Magnific AI : Master Advanced Image & Video generation

1. Introduction: Landscape of digital creation is moving faster than ever. As Freepik evolves into magnific, the creative possibilities for generative AI has become incredibly exciting. In this class, I'm going to guide you through this revolutionary platform, showing you exactly how to generate mind blowing images, cinematic videos from scratch. At first glance, advanced AI interface can look completely overwhelming. But don't worry, by the end of this course, everything will feel incredibly easy and intuitive. We will dive deep into advanced techniques like maintaining absolute character consistency across your frames. I will also break down the past image-to-video models to use for your specific workflows. We aren't just adjusting sliders. Together, we are going to build a complete production ready storytelling projects. I will take you from the absolute basics to advanced tools step by step, whether you are freelancer, creative professional, a video editor, or a digital marketer. This course is built to level up your toolkit. Let's get started. See you in that course. 2. Navigating the UI A Beginner's Guide to the Interface Freepik/Magnific: This lesson, you will learn how to set up your free Pee account. But more importantly, I'm going to give you a full guide too for a Freepik dashboard. You will see why it is now the ultimate all in One solution for artists, video creators, and business owners. I have been using Freepik for quite a very long time. Years ago, it was just a nice website for graphic designers to find vector fs. But today, it is a Powerhouse. You name it and you will find it in Freepik. Setting up your account is incredibly easy. It is just like setting up a Gmail account. So first, you have to go to your browser and you have to click on this link. That is freepik.com. Once you click and this is the window, we'll pop up. So they have regularly changed the layout of the website just like they have changed the Spaces intro, the video generation. Right now, they have used this kind of video generation and image generation on their front page. But before they only show you the graphics and videos. But right now it has changed a lot. For the log in purpos you simply have to go to your Freepik and it will automatically ask you to log in. Just like any Gmail account, you have to go to here and sign it by using Gmail or you can create a password. So you simply have to sign up here. That's all you have to do. It's not a hard and fast rule here. Once you logged in, you will see a premium here. So you will see a upgrade to premium here. I have already upgraded to I have already upgraded it to the premium and subscription billing, so it's not showing to me. After you have sign up for it, you will see this crown image. This indicates that you are a premium user of Freepik. This is the credit user that I have spent already. 158000 carries left in my account and I can use it anywhere on Freepik on the spaces image generator, cinematic shot, video generator, upscale anywhere I want to. Once you click here, and you will see uh lot of new option that has been added and they are actually it on weekly or monthly basis. They have now reach out to three D scenes and three d generator. We are going to talk about it everything one by one. Now, I will show you around. Imagine we are walking into a massive three D supermarket. I want to show you every single Ashley so you know exactly where the good stuff. Right in the middle of the scene, you will see a giant search bar. On the left bar, you will see so many options here. You can minimize or maximize this depending on your uses also a search bar here, once you click it and you will see so many options that will pop up. So we are going to let's simply click on vector, and once you click vector, it will specifically search for vectors. Now once you search for bunny and it will show you so many bunny that is created in vector. You can check for the air generated bunny, license bunny orientate, or you can check for the orientation. If you want a horizontal vertical square or panoramic, it depends on your project. If you want a specific color, you can do so a people file type style advance. Once you clear this out, you will again, click here, and you will see so many of Sincere create a image icon. Now we are going back to our homepage, just like again, Freepik. And now, depending on your project, you can select for the Illustration, photos, PSD files, templates, mockup, videos, three D, audio phones, anything you want. You can search it in its massive library. Also, you can check for the Spaces, image generator, video generator, voice generator, assistant old tools. And once you hover over old tools, a massive mega menu that has been pop up in front of you and all the options that you can use within the free pick. Also, once you go to other Spaces design editor, mockup generator, icon generator, background remover, skin enhancer, change camera, rete sketch to image. There are so many options, we are going to look into all these options one by one. And let me tell you you can test this option within your free account also, or if you want to test it out, they will also provide on a simple monthly basis subscription, and they will give you some cart so you can test it out. You will get the amount of cart and you can test it out. And if you do not want to test it out, use your credit. You can simply watch all my lectures and see which model is best for you. After that, you have at least the idea to use the credit, and where you have to credit, you will get the idea of it. And let me tell you, after completing this course, you will have confidence to use Freepik AI suit. So let's start by a simple image generator. First of all, you can create a project if you want to create a specific project for your specific lint, you can create a project here, and it will automatically show you the images and videos history related to that project only. But right now, I have simply created a new project of AI generation, and once you click it, you have to simply hover over to Image generator and click it. And it will automatically redirect you to a new window that is image generator. 3. First Steps in AI How to Generate Your Very First AI Image: So this is the image generation that I have already done, and it is not image generated. It is actually I have used this image for another case that is about Adobe Firefly, and I have tried to expand it from left to right side. Also, this is the image that I have used, and also I have created a video out of it. It actually created my face and turn it into this with a creator Hodie. And I don't like the face and it's not actually my face, but yes, it's somewhat similar to mine, but it's okay. It's AI and we can't say that AI is perfect. Once you click it, I have actually expand this from left to right, and I have uploaded this image. Now let's try to go to image generator, and one by one, we are going to actually test all the models that available, and one by one, we are going to test. It might get boring, but it might not get boring because if you are enthusiastic enough to see all the generation and what output they give and what type of cinematic or what type of image generation they give according to the model, you will understand which model you are going to use for your next project. So it is really important for you to understand image generator of which AI model give you what type of result. Now let's start from the auto generation. So it is just a suggestion though it will automatically check your prompt and give you a desired result. But first of all, but first of all, improve short prompt and generate diverse variation. So this is the AI prom. Let's start with the AI prompt by toggling it off. Just a simple image generation we are going to use. What we are going to use, let's start by acute bunny, for coffee, four, lady. Now let's generate it, and let's try to generate this image in the ID screen so that we can look into it better. And also, you have to select for only one generation as it will give you so many generation depending on your model. And I recommend you to use only one generation because it will deplete your credit really fast. So let's close this bar on the left hand side and we have the full view. Also, if you want to delete this one, you can delete. If you want to save it, you can do so. And if you want to move this asset to a new folder, you can do. So if you want to publish it, you can do so. If you want to save it, you can save it as template element or save as character. It's up to you. We are going to look into it every single option, we are going to look into in future generation. So now let's generate it. And it's actually using 50 carats of mine. And once you click it will 100 go to 100. Now it will go to 200. So I recommend you to use generate only simple one generation. So now let's generate it with a acute bunny ports coffee for a lady. Now let's generate. And let me tell you the generation is really fast and sometimes it get really no carat word use and moderated. For our lady. So now let's make it, wrap it in a small house. Now let's try to generate it again. If it get moderated, I have to actually change the prompt. Model it has used, it has used Google image entry, and I really don't like the result. Now let's try to use our specific model. Now let's try to make it a cinematic look. And with the cinematic, let's try to add a camera here. What type of camera we are going to understand which type of camera and which type of focal length it will give us. But let's not go to this really advanced feature. Let's try to simple image generator I want it. Flux two P. Now let's try to make it possible. And I want image. A simple image I wanted. With a simple model, Flux 02 Pro, I want a simple image that I have imagined of and let's see. So this is a prompt that they understand, and I want it to be acute rabbit coffee, not let's see, pour coffee for a lady in a small house. Let's try to generate it. Now it's understand better. Now let's try to make it a little bit dramatic by using a simple AI prompt. Once you use AI prompt here and let's try to generate here the difference of image generator, AI prompt enhancer, when the AI prompt is of a result. But when I use AI prompt, it actually gave me the result, and this is what I wanted, and it actually really cool to 4. Storytelling with AI How to Generate Consistent Characters for Your Projects: Now you have understand the difference between this Toble switch, and the Flux door two pro is really awesome. But now let's try to option that is upload image to describe. Now let's try to find a bunny. Within the Phi Peak, let's try to find cute bunny. Now let's try to find it within the Pipek and it has given me an illustrator. Now let's try to find and vector if we can find it. Now let's try to find it in a three D model, if we can find it. No. Now let's try to all images and only AI I wanted. Yes, this is a cute puny this is cute. This is cute. Now let's try to use any of the bunny here. Let's try to use this specific bunny. I want. This is cute. It has a cute eyes. Now let's try to download this image. Once you click it, it will automatically pop up a window and you can download this image depending on your specific need or the size. Now let's save this one. I have saved this image, and I'm going to upload this into the image description. Now let's try to upload it here. You can simply drag it here or you can simply the image that you have already generated, you can use. But right now let's drop this image into this area. Now, once you drag it and use this image, as the image has been used prominently, and it has been described as a prompt. And also, let's try to shift enter once you enter it, and you will see and you have to add bunny pour coffee for a lady. Now let's try to see generate again. And within just seconds, you will see the generation here. This is the bunny and this is the actual bunny that we have used, and it's not actually working, right? So now, don't worry, we are going to dissolve it in our images, understand how all these thing prompt. This is to give an idea that you can use this prompt, how you can familiar with this area. Don't worry, we are going to learn everything about it. We are going to use that same character bunny in our image. Now let's try to use. So we have used this option, so now let's try to go to prompt Editor, on you go to prompt Editor, ask me about prompt, or you can simply ask for then prompt, once you click for the Nance prompt, you will see a simple magic happen here. Now let's try to apply prompt or if you want to make a more detailed prompt, you can do so. And if you want to add a image for description, you can also use it as a prompt. So we have used prompt and apply this prompt, and now let's try to generate. Again, and I have I have 6,309 images remaining to generate. Now let's try to generate again. So this is a image that has been generated, and Rabbit O Boni is actually serving a coffee for a lady. Now let's try to use now let's try to remove this prompt or if you want to generate with the seed, you can do so. If you want to fix seed, let me tell you what is seed. This is unable this to get consistent result every time you use the same prompt. If you want a specific result, every time you use this prompt, like we have generated this image, if you want specifically this image for every time you generate using Flux two P, and this is the prompt and you want this result every time, you have to disable this seed option. And if you want to and if you want a different generation, every time you click on generate, you have to enable it. Once you enable it, you have to remember the seed number. With the same seed number, you will get the same result. But with a different seed number, example, let's say, use five, you will get a different result. Once you click here, you will get a different result here. And also, if you click here and again, type four, you will get the same result like this. What is the use of this fixed set and variable set? It is used for consistent character and the image generation, consistent image generation. Now let's try to while generating, let's