Character Animation with Magnific AI : Bring Your Images to Life

1. Introduction: Creating motion graphics like this used to require an army of animator and week of rendering. But the wall between the great idea and a final render just came down. Hi, I'm Karen your instructor for this class. I have spent my career bridging the gap between complex software and creative vision. And in this class, I'm showing you the ultimate shortcut. We are going to master AI native motion graphics using the cutting edge power of magnific formally known as freepik. This class is for digital market content creator, video editor, and a TD artist and anyone who has ever felt locked out of motion design due to technical complexity. You don't need a high end studio or years of experience, your computer and your curiosity. Your project for this class, you are going to build this exact intro video. I'm pulling back the curtain and giving you the exact prom, the visual ideas that feel cinematically effortless and time transitions needed to recreate exactly what you are watching right now, you will learn how to master high and AI prompting, cinematic upscaling and slimless scene transition. We will go step by step. We will use magnific to generate our base scene and establish a rock solid story line or we can say storyboard. Then we will breathe life into those assets to create a professional motion design sequence. Don't worry, take a deep breath and just jump right in. I will see you on the top floor. Our first lecture is waiting. Let's get started. 2. Understand the interface of Magnific Spaces: In this particular lecture, we are going to learn how to use Magnific Spaces. After that, we will go to Advanced version of creating images and videos. I'm sharing the UI with you so that you can interact with it later on. So once you go to magnific.com or freepik.com, they will redirect you to same page because Freepik has been rebranded to Magnific. Now, when you go to Spaces, you simply have to go to new space and you have to click on New space, and it will generate a brand new space for you. What you have to do is simply have to change the name of the space. If you want to create a new space, you can do so and if you want to create or rename it, you simply have to demand it. That is sample. So the space has been created and it will show up in your front page of your magnific account when you go to Spaces. When over over this, you can see the nodes, cut, stick node, anything you want to go to a PV section, undo it, redo it. If you want to do anything, you can do so. From here, you do not have to do anything, you just have to drag a simple image or if you want to drag a node here, simply have to right click or select here, right click or select here, it will open the same thing. Image generator, if you want to generate a image, you can do so. That is cute cat jumping. On building anime style. Okay. Now, this is a image generator node, and this is a video generator node. If you want to give a reference to it, you can simply upload. You simply have to drag and drop the image just like this and connect it like this. That's all you have to do. And it will automatically connect it like this. You can see this is a connection. And if you remove the connection, you can do so like this. It's really easy to do that. If you want to add external text, if you want to add external text, like right click text, you can do so like this. Simply have to type anything here. If you want to add assistant, right click, add assistant here, you can do so like this and assistant will help you with the generation just like this, double copy and paste it like this, paste it, and now this is free, meaning, and once you run it, it will run on this node. With this, you can simply connect it, and after that, if you want to run the prompt here, you can do so simply have to run it. And if you do not want to run it, you simply have to if you want to generation two or three generations, it's up to you. I just want it one generation. Once you click will see so many, you will see so many models available here. I'm using Nano Banana, so we are going to use Nano Banana and some of the GPT models in the course. If you want to change it Image size you can do so simply have to run it. That's sold. Once you run it, just like image generator, it's generating a image, and just like video generation, it will generate the video. And if you want to connect a reference frame here, you can add a start image, you can go to image, you can add reference image. If you want to add input, audio input, you can do so. If you want to reference image, you can do so. Once you say will see the image has been generated, and it looks cute. So this is just a basic so that you will understand how to use the spaces in our upcoming lecture. Once you understand this, I recommend you to practice at least 15 to 20 minutes after this lecture, how to add all these things. It is really easy to do that. Once you get the hang out of it, you will going to love the upcoming lecture. See the upcoming lecture. Thank you for watching. 