to do incredible things complex things. But You can go there to study. There it is all Hollywood is sitting on a nut. Nuke is cleanliness creation of special effects. And so if you have there is some short video you want make everything super perfect with complexity compositing, then you can do it there nuke. But again, these are the programs NK level - this is what life needs dedicate. That is, you will dedicate your life neural networks, and then another life, the next life can be dedicated purely nuku. It's so deep programs, just like blender. Blender works with 3D, but that's also uh program within the complex level, which allows you to do more complex things. That's actually all these programs that are here if we say things like, you know, allin-one artist, artist who creates the special effects there are incredibly high quality, who is rummaging around for Da Vinci Resolv, who is rummaging for Nuke, he is most likely he will also be dabbling in Photoshop, he is more likely to in total, he will be rummaging around for sound some programs, all of this together, because, well, that one thing depends on another, and they complement each other. Again, if we are talking about the ecreator, accordingly, it's just on the future, so you have a picture in your head appeared, you saw some programs. You can add more blue here add for 3D graphics. the same 3D Studio Max. I think you know if you work architect, then all sorts of ArchiCAD, AutoCAD too you can drag it in because any program tied to certain niches, if it in any way connected with fine arts, with visuals, they will intersect, and knowledge of one the program will be a plus for studying another program. This is regarding the complex. Next, let's assume, Now you and I have understood and accepted all of this, forgiven, studied. What's next? Further the acceptance stage is underway. Next comes the stage acceptance. We are now moving on to specifics pipelines. I'll show you a couple right away examples, and we are with you today let's finish. I will answer your questions quickly. and let's go have a rest. But first, I I'll show you some work related to Photoshop. Now we are talking about Photoshop so that you understood the seriousness of the situation. Let's say I often use Photoshop NanoBana, I'm adding angelic models. wings. I have a series of works with angel wings. Here I even I added feathers to the bra like this. Here too, uh, here are angel wings also, uh, quite realistic. And if Look at it this way, everything looks fine really. But if we look even here here, then in addition to the realism of these wings, we Let us note one more distinctive situation. Look at the resolution of this images. Resolution of this image incredibly large. The is 6,000 pixels. at 4,000 points. So that you understand, Our nanobana generates 2048 by 2048 pixels. And if I try to feed nanobanana this whole image, this whole picture, then nanobana will tell me: "Maxim, I'm sorry, but I can't... generate the image normally, and Nanobana will give me soap. " In order, To maintain quality, I nanobana, Look, I'm feeding you a small piece. That there is this specific cut out a small piece. Next, promt is written. Again, add to the woman, not the girl. E add angel wings. These angels wings are added to the mask, like this mask, they are glued in. This situation You can see it in many places. Here, to For example, the wing mask on this one photos. Hard wing mask all feathers, plus a slightly softer one, shaded over the rest. And in As a result, we get this, well, again, quite realistic, without reduction in quality, without loss details, again, please pay attention attention, the detail is preserved image. If you just, again I repeat, throw all this on nanoban, Nanoban will eat up your quality, and you Get some soap. And they can wear masks duplicate, they can overlap on top of each other so that it would all be e more realistic, so that you don't have a breakdown loss of texture and its presence, you, Of course, you need to understand Photoshop, How it all works. This one is also like this I'll show you a small example. Here is a girl with beautiful eyes. I added it to her purely there is makeup on her face. This is it generation was achieved. If her look, it's a square. Ordinary square loaded into nanobanana in resolution 1024 pixels. By the way, on that one the moment when I made this PSD file, Photoshop file, I took the 1024 generation pixels, because there weren't any back then Nanobon Pro, which was already able to do 2048, that is, this is the previous model. Now there are more professional model, that is, it is possible generate even more detailing, but I still inserted it image, I only drew it using this mask just the face. Then I inserted it already wings. And also, if you look at mask, just pay attention, mask, which uh moves everything into the picture, except for the face. I left the face to maintain facial texture. This is it the situation may occur. That's why one moment. Nanobana is very picky just like with this element. That is If you upload something like this, you'll get a nano banana. will send because she's naked. For to fix this, you need to load it on her, uh, patches. Let's say we can take some brush, cover this one the situation like this and dump it on the nanobanana. As you can see, I have in generation Nanobana appeared like this space is black. Why? This the patch that I originally I dropped the nanoban. And then you can Carefully cut out the mask here like this with wings, like this pixelprtnym By masking you can cut out background layer. And it turns out already the resulting image. Also a mask. one, mask two and the final result. We also have a fairly high resolution we get wings. This is what concerns inta. If you want something a little more serious to get down to business, you can use various upscales. You can take again, girl, write a prom, add angel wings. I have this one frame, and by the way, it's open. Here he is I'm showing the moment. Here you can insert wings, and you will get something like this generation. Here is the original generation. He is not very high quality, but again, having shown the face from the background models, we can do it in some easy way Composing will fix all this. Besides the fact that you put on wings, you received low generation quality. You can use upscalers. There is a syntax upscale, which allows you to increase image resolution with coefficient, let's say 3x. That is, three times I increased detail. And already this one detailing, highly detailed I'm uploading the image after upscaling. Photoshop and overlay it on the original images with a specific mask and I get higher quality, more pixel quality. Same thing, eh, It's happening here too. I have e the wife I want to dress up as elven image. What am I doing? Understandable The problem is, I can't transfer it to Banana. the image is original because it will block because the underwear is there All. And just get the hell out of here. Accordingly, I make a patch. Once, patch two. I take the