understand what style, character, and additional information. Look, now let's start by clicking EtyleOnce you click on Estyle, you will see which type of style do you want in the image? Which type of character you want in the image. And if you want a design or a three D, you can also select the type of the image generation. All these are actually preset and you can actually test it out and run it out with the image. And let's try to make it a classic animee style. Let's click it, and this is the style I wanted. Sometimes you will see this type of error within the Freepik or anywhere on the Internet that image generation and video generation, you will see this type of error sometime. But why this happened because at the bank of this image generation, there are so many calculations. So they might give you error sometimes like this you get here, right? So sometimes it's work, sometime sometimes it will not work or it will actually not give you any results. So you have to try it again with the clicking on generation. And let me tell you that is I love about FEI PI. Once the generation is done, they actually decrease the amount of the credit from your account. But if it not actually generated anything, they will not actually deplete any credit from your account. This is what I love from Freepik. It has changed the variation of the image. And yes, this is for the reference, we have used a style here, like how we have selected. This is how we have selected here. And for the illustration, I have used this classic anime, and if you want to use any of the media or the type of the image, you can do so. Now let's try to minimize the window and now simply generate this image. We are going to understand everything here. After that, we are going to jump to different models one by one with the same prompt, so you will understand better. Now, this is the image generation we got. The anime style is really awesome. Now let's try to decrease this one. And now let's try to use a character a and let's remove this one, only a character we want here. For the character, let me upload the image here. This is and this is the image that I want to use as a character. Now let's try to add a character here. Now, let's try to generate. Now the image has been generated, and let me tell you this is how you can use the same character in your image that you have downloaded. Now this is the same, and this is the same. So before we have tried to use a prompt and described our image, right, and it has gave us a bunny, but not the same or exact bunny that we want it, right? But right now, we have explained the AI model that we want This specific character for our image. Now it has generated that also, you can also use let's delete this prompt and also you can use that once you press Add at, and it will automatically add a character here once you click here. The image character has been addi here within the prompt, and now let's add. So now you don't have to add a bunny or rabbit hair. You simply have to type pause coffee for a lady. Now let's try to generate. And why using an extra feature, we got the generation of 75 credit. According to according to the future, they charge you different credit. Now you have not actually specifically tell what type of bunny or rabbit I want yet. Simply add a simple prompt, and the image is actually really good, right? The bunny has posed the coffee already for the lady, and I like it. Now let's try to. Now you have understand the very basics of this prompt generator, and you can also add another character if you want to add in your character list or into your prompt, you can do so. Right now, my objective is to explain you how this prompt actually works. Now you can actually delete this prompt and use another prompt. Here, thank you for watching. We are going to explore every model generator out there one by one and understand every generator what type of generation it give and when to use this generation, we are going to explore them. See you in the next lecture. Thank you for watching. 5. Understanding Your Models: I want you to understand what each model give you and how you can actually test the model and how you can understand a model. You have to simply go to the model website like mistake 2.5 model. Once you go to this and you have to actually check for the information on the Github, on the hugging face, anywhere you want, you can actually you can actually check for the information, what type of high quality images they give what type of information they are using. After that, you can actually and determine a specific model for your project. So now let's for the mystic AI is cutting edge imagination model. We are actually working. So I'm trying to find out the information about mystic. I'm trying to find out and I'm trying to teach you how you can find the information about different models. Right now, I'm searching for mystic, and this is how you can dig into the Internet and find a specific model in information. So this is a high quality aesthetical pleasing image with minimal effort designed to perform. Effectively, mystic I can generate a stunning visual as few as eight to 16 inter fans steps. So this is what it tell me that it is using a simple step of eight to 16, and that is why it gave us a really fast result here, here. And the results are also really cinematic. I loved it. Now, once you check this one and make powerful tool creative professional enthusiast like, so it is a high quality output, efficient interface, and versatile application, idle for raising for creative projects, including fashion photography and portrait art and more. So it is really good for fashion photography and portrait, like we are using here. I actually good for portrait, right? This is how you can check for the specific model that you are using or if you are using it into your CFUI or offline project, you can do and understand and test it out and understand and test it out with the different model. So now let's try to go with AI model with the 2.5 fluid, and let's try to generate this one. 6. Different model different output: So I'm going to test a single prompt that Gemini has given me, and I'm going to use it with every model available within the free peak. So there are 40 models available right now and we are going to test it. Now that's how we can test it. And let me tell you what is the prompt here. So this is prompt a cinematic portrait of a guardian warrior made entirely of moss ancient distort and glowing teal fungi. They are standing in a mist, dense forest, rain is falling, creating tiny water splashes on their stone skin. The armor is incredibly carved from the dark rock, dramatic moody lightning, macrophotography, style, hyperalistic, and eight K resolution. Now we are going to use this specific prompt for every image generator out there. Within the free pick. Now let's try to use it. First of all, let's start from the Flux dot to P. Let's try to generate it. And also, while it's generating, let me tell you, you can also increase the size of the image. Once you increase the size, you can also, and the carat will also increase. But right now, we want it in the four K resolution and all and let me tell you, and this is the image that has been generated. It looked mythical and it looks really cool, right? Now, let's try to cut the image and now let's try with a cinematic look and try to change the clean and try to change the cinema digital. I'm using a camera that is cinema digital 35 and with the prime lens classic, and it will give me a tap with 200 M. No, I want it with a 35 M with a crop. And right now, let's try to make it so it will give me a really great bouquet effect behind my subject. You can actually try it out with different different focal length aperture, lens and camera. You have to try all the combination here and you will understand how you can use it. But before that, you have to understand how camera works and you have to understand how all these camera actually do in real life. After that, you will understand which type of image generation you want it. Let's remove this reference, right? Remove this reference and generate. Our image generating, let's try it with again, Flux dot model. Why? Because I have used that character and I don't want it to be actually affect my image generation, and let's try to create again. Meanwhile, so you can give more command to generate also, generation, while it's attempting to create your image with the one generation, you can also try it with the second generation. Also you can queue it every time, and it can actually do a lot of generation at one time. So you don't have to worry about that. So let's try to click it and adapt or field it's cool, right? Yes, image generation is really amazing. Now let's try to cut it and see our second image. And now this is the image that has been generated. In the reference image, we have forgot to remove the bunny, right? So it has generated this type of image. But yes, gave me a really good result. Now it has created a human figure. Now let's go to our second generation and Google Nano Brana True. And again, let's try to generate it. With the Google Nano Brana True, let's use another model and C dream five light. Let's try this one, and it's actually using 50 credit. Again, we have made it in Queue, and meanwhile it's been generating, let's look into this. It's good, right? Now let's try to remove this one and using a different model here, meanwhile, CT AI is working, and we are going to us Flux two max. Now let's try to generate with this one. And the generation of Flux Flux two max is 195 arts, right? Now, let's try to make it one K, and it's actually using 130 arts. The Flux two max is really going to be a great model, right? With the C dream one, but with the C dream, the generation is actually good. It seems like a gaming character. A powerful game character, and I can see it's going to be a villain. So this is so this is the Flux model we have used, and this is the Flux pro. The image generation is really dramatic. As we have described in our prompt, that interact with cave dampd moody light, and like we have described, that creating tiny water splaces on the stony skin, it's working really well, right? Now let's try to click this one. Yes, it's done really well. But it has the texture, the texture on this one, and on this one is really different. The lightning effect is actually different. That is why it's charging so much, right? Now let's try it with another model with the Flux dot two flax and now let's try to generate it with 80 credit. This is the result we got from the flax. Now let's try with another now let's try with another model that is context Max. And now let's generate it and it's using 150. It has given me a result that we have never expected, right? It is a really cinematic look, and it's really awesome to see. Now let's try with a different model. So as far as I can understand, the Flux Context Max has given me a different, really different result now let's try to use KlNFlux Flux dot to Klan, and this is the result that we got. And let me tell you I don't want any prompt to interact with me, right, with my prompt. Now let's try to generate it again with the Klane, and let's see what we got here. As far as I can understand, the AI prompt is actually interacting with my prompt and I don't want it to be happen. This is the Klan before and after results are not bad at all, but I love this result. Now let's try to use a different model that is Flux Context Pro. And let's try this one. So the Flux context is actually giving us a result that we never expected. So if you want a result that is totally different from the same prompt, but from a different model, if you want a different result, totally different result that you can't even imagine or AI can't even imagine. You can use Flux Context P. And it actually gave me a really awesome result. Now let's try to generate it with another model that is Mystic 2.5. Now let's try to use this one. Mystic 2.5 has given a result. That is, it looked like iron man made of stone, but I loved it. The results are really good. Mysti has given me a different result and I loved it. We can use it in our upcoming prompt testing. Now let's try to use it flexible, 2.5 flexible mesti. And let me tell you we are going to understand all the model one by one. 7. Testing Advanced Models and Style Checkpoints: Result from misty 2.5 is look really realistic. While looking at this one, it looks like it's coming towards me, and anytime it's going to play like a video. It's so much realistic, I love the result. With the Flux, one, the generation is using only ten cadets. Let me tell you, Flux one, Flux two, there's a lot more difference in between them, and Flux one is really fast to use, and you just create it within just seconds, right? Information is not that much good as compared to Flux two, but the results are good. Also, is really fast. Ten or 5 seconds just generated the image for me. Flux 1.1. Let's try to generate this one. Meanwhile, it's generating let's try to use another model. Because Freepik is really fast to generate. And now Flux one realism, let's try to generate it. Meanwhile, we are generating, it actually generated another image. It's really fast to generate. And with Flux 1.1, the results is awesome. We can see so many images that has been generated, and we can actually differentiate what type of model we want it and for which product we can use. The graphic design, negative prompt, let's try to idiogram also generate it. Meanwhile, idiogram is working on our prompt. We have to check for another model that is Google image and three. Also, let's try to generate with this one. Let's try to again, generate with all models en four, let's try to generate this one. Idiogram has cia result actually created a human face, and this is what I loved about this model. I actually work on portrait. And now let's go to this one, Google Image and three, it has generated a vague result, but as far as I understand, it's good. Because of the contrast, the details, it's not very visible. And about this one, Google Image and four, it looks like that beginner Photoshop artists have tried to create a image or tried to add a creative. As far as I understand, it looks like image