3. Getting Inspired from Pinterest and Create your unique character: This lesson, I want to show you how powerful Pinterest can be as a source for inspiration for your creative projects. All the inspiration I'm using in this workflow, it all started from Pintrust. That's where I found the reference that helped me build this character. Here what I did. I collected reference images from Pinterest and then I took those reference into Gemini Google AI tool. I simply uploaded my Pinter reference and asked Gemini to create a character for me that would fit my course style. After a few trials, Gemini gave me an output I really liked. So I took that image and drag it into my magneific workspace that we call spaces. Now, alongside the AI generated character, I also found a character reference on PintrS. Using that reference, I generated a prom inside the image generator. And for this, I use Google nanobanana too. I connected both reference together and the AI character and the PintrT D reference and added a simple prom to guide the generation. Now the original image that Google's image generator gave me was low quality, but I love the green version the most. So I worked on building a mood board around it, and then I explore different background in Photoshop. To see what look best with my character, I exported it as a PNG, which you can also do directly within Magnific. It's completely up to you. Once I had the background and the character ready, it was time to upscale the image because the original generation was low resolution. For this, I used magnetic AIs image upscaler. A is the key technique I want you to remember. When you use magnetic AI with multiple reference images, you need to clearly tell it three things in your prompt. Which image to use as the style reference, which image is the best subject you want to transform, what background you want to use. Once you give these three ingredients, magnific AI works like magic. It combines everything intelligently and generated a beautifully unified image. After upscaling, I replaced the low quality version with the new high resolution image. Repositioned it and fed all three references the main character, Toby, his background, and the painter's texture reference. My goal was to give Toby the same kind of texture and three D feel I found in that painters reference. I dit a prompt, recreate into style and three D, and now we wait for the magic to happen. In the next video, we will see exactly what output magnific AI give us, and I think you are going to love. 4. Convert 2D Character to 3D Character: Right, so let's look at the result we got. And honestly, this is not what I was expecting and the character looks like it's made out of clay. That's clearly not the TD render quality. So the problem here is not the tool. The problem is our prompt. We need to rewrite it and be more precise. Let me show you something very important about how prompting works in magneific AI, and this applies to any image tool you use. When you are using multiple reference images, you must clearly tag each image in your prompt. You need to tell the AI. This image that is Image one is my style reference. This image that is Image two is a character I want to transform, and Image three is the background. If you do not tag them clearly, the AI gets confused, and that's exactly what happened in our first result. So I have now rewrite the prompt, and this time, I'm also switching to chat GPT two image model to guide the image generator with better control. Let's run it and see what we got. This second result is better, but still we are not quite there. The character is now rendering in two D with some shadow, but I want it complete full three D render. The D and volume are not coming through the way I want it. So let's try one more time. So this time what I'm doing, I have add a new box that is image generator, and I'm using the generated image that we have already generated, and I connected this image with a new generator and I'm asking it to create it into a three d style render. Let's see what we got. This time, with a simpler, more focused prompt, sometimes with AI tools, simpler is actually better. When you over explain the AI tries to do too many things at once and loses focus. Wow. Look at that. This is the result we want it. The character looks amazing. Proper TD render, gate depth, clean lightning, and the style feels consistent. This is exactly what I was going for. No AI generated images is over 100% perfect stay out of the tool. There are always small things you want to fix, a little color correction, small adjustment, a small detail there. I'm bringing this into Photoshop for some minor touch ups. On the final image, and this is actually a great habit to build into your workflow. A generates a heavy lifting and Photoshop handles the finishing. So what's the big lesson from this section? A bad prompt gives you clay. A clear tagged focus prompt gives you a stunning three D character. Don't get discouraged. When the first result isn't right, that's completely normal. Rewrite, simplify and try again. That's a real creative process with AI tools. Same in the next lecture. Thank you for watching. 