face separately. I have a separate face and additionally nanoban I feed a donor, so to speak, a reference. Please replace the image. As a result As a result, I receive a replacement. Wearing a tiara, ears, elf, white hair. Next I will upscale this image. and I throw it here, thrown separately plate, square plate images. And plus thanks to these I put on the painted masks again white shirt. Well, it works out. final picture. Accordingly, this is everything is being done. Again, I have file. I can get it for you quickly. show, uh, so you understand how it is happened. That is, here we have it original file, and quite high resolution. What resolution? E 2x for 3,000 points. That is, well, again well, this resolution is more than Full HD is more than just a screen, it's there iPhone, which when you look stories, stories are 1920 in size 1080 pixels, that is Full HD, and here This is more than twice as much. At two times more. Accordingly, the first one This is the picture we got. I just chose a rectangular area with the correct aspect ratio is 4 kp. Many models, they learn from certain aspect ratios and better feed them certain ratios sides. I abandoned this matter and did it. mask. The mask passed through carefully along the neck line so that we don't have a cut bodies, what oh, I'm sorry, so that the body was more even. Next, uh, the generation started with the head. I generated the head again square areas. I chose separately square area and looked at different generation. For example, I changed the color hair. Again, it's all square and uh square that is fed to on a banana. Changed the hair color again, by mask. And notice when we we're throwing it off and generation after nanobanana, nanobanana eats away at quality. Just look at teeth. That is, we initially had teeth clearer than after. AND by throwing your hair over it, you can create a mask, which will restore the original texture. Can we add more to this generation? throw an upscale that will increase detailing of the hair itself. I tried different hair types, there are a couple of them hairstyle, uh, with a turn of the head. OK. Or by reference, as I told you before I showed it, I put on this mask reference image. By the way, it is after upscaling. And then I have, as far as you remember, if it's like this look, we have no hair from below. They are cut, they work. If we just If we insert an image, we'll have, uh hair cut off. Even if I wear a mask I’ll draw it, delete the mask from below, yes, mask, Of course, it is drawn by hand. That is, she painted by hand with a brush. All this It's not that difficult to do. Accordingly, we have the shaving brush left only hair on top. And more is being done one generation only for hair. That is Here again I took the rectangular one area, took and asked for a nanobanana make realistic hair. Accordingly, nanobana finished drawing strands of hair, and I got e more finished look. I also have a smear I went over the texture of the face and left the face in place, and it turned out to be so piecey there would be a compositing that would result in allowed to obtain a higher resolution with local replacements objects before after. Before and after. How can I get a 25% discount? 25% discount on subscription when purchasing a course. That is, If you buy the course, come to course, you automatically receive twenty-five percent discount. Ah, guys, this is the last, third example, which we will analyze today and we will, I I'll already read your questions and we'll be with you let's go have a rest. We've been chatting for 2 hours already. One last example. The same thing works with video. The same works with video. Initially my frame has 4K resolution, 3. 840x by 2. 160 pixels. It gigantic. And this is a gigantic resolution, ah, it is will be eaten by a neuron. If you throw this away into the neural network, it will simply get eaten. That's why I take a square area. I throw this square separately area in the neural network, specifically the small one region. Next I ask to replace the blue ones hair. I replace the blue hair and get a frame that is compressed. It can be driven away separately via upscale. That is, here You can add one more step, run it via upscale. By the way, Syntax will be launching soon such a small insight, coming soon Magnfiic, one of the best upscalers EVAs created by man. Respectively, already finished video. I insert on video sequence, I create a mask and based on this mask here is the purely central part, uh, without a face I insert it like this. And it turns out realistic generation of me with altered hair. At the same time, the picture does not fall apart into pixels. At the same time the picture remains sharp because it was originally, uh, generated on pieces from 4K video. That there are these local corrections, they are just some small life hack that you use. And here it is these life hacks in There are a huge number of them in folk art. There will always be compositing, uh, always. compositing will be useful for any kind of video and pro during photogeneration. Like this. Accordingly, and guys, I also tell you I want to say, this is the road map it worked out. I will definitely give you this PDF. I'll send it to the guys from syntax, they'll will be posted. A What is magnific? Magnific is a very cool upscaler that does image, ascale image realistic. That is, if you take, you feed him a photo of a person in Full HD resolution, then it will give you The output is not just 4K, it's 8K. And each one it's time, even if there were no pores on the face, then every pore will be visible. Here. Very cool thing. That's how it is, guys. If you have any questions, you Feel free to ask. You have courage, let's go on topic, off topic, I'm happy to I will answer them. Oh, I forgot what I have webcam, the quality here is crap. So I'll turn it on again like this, so you can see the roadmap. And on Today we will be finishing. Enroll in the course. You have it in There's a big button at the top of the chat This blue one, sign up for the course. Sign up. Again, there will be a lot useful information. If you thought, with where to start neural network study, here You need to start with this course. This is the most important thing. OK. I'm very glad that you are here today came. Thank you so much for that, that joined Strimchansky. Here, uh, subscribe to YouTube, Subscribe to the Synx Telegram channel Family. In general, in Synnks Family No, it's not just subscribe to Syntex Family, don't be lazy, scroll straight, yes, scroll up far. There are a lot of interesting things there. life hacks in terms of generation, styles, prompts that you can use. You will be 100% inspired. guys that I showed you in the presentation, you can do it too subscribe to them because it is cool guys who make cool content, beautiful content. He, well, no not only AI slop content, but also artistic, beautiful. I'm sure of it You will also be inspired. Here. Big thanks to everyone Thank you. I was glad to see everyone. To everyone Bye, hugs to everyone, cya-kao to everyone. All.