and four is really fast, but it has generated a really vague results and I don't like it that much compared to all these generation, I don't really like it. But now let's try to again, generate with the model ImageN four, and let's try to generate this one. And again, let's try to generate after the image Gen four, we are using ImageN four ultra. Let's try to generate this one and with the image Gen four. So it has generated a gladiator for me. It's look Cool, but not that detailed. So now, image and four is also good, but let's try to use Ultra. Okay. Ultra has given me a different result, and I loved it, yes. But one thing I don't like about the image generation from the Google is that it actually gave me a result is that has so much contrast. So because of that, the details we might not able to see because of so much shadow and so much and the shadow has a contrast, the darkness in the image, right? Now, let's try to use another model that is Google Nano Banana, and let's try to generate with this one. And meanwhile it's generating, let's try to generate with the Nogle banana P, and again, generate Google Nano Banana true. Let's try to use Google Nano Banana. No, we don't have Google Nano Banana two right now, but it has shared us, so let's try to use it. Now we have used it. Now let's try to generate with Google Nano Banana two. And meanwhile, Google Nano Banana has given me a result. So it looked like a alien predator to me, and it's look good, not bad. But the results are not that detailed. But Google Nano Banana Pro is really good. The results here and the details. And now let's try to use a different model. Meanwhile, Nano Banana is working on our image. Now let's try to Flux one fast. Let's see if we have selected Flux one before. Yes, we have Flux one now it is Flux one fast, and its generation time is actually 7 seconds. Wow. Now let's try to use this one and generate. Let's try to use GPD and generate. Again, Google Nano Banana two is really slow right now, but meanwhile, Flux has given us a result. It's not that detailed, also because it might be a four step image in the generation, but results are okay, but the details are not good. The detailing is not that much up to the mark. Now let's try to check this one. We have used a GPT. Cool, it has a different aspect ratio. But yes, the results are actually it's looked like a statue. Don't worry, but still working. Okay, no problem. Now let's try to use another model that is Flux GPT one HQ, and let's try to generate it with the 500 karats, it's using actually 500 karats. Wow. So this is really important for you to understand that GPT one HQ is actually using 500 karats. It's way too much, right? Now let's try to use it from a standard the two ratio, and Okay, so Google Nano Banana two has been failed to generate the result. Now let's meanwhile, generating again, let's try to use a GPT four. Let's see it. And also it's actually getting really interesting for me to see what type of results they will both give. Google Nano Banana two has given me this result, the details are good, it's actually a good result. But again, like we have discussed before, Google image generation is not that much good because of the contrast of the images. But right now, the GPT HQ has generated our image. That is actually really detailed one. I love this one. It looks really realistic, and it has been generated within 500 arts. Now let's try to use another model that is GPT 1.5, and let's see it's using 150 arts. Meanwhile, it's in Q. We have to check for another model, 1.5 his it's using 500 arts again. Wow. Only Image generation is 500 arts. We can actually generate a video out of these carats. But don't worry. Let's try to generate this one. But also, with the GPT 1.5, it is really good. The results are really good. I loved it. The details of raindrop itself is really realistic. And we can see that and we can tell that it has been captioned from the camera, right? Now, let's try to use another model. Meanwhile, it's working on another model. Let's try to use C dream four. With C dream four, we are using a wide screen camera. Here, 50 generation with the model of C dream four, let's try to use four K, CDream four K, and let's try it again. GPT 1.5 hi also has given me a result. The raindrop. The raindrop is actually a realistic one. I loved it. No doubt about it. They have generated within 500 arts, but they have gave me a result that is more satisfying to see. Wow. Now, CDM four has given me a result that I like it with the horn. Yes, it's actually a different result. It is actually me and so now it is actual gaming character. A realistic gaming character, right? Now let's try to see a see dream four K. Yes, it is kind of a foggy mystic image, dreamy image, but results are good to see. Now let's try to run it with a different model that is Okay. And now I'm using 4.5 generate. They have marked runway as duplicated. So right now, image generation using runway is actually with a less value. This model has been duplicated this month please use C dream 4.5, so we can't use runway this time. Why? Because the runway model has been duplicated because maybe of less popularity, they have deprecated it. But now let's try to use a classic first, and let's try it with the generation of one. Whenever I see a 500 carats, it's really bad to see, but whenever I see only one carat is really good to see. But classic fast has given me a result that is almost a good. It's really good. No doubt about it. It gave me a fast and really good result, right? But the detail is not that much good, but the results that it gave me and the speed in which it has been generated is awesome. It's really cool to see it. And one more thing, see dream 4.5, it's good. Now let's again, with the model of classic First and we are using classic, and I'm using only five Carter with the classic model. I recommend you to use classic First so that you get the idea. Now let's use mystic 1.0 generate classic results. This is really bad. I recommend you not to use this one with a classic. You can use a classic fast. It actually gave you a good result. So now you can see mistake 1.0. The results are good. Let's try to generate with the Z image model. Now let's go to our in and generate. Z image has given me a result. It's good but not that detailed, right? So now check for the Quin image. Quin has given me a result. It's okay. No doubt it's okay. And with the defend model we have only three model to check. Let's try for the rock one and generate. With the defend model, let's select, meanwhile, re craft V four, and 60 carried with the models, let's try to use it with a pro. And these are new models they have introduced. And also grok results are good. It's actually good, right. Now let's see what it has given me. So V four is really good to see. No doubt, it's really awesome to see this result. It has actually given me a different result. Re craft v4p, the results of this one is totally different out of this world. The way it holds its hand and the way it sitting inside an is really love to see. The details are really good. The results are really hyperrealistic. 8. Manage your generations within UI: So now we have generated all the image using all the models. After that, if you want to delete a specific one or if you want to delete all of them, how you can do so. You simply have to you simply have to go here. Once you hover over on image, you will see a circle here. You simply have to click it, check, check, check, and the information will pop up. If you want to upscale specific image, you can do so and you can upscale up to 20 image at once. You have to select 20 image and you have to upscale them, and it will upscale for you. You can check here also. If you want to check one, two, three, you can either delete, you can download from here. Once you click here, if you want to download it in PNG, JPAC or original, you can download according to your need and according to options given. Or also you can delete them at once. This is how you can move to Trash, and after that, it will go into your history or trash folder. Once you go to your history, your favorite ones, if you have one, or you have a history, you can do and if you have any upload, this is the upload. All these are uploads. And after that, if you want to specific download, do you have any creation that you have downloaded recently here, and here is a trash folder that recently you can see anything that you have moved to trash. So both these images, I have moved to dress. If you want everything to be deleted, you can select and delete them. And if you want empty trash, just like Google Drive, you can empty your trash at once. You simply have to permanently delete everything in your trash. So this option will permanently delete everything in your trash. Now, go to our history. If you want to see everything in your homepage, you simply have to go to homepage and all the generation. Here. And if you want to if you click all, again, a window will pop up. And if you want to move them into a specific folder, you can do. So you can move into a specific folder, just like Google Type, it is working like that. It is not difficult to use that. So right now, I'm selecting this this one, two, three, four, five, six, seven, eight, nine, ten, 11, 12. All these are generation you can simply click and select. After that, we will going to download them at once. Download all the 50 ones in their original format. Let's click Download, and it will going to download. Your folder and a zip folder has been created. Now, save it here in Picasso and extract all. A folder has been created, and now you can see your generation. All my generation has been downloaded at once. It is really easy to use, right? 9. Overview of Video generation : We have tested out image generator. Now, we are going to test video generator for it. Let's go to Video generator. And once you go to Video generator, you will see many models are available for the Video generation. Just like image generation, there are so many models available for video generation also. Video generation is somewhat advanced. So we need to understand how video generation work and where we have to use video generation because video generation uses at least 500,000 or 2000 credits from your account. And I recommend you to choose a video generation that is best for you. So I recommend you to first read about every video generation or every model first, which type of model give what type of output are going to so according to your need or your project, you have to specifically use them. So here's the prompt that Gemini has given to me, a happy golden retriever dog running fast across a wet sandy beach at sunset. Water is splashing under its paw camera tracking alongside the dog cinematic golden lightning and that this prompt is really detailed one. So we need a specific model, a video generation model that will work specifically with camera tracking an important one and the lightning if it says wet sandy beach at sunset and water splashing, we need a model that work really well with the small detailed graphics. Everything has been covered within this prompt. This is a lightning, camera tracking, and detailed graphics detail. It's work with photo realism and what I want, I want specific video model that work really well, and we will compare it with other model so that we can understand that our research is working with the model or not. So let's understand the model first. So we have to work hard for the video generation model. Unlike image generation, it is really easy to do that. In video generation, they are charging us the credit. It's about TNAX the image generation model. This is the latest model that has been introduced in Freepik, and it's working really well with the graphics. The details they have showed. And if you are using a camera tracking, you can use this one also realistic videos, and camera tracking is working really well with it. With this one, it's not for the camera tracking right now, clink three and camera tracking, photorealism and sand detail in their output. Yeah, now we can see better by zooming in, it's working really well. Okay. With this one, it's 2.7 cling 2.6 motion control. It will not work for us because it is working on motion control and we have to give our output or we have to give a reference video so that it can actually copy the motion as it is for motion control. And it is for realistic videos, Kling Kling 2.5, it is for realistic videos. Kling 2.1 will not work for us Kling 2.1 masters. I will not work for a specific model, just like the detail we have added, it will not work like that with this different camera angle, Google Vo three with the lip sync, we don't want lip sync right now. Google Vo three is duplicated, so it has been. So some of the model has been duplicated, they are not really well with the generation, so we are not using them. Yes, it will work a little bit with seed dream, see tense, 1.5 pro. And yes, we can see the output here is really good. For this one, yes, it is really good. But it will work with the motion control. It can work with seed dream fast. Here is a detailed one, Sedran. It will work with human figure, but we don't want it right now. It's good but not that much good. I can see. We fabric 1.0 fabric here. I will specific work for fabric also. Two weed fabric, minimax, hello, illustration and animation. And is Pix six realistic video, Minimax, Yes, it is Minimax Hello two, and it is actually good. This is a fast motion control. It is good for time lapse video. I don't like the result of it that much. This is good. So it has been generated to Pixes. Okay. Pick into the video has been created into Pixel world. The pixel was 5.5, illustration and animation Native audio. You can add an audio here. It is for Native audio one, 2.5. For 2.2, it is good for camera tracking. Animate, move, cool. Good for camera tracking, I guess. So right now they have introduced us with the camera tracking, and I can see that the generation with this output is going to be good. So now let's try to generate it with the Pacific prompt, and we can see if they're in generation. Joe, they are using 1,400 of generation, let's try to reduce the seconds here so that we get a little bit of 480. Now, the