5. Using Photoshop the output for enhanced version: Now I'm working inside Photoshop to enhance our design. I'm going to take advantage of its built in AI capabilities. Photoshop utilizes Adobe Firefly to handle generative tasks directly within the workspace. I'm trying to do here is to add the letter T onto character T shirt, giving this character a distinct recognizable identity. The base character looks incredibly cute as it is, but adding custom writing will elevate the overall resign. Let's run the generation. As you can see, the letter T has appeared on the shirt. However, the initial style doesn't quite match what I have in mind. To fix this, I'm going to refine my prompt, state that I want the letter T to blend seamlessly with the fabric existing texture. Now I'm going to switch models to try a different approach similar to how we utilize upscale tools. Let's test this alternate method, look at the workspace. Now this is significantly better. To make it look flawless, I just need to brush up the edges slightly so the branding looks perfectly integrated into the clothing. When we zoom out, the latter t looks completely natural and perfectly blended into the graphic. With that fix, I will save this image and upload it back into magnific. At the same time, we can clean up our project space by removing the old placeholder elements. Just like this, I increase the size of it, and let's select all of them and move them aside so that it will not take up my space. Clean 6. Using Gemini for Story board and visual ideas: And right now, I'm asking Jem andi that, let's meet Toby, and we are going to learn how to meet Toby. Basically, what I'm trying to say, I'm trying to create a script, and I have a small script. In my mind, I'm sharing it with Gemini, and Gemini will give me a complete script out of it. And it will help me a lot, and we are going to create a video and out of this script. You can read it out easily. All these information that I'm typing right now, it's totally simple to follow. This is the script that we got from the Gemini, and you can read it out. It's really easy. And also Gemini has given me that visual information like what type of visual we can create with using the AI. And now we are going to use it in our video generation. Next, let's head over to Gemini to generate our next sequence prompt. To give you some context, I gather these initial visual ideas from Printerst. Now using those references, we are ready to generate our main image. I have already uploaded the modified character we customize in our previous photoshop lesson. Next, we will push this workflow forward by using our image generation tools. I'm adding the image generation node and inserting the core descriptive prompt text right here. This text is based on the narrative sequence we requested from Google Gemini earlier. Gemini provided a beautifully structured layout sequence for us. I'm setting up the text using exact word for word instruction. To guide the AI effectively, our main objective is to visualize the specific scene concept in a motion design style. While maintaining absolute character consistency across every single frame. As you look at the interface, you will see exactly where and how to tag your reference images within the prompt structure. Telling the AI exactly when to leverage specific reference points is crucial. Depending on your active workspace model and available system credits, you can select the best variant to render out the shot. The generation process is now complete. The tool has delivered an absolutely perfect storyboard layout for us. Every detail we requested is clearly defined here, including how the text integrated with the visual elements. This turned out exceptionally well, but I want to show you how we can push it even further to add extra distinct details. For establishing a basic layout, this is an excellent starting point for our video project. I will go ahead and download this image and then re upload it so we can utilize it as a continuous stylist reference. This workflow gives you a clear strategy for using Gemini to build baseline prompt concept. Now let's head back to our printers to pull out next set of composition, references for the upcoming sequence. I will see you right there in the next video. 7. Using Nano Banana 2 for Image Generator Node: Of this sequence, I spent some time brainstorming concept direction over on Printerust. I had this vision of creating a stylized motion graphic scene featuring a cute character Army busy at their workstations. To bring this to life with high quality vocabulary, I asked Gemina to build out a detailed prompt sequence. Now let's jump right into our canvas. Here, inside magnific spaces, we are going to create a fresh workflow from scratch. First, I will spawn an image generation node and immediately link it up to our baseline. Character reference sheet. Next, I will take that complete high descriptive prompt sequence we got from Gemini and paste it directly into our prompt input box. For the heavy lifting on this render, we are going to deploy the brand new Nanobrana two image model with our node connection logged in and model selected. Let's hit generate and see what we got. And look at that. The final render is absolutely incredible. Let's break down. How the model handle the composition details. I have created a new node that is image netter node and connect it with our image reference and add a new prompt that is by addressing the image by tagging it at the rate and tag the image that is toby I have asked that change this character to Rm outfit with lots of same character setting on computer. Computer with gaming computer high end computer in a studio, keep the background color consistent. By selecting a nano Banana to model, I have hit Gent again. And this time we got an army of our creator, and you can see this is absolutely stunning. Mean while generating the image, I have been storm some ideas from the Pinter itself. Again, I like this character sitting on the computer and very tired working on it. For this reference, I have simply copy and paste it into spaces. Here is the character I added. And again, I have added a new image generator with the reference image connected with it, that is our Toby and now, again, a second reference image that we have copied just now from Pintrus have connected with the image generator. So now we have connected two reference images here and ask our image model that follow the style of the image one, that is, this is the image that we have copied from Pintrus and change the character to Toby with same background color similar to Toby. This is a simple prompt, we do not have to be a master in the prompting. Simple enough is good. So this is the image. And again, I am creating an image for another sequence because we can generate up to eight image in a queue or video generation in a queue at the same time. So it is really handle to use this much generation. And in the second generation, I'm using Toby as a reference for the image generator character looking from left side of f, looking like watching at camera only had visible, but little test. Meanwhile, it's generating the third image, and this is the image that we have got here, and it looks really amazing. The texture is good and the character is consistent. Right now it looks happy while working, but I want it to be very tired working on the desktop. Let's hit again. And right now, meanwhile, this generating the second image. I have left check our later generation. It's work, and it looks really cool. This is what exactly I wanted and it has created. So right now, our image generation is working really fine, and it is working according to my imagination that I have thought of. Let's get to our next lecture and create some more images for our new sequence. Same in the next lecture. 8. How to copy one concept of image to another image: Following the exact same node based process. Creating a brand new image generator node. We will start by pulling a reference connection directly from our main character, Toby. Next, I'm bringing in that new composition reference asset. I discovered and copied over from Pinterest. Just like our previous setup, we will link its image output line estate into our generator node to combine our asset. Now let's look at the specific prompt I have added here. We are telling the AI to place Toby right into center of our new reference image image. We want to visualize capture a sense of total confusion with elements revolving around him and add a stylised black flower right on his head. I have also included a cleanup command to strip away any stray hands and ensure the entire composition is rendered with absolute sharpness and clarity. Let's hit the play button, run the model and wow take a look at that. The final output we got is absolutely amazing. 9. Using character sheet for combining your face with the scene: This time, I wanted to head in a completely different creative direction. While brainstorming on Pintrs, I decided to move away from modern workspaces and go with a retro old school max style instead. Following up user step, we will simply copy and paste that design asset directly into our canvas. Now, things are different and it's getting incredibly excited. Because we are connecting three distinct reference images into a single generator node simultaneously. We are linking up a main companion Toby, my custom face character sheet, and the vintage style reference we just brought in. Let's take a look at the custom prompt layout and structure inside the node. We are directing the AI to place my character set persona into the scene dressed in a sharp profession suit with a clean haircut. Sitting comfortably surrounding by a multiple computer setup inspired by our retro reference. We also want both characters, myself and Toby looking directly into the camera lens for the generation engine. We are sticking with our heavy hitter, Nanobranat. The first render came out incredibly solid, but to unlock its full potential, I went ahead and took a tiny bit of prompt detail for added flavor. Let's execute the final random node and see what we get. And now, look at this masterpiece. The blend of character consistency, wardrobe adjustment, and that retro coomputer environment turned out absolutely brilliant. 10. Using different share and scene combine them into 1 scene: For our next sequence, I found some incredibly abstract shapes on pinters and brought them straight into magnific spaces. Following the exact same steps as before, we are going to connect three completely different visual assets into our new generator model. Our retro computer, the new geometric shape, and our main character, Toby. The key to getting the pascalit from AI models is keeping your prompt simple, but making sure you tag your reference images precisely every single time. You tell the model exactly what you want, drop the tags in the right spots and let the engine weave them together. Now, honestly, the final result was something I wasn't expecting at all, but it turned out absolutely amazing. The way it planted the hard tech of the intagmc with the flute shape and toby design is incredible. That's the beauty of this workflow. Every single time you hit generate it, it sparks brand new ideas for your next video sequence. Let's keep this momentum going, move forward and complete our entire storytelling sequence. I will see you right there in the next lecture. 