generation has been reduced to 533, so we can upscale it later but right now, let's try it with the simple generation. Now let's try it. Let hit generate. So content acknowledgment by continuing confirm that you all contain upload or generate character, use of tools of generate connect that violet third party we is strictly prohibited. Misuse may result in immediate suspension of your account and potential legal consequences in accordance with PP. So content acknowledgment has been so there is a term and condition for us to see that they have a specific third party tool that has not been generated with the prohibited of third party hikers. So we have to simply check this. And just like everything, every company will show you. So it will take about three to 5 minutes for generation, and the bar has been increases. So when Freepik shows you a third party prohibited oning, it means that you ask the AI to draw something that belongs to another company or person. So it can be a famous character, real celebrities or politicians or brand logos and products. So Freepik put this block in place to protect you. If you generate and sell an image of Batman, the creator of Batman could show you. So the AI simply refuses to make it to keep everyone safe. So you have to use the specific image or character if you have the rights to use that character. Got it. So now let's check the output that that seed dream has given. With the sound, it's cool. The detail of this is really good. The video generation is good. So now let's try to compare it with the different model. For the video generation, I love killing models because it's work really well in my case. So right now they have introduced with a Croqlo let's try to use and generate generates 320 carats of generation, and it is same for both the generations. And we can use 1 second, 3 seconds. Let's try to use it 3 seconds and generate. The generation time is really fast. The result of Croc, this is really good. If you want a cinematic look, how see dance 2.0 is working, a cinematic look. But for the photorealism, I think Grock has the upper hand here. For a cinematic look, I go for the seed dance and for the photorealism, I go for the Croc. And we can test it out with every model, but right now the result we wanted has been given from both the model, and we can use either one from grok or either from the C dense 2.2 dot. Oh, so we are going to generate so many things in upcoming lecture. Same in the next lecture. Thank you for watching. 10. Generation your first Ai video : We have tested out image generator. Now, we are going to test video generator for it. Let's go to Video generator. And once you go to Video generator, you will see many models are available for the Video generation. Just like image generation, there are so many models available for video generation also. Video generation is somewhat advanced. So we need to understand how video generation work and where we have to use video generation because video generation uses at least 500,000 or 2000 credits from your account. And I recommend you to choose a video generation that is best for you. So I recommend you to first read about every video generation or every model first, which type of model give what type of output are going to so according to your need or your project, you have to specifically use them. So here's the prompt that Gemini has given to me, a happy golden retriever dog running fast across a wet sandy beach at sunset. Water is splashing under its paw camera tracking alongside the dog cinematic golden lightning and that this prompt is really detailed one. So we need a specific model, a video generation model that will work specifically with camera tracking an important one and the lightning if it says wet sandy beach at sunset and water splashing, we need a model that work really well with the small detailed graphics. Everything has been covered within this prompt. This is a lightning, camera tracking, and detailed graphics detail. It's work with photo realism and what I want, I want specific video model that work really well, and we will compare it with other model so that we can understand that our research is working with the model or not. So let's understand the model first. So we have to work hard for the video generation model. Unlike image generation, it is really easy to do that. In video generation, they are charging us the credit. It's about TNAX the image generation model. This is the latest model that has been introduced in Freepik, and it's working really well with the graphics. The details they have showed. And if you are using a camera tracking, you can use this one also realistic videos, and camera tracking is working really well with it. With this one, it's not for the camera tracking right now, clink three and camera tracking, photorealism and sand detail in their output. Yeah, now we can see better by zooming in, it's working really well. Okay. With this one, it's 2.7 cling 2.6 motion control. It will not work for us because it is working on motion control and we have to give our output or we have to give a reference video so that it can actually copy the motion as it is for motion control. And it is for realistic videos, Kling Kling 2.5, it is for realistic videos. Kling 2.1 will not work for us Kling 2.1 masters. I will not work for a specific model, just like the detail we have added, it will not work like that with this different camera angle, Google Vo three with the lip sync, we don't want lip sync right now. Google Vo three is duplicated, so it has been. So some of the model has been duplicated, they are not really well with the generation, so we are not using them. Yes, it will work a little bit with seed dream, see tense, 1.5 pro. And yes, we can see the output here is really good. For this one, yes, it is really good. But it will work with the motion control. It can work with seed dream fast. Here is a detailed one, Sedran. It will work with human figure, but we don't want it right now. It's good but not that much good. I can see. We fabric 1.0 fabric here. I will specific work for fabric also. Two weed fabric, minimax, hello, illustration and animation. And is Pix six realistic video, Minimax, Yes, it is Minimax Hello two, and it is actually good. This is a fast motion control. It is good for time lapse video. I don't like the result of it that much. This is good. So it has been generated to Pixes. Okay. Pick into the video has been created into Pixel world. The pixel was 5.5, illustration and animation Native audio. You can add an audio here. It is for Native audio one, 2.5. For 2.2, it is good for camera tracking. Animate, move, cool. Good for camera tracking, I guess. So right now they have introduced us with the camera tracking, and I can see that the generation with this output is going to be good. So now let's try to generate it with the Pacific prompt, and we can see if they're in generation. Joe, they are using 1,400 of generation, let's try to reduce the seconds here so that we get a little bit of 480. Now, the generation has been reduced to 533, so we can upscale it later but right now, let's try it with the simple generation. Now let's try it. Let hit generate. So content acknowledgment by continuing confirm that you all contain upload or generate character, use of tools of generate connect that violet third party we is strictly prohibited. Misuse may result in immediate suspension of your account and potential legal consequences in accordance with PP. So content acknowledgment has been so there is a term and condition for us to see that they have a specific third party tool that has not been generated with the prohibited of third party hikers. So we have to simply check this. And just like everything, every company will show you. So it will take about three to 5 minutes for generation, and the bar has been increases. So when Freepik shows you a third party prohibited oning, it means that you ask the AI to draw something that belongs to another company or person. So it can be a famous character, real celebrities or politicians or brand logos and products. So Freepik put this block in place to protect you. If you generate and sell an image of Batman, the creator of Batman could show you. So the AI simply refuses to make it to keep everyone safe. So you have to use the specific image or character if you have the rights to use that character. Got it. So now let's check the output that that seed dream has given. With the sound, it's cool. The detail of this is really good. The video generation is good. So now let's try to compare it with the different model. For the video generation, I love killing models because it's work really well in my case. So right now they have introduced with a Croqlo let's try to use and generate generates 320 carats of generation, and it is same for both the generations. And we can use 1 second, 3 seconds. Let's try to use it 3 seconds and generate. The generation time is really fast. The result of Croc, this is really good. If you want a cinematic look, how see dance 2.0 is working, a cinematic look. But for the photorealism, I think Grock has the upper hand here. For a cinematic look, I go for the seed dance and for the photorealism, I go for the Croc. And we can test it out with every model, but right now the result we wanted has been given from both the model, and we can use either one from grok or either from the C dense 2.2 dot. Oh, so we are going to generate so many things in upcoming lecture. Same in the next lecture. Thank you for watching. 11. AI Character Design How to Create and Concept Your Own Character : Now let's try to generate a character. So I simply ask that I want a character just like big hero six, but different. So help me with imaginary character generator for my course so I can test run prompt on it. Character should be like my big hero six, but I want it to be with texture different. Can you suggest some options? And I will ask in the end, give me prompt for it. Let's go for the ChatGPT and ChatGPT has given me some options. Copy this one, and in our image generator, let's go for the specific character and for the models. I think has photoism, I want it with Flux two P, and it's using only 50 character. But without AI prompt, I want it will work or not. The character is good but not that much good. Now let's try it with a different model, and if it will not work, I will ask for the different character. But with the Context Max, I like this so much that it has given me a different characteristics. Let's try to make it. Using Context, Max, and I hope it will give me a different result. Other than this, at least. Yes. As expected, Flux context is really good. So we are going to use this character at least for our image. Also, let's try to see this one, the lava energy core, a mechanical hybrid. Let's see if it will work with contracts pro and I want the contract pro to work for me with this character also. Let's generate. No, I don't want it. To be this is scary, I want it to be like this. We can actually use this as a character, also, right? A mechanical hybrid. Let's see if it will work for this one also. Actually, I'm trying to generate a simple character so we can use it in our upcoming lectures. I'm sharing you all the process how we can generate a character and how it will work for us. But right now, no, it's not working at all. It's doing something else. Now let's try it with the different CD light. Let's try it with this one and with the Okay, let's try it with this see dream. Basically, my motive is here is to show you how we can generate a specific arrector and use them in our video. And if there is any specific advertising video or anything else, you can use that character and generate a specific advertising video, product video, any kind of video you want to generate. It has generated something but not up to the mark, and the generation that I got here is really cool to see. So what we can do here, we simply have to copy this prompt and paste it here. After that, we have to paste the style he just like lava energy core, fabric mechanical hybrid. Let's try to use lava mechanical core hair and paste it here. As powerful presence. I think that no, it will not work like that. And I think at the first result we got here is really good to see. We have to use this with our generation. So how we can use this first of all, let's download this download this, this, and this. And this. Okay. And this. Now let's download Extract All. So this is the generation that we have created, and I like this one. And we can actually use it within our free pick. I want it to be edit it first, how we can do that. First of all, you have to click it, and if you want to edit the details, you can edit, use this image, edit it image. If you want to image if you want to edit the image, you can use edit image. Once you go to your dit image, you can see there are so many options here. And if you want and once you click here, you can see the lightning top, front, right, left, back, bottom, front and top. We can generate a image by using lightning also. It's really cool to see. And wow, it is a new option that has been introduced really quick. So first of all, let's go to this. And if you want to see any change, you can do so. And if you want to see by using any prompt, you can actually change the prompt by typing the prompt here. And if you want to see anything by using visually, let's try to click here and see, remove this robot. A remove this robot also. And now with this one, also, you can actually remove this by using this also. There are so many options to remove this robot and you can simply remove by selecting this specific area, and it will automatically remove or expand for you. And if you want to remove this one, you can remove this one also. Now let's try to erase. And once you click Generate, it will automatically select a model for you and automatically remove the character. So the generation has been done, and you can see we have a specific character that has been generated. Now let's save it. Save and close, save and continue. Let's go to save and continue. The save changes have been done, and now you can download this. Let's cut this one and export. Export in SVG, JPCPNG I want it in PNG, let's download this one. And I have downloaded it in my folder. Oh, it has not been downloaded, but Okay, let's try to make it back. And this is the image that we have saved. You must have to save the image as it will not appear in your generation history. So you simply have to click the download button here and download directly into your computer. Now you can see the downloaded image. This is the edited and this one is the image. So we have generated a character that will work really well with our upcoming generation, and I will try to use it in different scenario. 