11. Understand Variation node for better image generation : In this lesson, we are stepping things up by exploring variation nodes. While it function similar to standard image generator, it is vastly more advanced and operates on a completely different level. Once you see the final output, you will immediately understand the true power of this feature. To get started, you first need to configure your node settings. Here you can select the specific node variant you want to deploy, customize your expect ratio, set up your grid layout and define the generation resolution for your output images. What makes this tool so incredibly powerful is its ability to take just a single reference image and automatically generate multiple distinct camera angles from it. For a filmmaker or storyteller, this is the ultimate tool for instantly creating extra role to plug into your video sequence. Now that the generation is complete, look at these frames. Each individual shots tell a unique story of its own with stunning consistency and detail. It is a really cool work for you to add to your tool kit. Let's keep moving forward, and I will see you in the next lecture. 12. Change the character and create similarly looking world : From Pinterest, I have copied this frame. As usual, for my next sequence, drag another node from our main image that is toby and ask it to generate a new image generator node. With this connected, I have connected the new reference image that we have just copied from. In Pinterest, now, let's add a simple prompt here again, that is our main character holding the folder, that is the image that we just copied, all the character and some character looks designer with cap holding camera artist looks similar to our main character, Toby, but different get up standing beside the main character that is holding the folder. On hands on hands and hands are above the shoulder. Hope you get the idea that how I used the prompt here, and this is the image that we got. This is exactly what I wanted, and it's exactly created the image that we wanted. A simple prompt yet powerful one. I'm adding a new extra detail that is magnific on the folder, so we can just spot it out that we are talking about the magneific. It's up to you if you want to add it or not. I totally depend on your sequence and your choice. I love to add small details on our images and videos. That is why I always try to make it more exciting. So after updated, after adding the small details, I have uploaded into our spaces. See you in the next lecture. Thank you for watching. 13. Why Reference Image is important for story telling: Trust help me so much for the ideas. So I use regularly for the frame ideas and using these as a reference images and combine it with our sequence and create our own sequence. Right now, I have used two images that is our main character, Toby and the reference image I just copy from the Pintrust again and add a small prompt that is character sitting in the middle, thinking, levitating eyes, close and tag our main image that we have just copied details surrounding the character. So what it does, it has created an amazing image that is our main character is levitating. After that, I have tried to add a background I have tried it with these different models, but it's not actually working. Until unless I give the reference to our image model, it will not work that way. So you have to keep in mind that you have to use a reference so that you can tell the AI that you want this type of animation, videos, or visual ideas. Once you give the idea, you will get the exact same result that you want it. But 14. Creating Series of images for sequence and transition : Sequence video, I have copied this motion graphic image. I have saved as frame for our next generation, and again, using the same step and a simple prompt. I have generated this. It looks very excited. Our Toby is very happy in this image. I've tried to generate it with a different sty, change a little bit of prompt in it, and it's generated like this. Now with the image that we have created already, the toby looks really stressed. And I have in mind that once the toby is stressed, it will go into its meditation states and everything feels calming and it feels really calm. You can see the expression has been changed. Now, this is the image, this is the image, and this one, I have also created another image with this jumping into the air and all the tabs remove them. But it's not looks that way. So I have to give the reference to image or that Toby looks exactly like this because once it opened the eyes, the eyes has been changed. So the character consistency has been loose also. Also, meanwhile, I have tried to add another sequence image or story line or we can say story board that Toby is with his friends standing in the lift that is standing in the middle, the image, this is the image and Toby, all the characters similar to Toby with different colors, size, outfit standing in the lift, happy, and we have to remove the human in the image that's sold. And you can see the similarity in this frame as per this generation is almost almost similar. So it has generated really good result. The image is also changed here. We have the toby after graphing