12. Advanced AI Video Generation Mastering the Seedance Model : The video generation, let's go to video generation part. And if you want the specific model to generate image for you, you can do so. How we can do that, we simply have to go to models, and here is the CDM. Photo is S for the grok you can see there is a simple reference image, and it will automatically animate that image. But once you go to all models and C for the specific C dream fast and C dream and C dense 2.2, you can see there are so many references that has been generated, and you can see if you want to add image camera motion and image of your character, you can also use that. Let's try to use this image and let's try to add a video if you want a video here. No, we don't want to video to work and let's try to add we have to specifically generate a image by using this, then we can apply that image with our character. First of all, let's let's go to image generation and with the image generation model, we have to select specifically a model that will work really well with our generations. As far as I can understand, re craft Pro and re craft, these are actually good for my image generation, and once I selected it, it actually asked me a style, but not for a specific character they are asking me. So I have to select a model that they will ask me for a character. Let's see if Pro has asked me the character, yes, Flux two has asked me the character, and I have to upload this character here. Once I drop the character and at Character one, I have selected Character one is standing in front of big human like structures made of wood. They are trying to scare the character in the big dense forest. Let's try to generate the image. And I hope whatever I have in mind, it will work like that. And if it not, I'm going to test on all the prompts and we are going to see what we will get. And if there is any style, we want, no, right now, I don't want any style to work. Okay, the lightning has been changed also. Cool. Let's see. Let's try to change the facial X pression of the character change to scared. Let's try to generate. No, it's not working, but I like this image also. Don't worry. Let's try to use this image here as we go to video and with the image, let's drop it here. And with the motion control, I want this F PV drone style, and let's try to remove the prompt here. If we want to add a shot, we can actually generate different shot here, short one, shot two and shot three. So how we are going to generate the shot one and shot two for C trends, we want the help of the ChatGPT for the prompt generation. Let's add this and ask ChatGPT that I want that prompt for shot one and shot two that first character jump and hit the ground very hard shock wave appear and run towards the humanit figures, punch one of them. So first shot. So I have to define it as a first and second shot. And the character appear with the camera movement in cinematic style short to cool. Now, I hope advity will give me something. So it is just like a storytelling. You can see I will ask my character to work like that. And I hope it will work so prompt. Chat has given me some prompt. The soft drop from the sky lands with massive force on the ground, creating a powerful shock wave that surface sends out flying camera start with the top and flying. Let's try to add the first generation should be this paste, and the second generation should be this. Let's paste it here. And for the generation, I want it to be three second. And for this one, I want it to be three second. And for this one, I want it to be four second. It is A advanced prom we are using, and I hope that seed dens tupino will work with this. Now let's try to generate. And right now it's using 2000 of my credit, and let's try to make it for ATP so that we will upscale it in our upcoming videos. Let's try to generate the video. So if it will not work, don't worry, we are going to rectify how we can do that. But let's see if it works or not. The video has been generated, and let's play this video. How smoothly and how cinematic it has done, connecting all the dots and the way it has generated using just single image is really amazing. The results is really awesome, and it is just like a cinematic movie Pixurs movie, whatever you suggest, it's just like a movie scene. And let me tell you how we can actually upscale this video, how we can the there. If you want to upscale it, you simply have to use this video and go to upscale. Once you go to upscale, they will show you how you can upscale the video. This is how you can upscale the video. You have to simply select the specific model. And if you want a pre season one, you simply have to select for the pre season. And if you want anything upscale, you can do so easily. Now, let's. So this is a video that has already been updated here, and we are going to use it creatively or precision. You can see the generation here and the generation here. Let's use and use Magnific and try to use it into ten AT P. Now let's with the strength or FPS boost, interpolate fame to increase the video frame rate for smooth and motion. If you want it, you can do so, but it will also increase your credit. But I recommend it depend on you. If you want a sharpness, you can increase the sharpness. If you want a smart grain, you can do so. Review 12 frame preview. It can also generate 12 frames for your video, and it will cost you this. But I want it to generate like this. So let's wait for some time, and it will automatically scale up the video for you. We have to wait until 20 minutes. We will come back. Let's jump to the future. So the high quality image has been generated, and you can see the video with the high quality, low quality video and high quality video. Now let's download this video. And download this one also. So we have understand how we can generate different fame videos. We have understood how we can upscale the video. We are going to generate so many things in upcoming video. Stay tuned and see you in the next lecture. Thank you for watching. 13. End to End AI Production From Gemini Script to Advanced Video Generation : And we have created a two frame sequence by using CD dance to Point, which is the latest model video model generator. So now let's go to our video generator, again. And now we are going to generate a sequence of videos that is more advanced feature to use. Now let's go to your CDansTPinOF or you can check for the Seedance to Pinto. Let's go for this right now. And with this selecting, we have to first we have to look into our script. So I have asked Gemini to create a frame by frame video. Let me show you what I have typed. The video is incorrect. I have asked Gemini to create a video for him, but it not give me a right frame or the sequence that I wanted. So I have asked it, it is incomplete and it's not correct. So give me a prompt that is at least with 12 frame video. After frame by frame generation, we will look into our video generation. Let's start by going to two hands of diver comes and hold the glowing p. This is the sequence that I have in my mind and I have typed it and ask Gemini to give me a actual prom with the correct vocabulary because as far as I can understand, vocabulary is not really good according to the prompts and according to the needs of the model so that it can understand precisely what we want to portray. This is my words that I have asked to rewrite prompt prompt generation of frame by frame generation for my upcoming video. Drop come from drop off tube from the cloud. Then comes on face off lady, I will provide a picture of lady or character sheets. I will tell you what is the character sheet and you will understand how it will work. Then drop from in hand and drop into C, drop turns into pearl. Two hands of diver comes and hold that glowing pearl camera move to the person in the sud character, face visible character looks into the large city, a beautiful mermaid comes and takes the pearl from his hand, put it into a small area that ps is energy source of the city city glows. Now give me prompt for this each frame by frame, I will animate it it using AI. You do not need to generate image-to-video, I want prom from film. As far as I can understand, Gemini has a disadvantage that whenever you generate image or generate video and after that you prompt. Even if you type prom that you want to prompt, it will automatically start generating image-to-video for you. And the credit that you have, it will automatically cut the credit from your account. And that is really bad or we can say this is the drawback of Gemini. From the day Gemini introduced, this is the same complaint I always have with the Gemini. Bottle is good, but this is really bad. It has generated a good result for us. Now let's read it first and understand the flow of the video. Is beautiful detail and complete narrative flow for this science fiction animation. So now, as requested, I have generated detailed prompt from each frame. Now with this deep sea echo and 11 frame video prompt sequence, a macro start establishing the origin, prompt a cinematic eight macro closer of a massive swirling incent Columbia cloud system in a deep sea blue sky at the core of the cloud formation a transparent glowing conducen of water resembling then ancient organic tube extend downward from its precise tip, a single large resident water droplet details and begin to accelerate downward, capture with extreme high speed photography, dramatic studio light. We have to start generating a frame first. For this one, I'm using. Again, video generator, let's not go or let's not jump into a video generator now. By generating into the image generator, let's go to first, image generator and with this, let's use Google Nano Banana. Or search for the Nano Banana pro with this one, try Nano Banana two better and faster. Let's try it with Nano Banana two. Let's try it with the prompt I have generated, and you have to understand that the prompt that you have added or you want to dit, or you want to add. In the prompt section, there might be some limit with the generation. So you have to keep that in mind. Also, this is the resolution I wanted, and this is the one K resolution I wanted for this one. Now let's try to generate. It has generated something, but now let's try to add it with the different model. Let's try it with this generate. Otherwise, we have to change the prompt or retouch it with another prompt, and it is looking like something it's good. No doubt. But let's try it with the context. As far as I understand that context is really better, yes. Context has generated something amazing and I want this type of generation. As far as I understand that context actually giving me a really good generation. Now let's try. Now let's try to save this image first in our system. With PNG, now let's download it one, download. Now let's save it. With this one, now we have to simply add a cloud system. No Simple, clean. Now let's go to Freewi and clear sky. Let's type it and you have to download image from it. This is the image that I wanted. Now let's download this image, a large image and save it in our system because we have a Freepik subscription. Why not use PEPC for simple image generation. Sometimes we can save our generation credits also. Now let's go to another image generation that is the contact that I wanted. The droplet makes impact with your character note ensure the generation is em provide character appearance, right? So the photo realistic closer portrait of female character matching the user provided surface image sees looking upward with a sa comparative cinematic depth of field. Now let's copy this prompt and let's paste it here. But with the character, we want right now a character, right? 14. How to Create and Use AI Character Sheets for Multi Image Consistency: This is a character sheet that I have created, and this is the image, just a simple image that I have clicked. Let me tell you how we can use this image. With the video generation, and selecting the tool that is video generator C dense with the motion and image, let's try to add image here. That is the image. I have uploaded this image. And now, so in the image generator, I have added the character with the context max. I want to use the same prompt that I have copied and now paste this prompt here. This is the same prompt for the second frame, the contact of the image. Now let's generate. If it not work, then we have to tell that we want to use the character for the image. So I have to add a character here if it's not work. As far as I understand, it created an image. I loved it, but I don't want this to be very simple. I want it to be a very dramatic image. The image, it generated a good image, but I wanted something else. Now let's drive it Google Nano Banana two if it will not work, then I will use AI prompt and answer. Image generation is good, but I don't like this. I want something else. Now let's try to use AI prompt and answer with the generation. No, I don't like this one. Now let's try to use image description for it comes on phase. Now let's try to use this for only I want it. This is I want it, this type of imagery I wanted, and this type of imagery I wanted, check what else we wanted are happy. This is a complete workflow that I'm sharing with you so that you can understand how we can generate a perfect video for ourself with this one yes, I want this type of imagery, but cinematic raindrop face of girl, not girl of lady. As far as I understand it girls means the age should be under 20. But I want the age at least about 25 to 20. I want this type of imagery. Yes, this is what I wanted. This is for the image generation. With this image, add it in the photo editor or upload as a description. With this one, let's select this one, upload and add it here. Simply drag it here. Use this image. Now it's generating a prompt. So now let's use this image. Or we can try another sequence of it. Water droppers place face and hair, dismiss, which is wet and dark. No, not wet and dark brown color, curly curly hair, the background blur, green pot. I wanted in a cinematic look blur of green photos with blur, C Bach is soft, out of focus, light, warm, yellow and white tone. Not yellow and white tone, creating a dreamy and eth atmosphere the lightning is soft, diffusing, highlight the water and the women expression. I have tried so many prompt. I got these many results. The first generation that I like the most, and after that, I have not liked that much. So I'm going to use the first generation. 15. How to Create scene matching the sequence: Right now I'm tating a specific workflow here. What I'm trying to do is to upload this character sheet along with the main character to see how it processes the reference. At the same time, I'm using a specific prompt. This is my active prompt right here and you can easily read it out on the screen. After that, once I run the generation, this is the result I am getting, it shows a hand holding something. It's a single water droplet. It looks exactly like a raindrop falling into the hand. Now I want to see what happened if we switch things up. So I'm trying to use a completely different AI model. Let's see how the output changes after using a different model. After switching to the second model, look what we got. This time, it generated a moon type object sitting right in the hand. Honestly, as a standalone element, I'm not a big fan of this loop. But let's see I will save this image anyway for reference. Moving forward, what should we try next? Let's try utilizing another prompt and this time, I'm going to generate it using Gemini. Inside Gemini, I will set up my requirement to pull up refined prompt, copy it and paste it back into our tool. Now with this new generation, look at the result we achieved. We got this beautiful pearl like object glistening inside the water. This is a major improvement, and visually, it looks absolutely stunning. I really like how this turned out, so I'm going to save this image right away. With that saved, let's move on our next step. We need a proper background. So I'm pasting our environment prompt right here to build out the scene. 16. How to generate and edit the image for story board: The PL has been downloaded, and now let's try to add this in the environment with the context Flux. Somehow I love the output of Flux, one Context Max. Now let's try to generate this with this one or if not, I will ask for another. No, it's not that one. I want this in the character. Let's add this in arrector. This is the PL add one, and now let's add the same prompter paste. Let's add this prompter, paste. And now let's try to add a generate. I don't want it. Let's try to use another one with this. And with the character, the size of the pun, let's try to add this one. Let's see if the pearl is very big or not. This time, it has generated like this, and I like this result. No doubt about it. And right now, let's go to our new generation. That is, let's remove this one. And now let's add my own character sheet with my own phase. I have created this character sheet of my face, and as you can see, I am really handsome today. Don't worry. Let's try to move forward with our image generator. Now let's go to our character. Again, I'm using Nano Banana two for this one, and for the image generator, for the phase, let go to this close up, simply add this one and try to generate. And I hope we will get a result. That we have never expected. I can either good or it can iter bad. Expectation can go either good or bad. Let's wait for some second. Final task has begun and the image that I look is really good. Let's try again try to generate it and the face character inside helmet clearly and must fine person and frame two, as they are illumined by the intense gray lightning effect, the diver face characteristic helmet, close up focus directly diver helmet. They are light. Pearl visible. Clearly, now let's try to again generate this one. Basically, what I wanted is that I want a perfect frame for every video generation. That's all I wanted. Right now, I want the focus on my face as well as on the pearl. It looks totally odd. Let's try to enhance the prompt. And again, why this total Lod, you can see the hand is looking as far as I can understand, it's a reflection of the pearl on my hand, right? But I don't want it to be like this. I want it to be hand P with a pearl focus and with my face. This one looks okay, but no, this one is not pearl visible car not this one. Now let's try to close up shot. Not a close up shot. Let's try to add it, remove the close up. Let's try to add up portrait shot here, trait shot focusing. And now let's try to generate. And if not, I will try to again, using a free peak reference, I will try to use a futuristic water diver. Let's try to search for it. If we get something else. Yes, it's look cool. Now let's try to see. It's cool. It's really cool. But I want something special with the AI. Let's try with this image, I guess, no. The image with the image I want it to be like this. Let's see. Let's change orientation with the horizontal water diver, holding pearl. It looks great, like this. It's look good. Let's try to again generate a different prompt here. Let's try to edit this image. Okay. Let's try to edit this one reference and Pearl visible, clearly. Now let's try to generate. Right now, let's try to generate with the Nano Banana search for Nano and Nano Banana Pro. I want it. This is totally different. The style is, again, different. Let's try to use this one. Okay, don't worry. Let's try to use this one. And again, with the pearl, let's see if we can get our reference image glowing pearl on hand in water. No, the reference is also not good, but let's do it with this one. And if not, let's try to again, generate a prompt. If we don't get what we want it, let's try to I don't get better result from this prompt as Pearl is not visible. Clearly, give me prompt for Nano Banana image. No, I don't want to use because it will generate for image. Let's try to see. And I hope it will not create an image. If it creates the image, I really don't like the gemini because sometimes it generate the image and I really don't like. No sometime it actually created the image mode of the time. What we get this close up focus directly on divers helmet visa making the face clear visible through the clear glass dome, your character person look profound with sound dramatic light, and I have position the intense Soja frame lower right, it's vibrant, right, blue light and deep, so she enclose the skin and reflexion in her. Let's try to generate with this one. Let's try to generate with Nano Banana too, as the result of this one is really great. Yes. This is a prompt that I wanted, and now let's try to helmet wiser. But let's try to add a prompt with AI also. The results are really not good and not satisfying. I like this one, but you can see we have to edit this image a little bit. Let's use this use as reference. No, let's not use it at reference. And now let's try to did this image first. Add this image. With this one, let's try to add retouch by selecting this one, diver Marine suit. Let's try to change to diver marine suit. Let's try to add it with the model. No, I don't want four variation, I just want two variation of it and with the model, I want it Nano Banana, two. Again, let's try it. I want it to really futuristic one. Now let's try to use this image. Let's save it. First save it. You have to save it first. You have to remember it. Now let's back. Use this other apparent, change the suit of futuristic diver suit. Generate. So this suit is like the heritage style and I can see that the suit is about 90s or 80s type of suit. This is good. This is really good. And I can see it's amazing. But the pearl has been changed also. I don't like this one. Let's try to change this pearl to simple pearl. Let's remove and remove and did this image. Use it, edit, add a glowing pearl. You. Lun let's try to generate Image and I want only one K resolution with one variant. Now let's try to generate it. Yes, this is good. Now let's try to download this image. First save it and download it and export this image and download as PNG or JPEG, it's up to you. Save this one. We have another sequence that is first, second, third, fourth, fifth, six. It will fifth, six, seven, eight, and before that, I have also generated another image that is, this one, let's download this one and save. This is the image that I have downloaded and one, two, three, four, five, six, seven. So these are seven sequences that I have downloaded. Now let's move to another sequence that is onto the frame, the pace is done, the vison, the massive change in scale of tab the goal and establishing video wide so looking over the shoulder of a diver, still holding the pearl paste, and in the distance, a massive futuristic drawings is done. But I want it with this character. This is the character I wanted, and I don't want it. And let's generate it what we will get. We have to understand it first, what it will give us. 17. Total Character Control Advanced AI Workflows for Persistent Actors: Image has been generated, you can see, I have used it at character and created image that is good to see, right. Let's see if it will work or not like this. Let's try to save this one. So this is the futuristic underwater city. We can see, and this is the pearl that will going to illuminate this. So this is the powers of this city. So we have saved this image also. You can see the character. This is the suit, this the suit, and this suit. All these are resembling to each other. That is what we want it. If the suit is not resemble to each other, so the video sequence might break and we don't like that one. Now let's try to generate an image. Emitting a beautiful mermaid approaches a cinematic mit sort or beautiful mermaid with intensequ that gently places the growing pearl into her palm. Now let's copy and paste it. And now let's try to generate. If it did not work, then we have to ask for the mermaid. Or we have to find a mermaid fails and then no carriage for use. Imagination. Let's retry it. If it not work, then we have to ask for another character asm beautiful mermaid with inter scale glowing pattern on her tail and long flowing hair, sewing gracefully towards the diver see extend a delicate hand with web finger, and the diver gently places the glowing pearl into her palm. No, it's totally bad to see. Now let's try to. Move this one and not the reference image I want it. You can see that cinematic or a beautiful mama glowing pattern on her. Let's try to use it style or this one I want it. Let's try to use. Glowing pattern. And for the reference I want it, let's mermd with glowing pattern. Let's try to add our character from the character sheets and try to add a merit like this. Let's try to make a first let's try to make a mormd. Let's remove this one. I want a character sheets. Let's add a character. From the uploads, this is the character. And for the image, I want character history. Uploads at character is did. Now let's make a Murmad with this one, I like this one. Let's try to download this specific one. I like this one. Let's download this and try to create as per our reference image. Now with this one, upload the image here and use this image for the reference and for the character. Now let's try to generate. And I hope that we got the result that we wanted. Don't like the result as it's not use the image, so I have to use the character let's delete this one. I don't like this. For the character, let's try to add the character here because it's totally adding image here. So I want it to work as a character. Let's try to add a character here. No, it's totally different. I don't know why it's not working. I have to work it with the style also. Let's try to add this image for the video generation, let's try to add it with a style. I have to add a style here, character. And if it will not work, then I have to check for the different prompt. It's totally different. It's funny. Let's delete this one better. But it's not like that futuristic with a smile. Let's try to remove this one, remove this. And I want the prompt to work like create with this. Let's try to add a prompt here again. Let's side to prompt here, use it. And after that, let's enhance this prompt, enhance the prompt. This is the next feature. We are using a pros and you can use 10,000 character for this one. Wow. Let's apply the prompt and let's generate. It's not a close up. I don't want it to be a close up shot. Let's make it more dramatic. Yes, this is what I wanted a close up sort of the moment. Yes. Now this is what I wanted. Now let's save this one with the hand. Now let's try to add this and at this from the rep, friends gave the pearl two image. We are using now two images. You can see this is daily advanced prompting we are using. Now for this style, use it style here, the background. They both are amused to see each other. And let's see it will work. And I hope it will work. So what I have done here, this is good. Let's save this one. But also for the image, let's try to Image is a mermaid. And goes into the city. Let's try to generate it again. The mermaid swim into the city with wide angle shot and from behind no face visible of the mermaid red glowing tail of mermaid visible from behind. Let's see if it will create or not. Yes. This is good. Try to remove it by using. Let's save it, save. And with this one, let's try to open with Photoshop. Let's try to edit with Photoshop as I don't want it to be like this one, let's remove this one, remove by selecting this. Okay, no. By removing tool, let's try to remove this one. Yes. Why am using it Because I don't want to use my cart in the Phillipe. That is why I'm using Photoshop. So it has been removed. And I think there might be a crown. Let's add a crown like this one. With this one, let's try to add a crown here, genitive pill by selecting partner Nano Banana, two, Nano Banana, two mermaid wearing a CROWN Crown. Yes, it's now working fine. Now let's try to save this image. 18. Next Level AI Illustration Keeping Characters Consistent in Complex Scenes: With the generation, we have to add a character here. Again, the character, the pearl. What I'm trying to see, I'm trying to look into a area where we can add the pl and I want it to be very futuristic and very generative. I like this one. Now let's try to create a prom for it as add the pearl onto the area is spatially created for the pearl scene looks ancient and dramatic mermaid put that pearl from her hand that s after that and other prompt will be for the close up of the other prompt ware, pearl, power the whole city and everything's glowing. Now let's try to generate it. So I'm asking for two generations, and I hope it will not give me a image here. Please don't or do this, which is the power sort. Defining. Okay, so it will not generating an image. I just want a prompt or here. It's look good, but I don't want like this one. The mermaid added. Yes, I want this one, but the marmd is different, but give me the reference. Let's do it. This closer so capture the historic drama. Now let's try to add this one. Now let's paste it here and ask for Nano Banana two for the image generation, and I hope that we got something. And for the second image, let's try to finish our first image. Once we got the output, better output for our video generation, then we will go for the next generation. Right now we are just working on the image generation part. Then we will work on the video generation part. It's really cool to see. I loved it. Now let's try to save this one. Let's save it. And with this one, let's try to add up. Let's paste what we got here with the image generation for the image second image, right? The grand white angle sort of wonder. Now let's see with this one, paste here. And for the character, let's remove the character, the p. And I want the style to be like this. This is the city I have in mind, and now let's try to add this sty layer with this one, let's try to generate activated the p from image one, I don't want it. Basically, I'm changing a prompt little bit according to my need. A great result that we got here. Now let's save this one also. So the sequence has been created, a complete sequence of the images has been created. Let's look into the images that we have created just now. 19. Using Claude for better prompt direction in Video generation models: So far, we have generated all this image. Check for the sequence first. This is the first image. This is the second image, and this is the third image. This is the fourth, fifth six, seven, eight, nine, ten, 11, 12, 13. So this is the 13 images we are going to animate using these 13 images. Now let's see what we got in our results. So video generator is a little bit different from the image generator, as we already know that video generator is very advanced to you. So we have to first understand how the video generator actually work. We have first understood image generator how image generator work, but now at this time, we are going to learn how to generate video by using C dense 2.2. This is the latest model that has been introduced. Recently, we are going to use and create the video by all these generation that we have done. This is the first image. We are going to actually add a simple frame to frame did video. This is the image. And for this one, let's start by adding a shot here. We can add multiple shot in between if we want, and you can see the generation point, here it is. So first, let's understand how we can generate a video without using a frame. Let's use any model available here. So first, I'm going to use any model. This is a random generation of the videos, how we can do that and where we can use and different model have the difference use. We have already talked about how the difference models work. We are going to actually use Motion Control 2.6 and cling 2.6. We are going to use different type of video generation and work like that. So for a simple video generation, we are going to actually use a nominal video generation tool that is Google view 3.1 for this one, for the reference, and we are going to actually create. First, we have to create a prompt to describe the image and the video. Then we will going to compare both the image that we got and the video and after that, we are going to animate that specific video. First of all, I'm using a cloud here. Don't worry, you can use your clod Gemini ChatGPT. It's depend totally on you. Social media, informational video for a single frame, not a single frame, Fa give me a prompt for social media informational video in which a man drink coffee and how it will react in its stomach and brain. Keep the video prompt, very detailed. I want that the animation should be like a three character like Manekuen or wearing modern clothe, then there is a transparent visual of its body, just like human body. So this is I have given a prompt that give prom for social media informational video, informational video in which a man drink coffee and how it react in its stomach and brain. Keep the video prompt very detailed. I want that the animation should be like a three character like Many quin wearing a modern clothe where then there is a transparent visual of its body just like human body. Now let's enter it out and wait for the prompt. So it had just suggested I can use Sora Kling and pica. We will going to do that. So this is a global style direction, a visualchctor. After that, it will give me some scene also for one to four second, four to ten second, and the journey begin soloing, stomach reaction. So it has given me the visual direction for my video. Let's use this as a prompt. Let's copy this. But the expect I have to change the expectation. We will do that. By Let's go too. It's now using 160 P, and let's see if it's well. Okay, it's working for this one, also, and I want this one, and expect should be around this. And this is the TD animated video. For this one, now I'm going to copy the scene, the coffee movement. Now let's paste it. I'm going to generate two videos with different prompt. Let's see if it will work with the same character and animation or not. Got it. Now let's try to animate it. And if I want the audio for it or not, I want the audio. Let's try with 260 credits. This is the video that has been generated with that It's okay. It's totally okay. The visualization that I have in mind, it's completely working with this. But we have to understand if we do not want this taxpayer, then we have to check for the subtitle because sometimes not sometime, but every time in informational videos, AI thought of the subtitle or the title for the better understanding of their viewers or students. So it must have added a subtitle that I forgot to remove. Then simply we have to find that out. And yes, as you can see, bold animated tax paid in at the bottom, what happened when you drink coffee? This is what happened when you drink coffee there. We have to remove this one. This is okay, but actually not able to see the face of our character. It just look like it's not opening the face at all. The glass is not coming towards the mouse, like it's not touching the lips at all. So it's not that consistent one. So we want a different type of video or animation or we can ask for another model to create for us. As far as I can understand, Google VO 3.1 it will not work like we have thought of. We have to take another route, rectify this problem. But don't worry, it is just an example to show you how we can create a video. Again, we are back to our main generation where we are using the see tense two point oh, not fast two point oh. This is the T is the actual scene we want it. Let's start by using start and end frame and see how it will fill our video. And what I love about seed dense is that it actually fill the frame between our images and it work really well. Let's see how it will work. And right now and for that one, I'm using Cloud. So this is what we want to generate by starting from so this is the start frame, and this one is the and frame. And let's add a Cloud. But also, we have to add this both frame to our cloud so that we can understand. It's up to you if you want to use Gemini cloud, Ja JBR, it's totally fine. I want a cinematic movie scene type prompt for video generation in which Killer Iki as a reference image turns into this drop of water. Here is a prom, let's copy it. Now once you paste it here, you simply have to read it out first so that you will understand that the AI actually understand what we are trying to portray in front of it. Let's read the prompt first. A cinematic time la begin with an open sun drenc blue sky fill scatter with cloud drifting. You can edit out fast and understand pure vast and peaceful and the camera slowly is up at the smotfhift. The cloud begins swirling inward, converging into majestic rotating votetex lit from within the electric and and bioluminescent blue light, spiral titan into a singular growing funnel from the heaven. So we have to actually add first and last film. So I simply drag first and last film here. That's all. And it's uploading. Don't worry, it will upload in some time. And if not, we will have to upload it again. So for this one, let's start with the motion against a twilight. So we have to add the image, the start image and the end image. The timelt scatter in different pathway peacefully upward south different beyond into a majestic vortex. So before it, we have to as in reference in start image. Now for this one, for and image the camera slow tiles. So I have ad. You have to understand the prompt and you have to decide where you want to put this thing. And if you have not put it and simply start generating, it will automatically adjust your prompt and both images or frames and start creating a video out of it. But for now, right now, let's start with the four second with this one, let's start with the 480 and we can upscale it later on. But right now, let's start with the 720. Don't worry about the credits, and let's do it. Right now, I am using the credit so you don't have to. You simply have to learn it and you have to create a awesome video out of it. You can actually copy it. If not, you can create your own. And I would love to see that. I'm a little bit nervous because this is our first generation and I don't know what it will generate for us. Let's play this out. I honestly don't like this at all. Let's start by typing ourselves. We are not using prompt here we are not using this type of prompt, we are using simple prompt start from clear sky to dynamic sky in which shot two will be a close up of the drop making in the clouds and the shot three, the aim and image will appear like this. So we have two and fin will appear with the rain drop in it. After that, I want fourth frame in which the raindrop drops drops towards the Earth. Let's see what we get here. We have to wait for at least for three to 5 minutes and we got the generation. If we not get the result, then we have to change our strategy. I will show you the entire workflow, how we can achieve a cinematic best result. The video has been generated and let's see what we get here. Just a simple transition, right? I don't like this transition to happen. Let's see, let's remove this first frame, and I want this as a end frame. Got it. And now let's try to use a different model where we got a best result for our video. And with this one, let's add the image here as from clear sky to vortex of clouds rain drop towards Earth. And we are using five second of frame here, and let's generate it. And right now we are using Kling 3.2. 20. Understand the Node based workflow FreepikMagnific Space: Is the issue we are facing and we are switching over to free peak spaces. When we generate videos on a large scale using a reference image, we aren't just making a single isolated video clip anymore. Instead, we are building an entire sequence storyboard. To do that, effectively, we need a lot of workspace. We are moving into free peak spaces because standard image and video generation workflows just aren't give us the precise result we need. That is exactly we are leveraging free peak spaces for this project. So how you can enter this space, let's go to back. And once you go to home you will see a space here or you can see the space here. From here, you can see the space. And once you go to Spaces, it will automatically pop out like this. You can create new or you can just open your last created space. Once you create your space, you will see a well connected web structure you can see, and this is the video. Let's first see this with the sound. You can see this is a 15 second video created using seed dens and the aspect ratio is 69 with the sound and one variation. And it is 720 P. Here is the second generation that I have created. It is without sound. So you can see here the character consistency is really great. This is my face totally working fine. Now, let's start by creating your first I space. So this is advanced I space where we are going to learn the prompt, as well as we are going to learn how you can create your own space. So first of all, if you want to create this type of space, you simply have to add a node. All these are node based structure. Whatever you want, you can just add a text, just like this. If you want to add image generator, you can add image generator. If you want to add video generator, you can add video generator like this. If you want to add assistant, you can add assistant. If you want to add image upscale, you can add image upscale. All these are nodes. Every node he their title, and with the title, you will understand what nodes does. So title itself says, I generate images, and the image will appear here. There are two types to generate the image. First, if you want to generate the image, describe it here like this and simply hit run, and it will cart you 100. If you want to select the specific model like this one, you will see the carats here, the K cardit for the ten carts for Flux. You can select the model here if you want and there is another way to add a text also. Like this. This is the text, and here is a node. You simply had to connect it like this. Now the connection has been done, and if you want to disconnect it, you can see a casar here. Once you click anywhere on the node, it will just automatically disconnect it. And if you want to simply connect it, type here, blah, blah, blah, and you will see the light up here. Now, for the example, let's try to create a cat with Watermelon heat. Now let's try to run it. With the Flux model or image anything, you can use it and you can see the cart has been used for each generation. You can test it out with different generation. For this one, I'm using Zmge and let's try it. You can zoom in and zoom out by pressing Control plus scroll down and scroll up. If you press Alt, you can see scroll down and scroll up. And if you don't press anything, simply scroll up and scroll down, you can see, just like web browser, you can scroll the page. And once by selecting left click, you can select the node like this. By selecting right click, you can see all the options has been list here like this node system. Now, if you select, once you press the button in scroll, you can simply the hand will automatically grab the Canva and you can move around the Canva, just like this. The Watermaner head has been created on the cat, and it's really good to see what you have to do. So we have understood the tax prompt here and we have understood the image generator. Now let's try to add a video generator. Let's try to add a video generator. Now, simply connect this one. You can connect it with the single or you can connect it with the multiple also. For this one, let's try to use a symbol. So for this one, let's try to use a simple video generator like 12.2. I'm using it, and now let's try to run it. As you can see, it already informed us that it a start frame. For the start frame, there are two types if the so for this one, it requires the start frame. So adding any once you click here, adding prompt any prompt here, it will not actually work at all because it requires a image for the reference. So for this one, we do not want this one. We want a simple image reference. So getting image reference is here. So I have added a start frame here. As you can see now, first, move cat move around in the house. Once I add the prompt here, the run key has been light up. If you do not want it, you can also add a text prompt here add text and connect it here and add simply I paste it here and now you can see the text has been appear. Now let's try to run it. What we have done, I didn't write any prompt here, I simply add extra prompt here and add the image that I have generated and connected it to my video generator and it starts working. And within one or 2 minutes, it will generate a video for you. Once we understand the basics of the spaces, you will understand this web design. Meanwhile, it's generating the video. We can let's simply check for the images. So all these are images that I have simply dragon drop and uploaded it. So this is the image. I simply drag it and drag it here, and you can see it's uploading. Once it's upload, the image will appear. And this is the same image, and this is the same image that I have uploaded. Now, if you want to delete it, you can simply delete from here and if you want to download it, you can download it again. So now let's delete this one. Just like that, I deleted I have uploaded all the images that we have generated before. And we have connected it with the assistant node. Just now, I have not used any text prompt here. I have used assistant here. So how you can add assistant, add and you can see the assistant here. We will see what assistant done. We will see what assistant did in this case. Meanwhile, we were talking about this node system, video has been generated, now let's play it. You can see the video has been generated and it's working really fine. Now let's try what the assistant does for us. For this one, let's try to right click and add a assistant here with the assistant. Let's copy this one and paste it here. With this pasting here, you can see the assistant has a prompt, the result and the prompt, and it has the different model for the different output. For export as text. So first of all, let's connect this with the text. And now you can see the text has been appeared and the tax is connected with this. Now you can see, same text has been appeared here. The prompt has appeared in the assistant. And if we want to use the assistant for our prompt. Let's do not do anything and start by generating a prompt. And you will see a simple magic will appear. A redefined provocura domestic tabby cat explorer cozy lived in house from morning into afternoon. It's spit through the sunlit living room, and this is the total prompt. Once you drag it here, you will see this is the prompt that has been created, and you can see three and one to 4 seconds each frame poses and spatial coordinate for moment path. So it has actually understand my input of the text and why using ChatGPT model, it has created a really advanced and detailed prompt for me. Now, if you add or if you connect it with the video generator, or a image generator, you will see the magic happen. Let's try to connect it with the image generator at this time. Once you connect it with the image generator, you can see the prompt we can run it. Now let's try to use classic. I'm not using extra carried for use for this generation because I just want you to understand how all these network work. Now let's play it after running it and let's run it in this also and you can see image has been created and it's just a random event it just created. With this one, let's try to generate the image here. For the reference we are using again, the start frame for our Van, as it will not work with just a prompt. We have to add a frame here. Now let's try to add a prompt. And I'm using four ATP for just now. Meanwhile, it's in queue and it will generate really soon. So what this is start frame. Let's understand it by going to Freepik again. And in the video, Video generator, once you go to video generator, you will see the magic is happening here. The magic is happening here also. Now, let's see this is the start frame. This is the reference that it wanted and it has showed us the same result. Wow. You can see the video is working really fine. This is just a simple prompt that it's working. And this is the advance from that our assistant help us as an assistant, and the result has been created. But the problem is, we do not have the water meanel hat anymore. Obviously, the cat has jumped and the hat has been removed automatically. Or we can say the gravity will also work in it. No problem jokes apart. So no worries. We just want it to understand how the prompt system and assistant work. So you have understood now you can see it just creating a web. This is the node structure you can see, and it's really easy to use once you understand it better. For this one, let's remove this. And for the upscaler, if you want to upscale any of the image, you can upscale it by connecting the image directly to it. Let's connect it. Now the image has been connected, and if you want to upscale it, you can upscale it by running through it. And if you want to use any of the magic here like creative Magnific precision clarity. And if you want to use another model, you can use. So if you want to see any of the detail here, grainy analog, portrait balance, if you want to see any of the result, you can just experiment it out. Which model gives the best result to you, you can simply use. 21. Magnific Spaces From a Single Prompt to a Full Video : You have understood the basic of the node system. It's really easy to understand it. Now let's try to group it. If you select all of them, you can simply group them just like that. Group has been created. This group also can be used as a reference image or as a reference video for any of the further generation if you want it. For the larger model and for the advanced model, you need to use a group, just like for the UGC model, we will talk about it in upcoming lectures. But first, we just wanted to understand the node system, how it's work. Now, let come back to our. You can just simply drag it out anywhere you like. And if you want to change the color, you can do so. Now, what we have done, we have simply added the images here. All these are images that has been updated here. Now after uploading the images, you can simply add the assistant here. This is the assistant. Now for the assistant, I just wanted to understand how you can actually use this prompt system, how you can use it. So this is the prompt. Let's export this prompt. Now, let's try to export it as a text or export it as a list. Now let's try to export the result. Once you export the result, you will see that prompt has been exported as a text. Here, you can increase the size of the text, and you can see our generation or our video has been exported as a list of the prompt. And if you want to export it as a text, now let's try to delete this one, delete. And if you want to export is as text, you can export it. You can see here is the result. This is the result that we got as a text. So here's a text. If you want to export it, you can do so just like this, and it will simply work for you just like any text prompt. Now, let's delete this text. And we have the assistant working for us right now. If you want to copy it, you can copy and paste the text here directly, just like copy and paste it, just like this. And also, if you want what we have done here, we have connected all the images, just like this. Simply I removed all the connection here. Just like this, you can either connect by simply drag it around here, just like this, drag it here, and it will automatically connect with the assistant. As you can simply select all of the node, select them. And after selecting it, you can simply drag it and connect it connect everything at once. This is how you can connect everything at once. After that, once you go to assistant, how assistant work? Let's see. All copy. Simply what I did, I simply past the prompt here and ask Gemini to create a prompt for me. That's what I did. It's not that hard and fast rule for it. You can actually and one more thing, this assistant is really advanced that it understand all your inputs and create a sequence out of it. If you understand the prompt better. Let's see, vortex formation, image to image node condition on image to prompt glowing blue vortex funnel from storm into C electric circuitry like energy pattern. And you can see that in the three to four second, the image and the generation will work as if X building deep resonant Hum layer B is raising in pitch and an amplitude. You can see the grand stabling white sort image generation node condition on image six. All these are images that has been added here and it automatically understand the images. And if you do not want to add the image or if you want to change the image, you can do so just like this. Once you click here and Shift A, you will see and it will automatically help you in the assistant. But with the assistant, so with the assistant, you can actually connect whatever it connected, you can simply connect whatever it generated by using your simple prompt into the advanced prompt and simply you have to connect it like this into the video generation part, or I have generated it with the video generation here also. So before that, I simply run the prompt so this is the output. This is the first output that we have created. This is the first output that we have generated by using image two assistant to video generator and simply run it using C 2.2. This is what I have done. And you can see this is the output that it has gave to us. For again, after that, I simply not satisfied because my face and my model face, that is my wife is not appear here. Now, also, if you want to remove this one, you can remove that and I have connected. I have dragged this layer here, the node here, and I simply drag a new video generator here and connect it with it and connect it with the text prompt here, and I did a new video generator added here. For this one, I have added the text here. This is the text I have added. I have selected the seed dense two. For this one, I have asked specifically, I have connected the reference image here. And this is the reference image, I have connected here. So both the reference image has been connected or we can see the character sheet has been connected with them. After that, I have added use reference image for Explorer at my face. This is my character sheet for Mermaid, use this use character six as reference. After that, I have told the AI that use my face as explorer and Mermaid face as this character sheets, and it automatically generate the output that I wanted with my face. This is what I did extra. So this is really important if you want to generate it. This is the output with my face and with my wife face. And also one more thing I want to share with you that you don't have to actually settle it with the one generation. I recommend you to use at least the generation for two or three times. If we see this generation, I can use this is the part I can use from this video, and it looks really amazing. After that, the generation is not that bad. It's really good to see the generation here. Just want you to know that do not settle everything with just one generation because AI will always amuse you with new generations. Because every time you generate it will give you a better result, or you can guess if you get the worst or best result, you have to decide from every iteration. I recommend you to work at least with two or three options or you can work with two or three iterations of the same output or the same prompt. This is it for this space, I hope you have understood the basic of 22. Project : Now that you have mastered the tools, it's time to put your skills to the test. For your class project, I want you to create your very own visual story using the techniques we have discussed. Once your image and video clips are ready, head over to the Project Gallery tab. I will personally review your submission and giving detailed feedback on your work. If you hit a roadblock or have any question, just put them in the discussion tab blow. I'm here to help. I can't wait to see your stories. Happy creating.