the toby, the actual toby. I now generated a perfect toby for us that is jumping on the stage. Or the platform. Now with this image generation, I have created another image that is instead of some tabs at different flower and beautiful smells in the air. I have created another variant of this image. What I'm trying to do, I'm trying to create another variation of same images so that we can change the frame in our video generation from start to end frame. We will talk about it in upcoming lecture for the video generation, and for this lecture, for the image generation part, we have actually created so many images using the same method of ptrast getting ideas, and created a small prompt changed into the character, the style and the main character. Now, again, I have created an image with the two reference images and the image generator that is character is sitting like the man is sitting in the reference image. You can see all this web has been created one by one. At first, it looks really messy, but once you start doing it, it's actually become really easy. And now you can see the similarity in the frame, the Atbi and the reference frames has been almost similar, but I love everything that we have created so far. 15. Clean up the interface: For this lecture, everything has been created. The image has been created. Now, I'm trying to clean up everything. I'm trying to separate all the generation on the right hand side and all the reference images on the left hand side so that we can use them in video generation as the sequence. You can see from the start to end, I'm arranging all the images. It is really helpful so that you don't have to so that all the frames that you have generated will not miss out or will not get lost. See in the next lecture, we'll talk about video generation in our upcoming lectures. Thank you for watching. See you in the next lecture. 16. Understand Video generation using Seedance: Now let's move on the video section. Creating videos here is actually quite simple. Very similar to image editing. It can be a little tricky at first. Right now, I'm using the first and last frame style for video generation, and we will be using C dance too. Let's select five second here, and as a five second duration is perfect for this animation. Before we hit generate, let's use prom assistant to use let's use prom assistant to see if it gives us a better result. Inside the assistant, I will connect both nodes we used earlier. I'm using Opus 4.7 today. I'm just going to explain it to the AI, exactly what I wanted to happen between these two frames, and then hit run. Now we have our generated prompt. Let's see what it came up with. I have a feeling it's going to be good. Next, let's connect all the nodes back to their respective places, and there we go. Look at this output. This is exactly what I wanted to see. It's honestly not rocket science, just a simple explanation to the AI. A does the trick. We are going to keep using this exact workflow for our upcoming video generations. See you there. 17. Understand Text Assistant using Claude for video generation: We are going to generate our very first video here. How we can do that, just like image generator, we need a video generator here. Here is a video generator once you add the node, and we have to connect it with the reference images. There are different type of that is the first and last frame or the reference images you can add. And right now we are using for the seed dense, and we will work with the five or six around seconds every time because we need the duration of animation. So right now, let's so what we are doing, I'm using a first and last frame here with the video generation. And selected both the frame. Right now, I have selected this simple frame and the three D character change. And I have asked JaminiO we can say I have asked the assistant for my video generation that help me with the generation and give me the exact prompt that I am asking for. So what it will do it will simply regenerate my prompt and elaborate or add vocabulary or we can say it extend my words. That's all it will do. And add its own creative vision in the process and sometimes it will work and sometimes it will not work. We have to look about that. In our mind that if it will work or not, according to the prompt and if it will not work, then then we will lose some cart, obviously. Right now, I have used first and last frame, set dense 25 second generation, and this is the text I have used at the rate, start walking from left to right. That's soul but right now I need the help of the Gemini also. For this generation, I have used Opus 4.7, and in the open second, it just generate the image for me. First, I have used the assistant, and then it will give me the video. So this is the video that has been generated. 18. How to get better prompt using Google Gemini and teach the model: Let's add another video generator node here and select C dense as usual. After that, connect the node of the image that we have already in two D. Now I'm trying to add the node of our army with the first and last frame. And I have in mind that my character Toby will come to the frame and the frame will change into the army of Toby. They just come from upside, just like helicopter. I have this thing in my mind. So for this one, now I'm using assistant again with the help of assistant. This time, I'm using GPT 5.4 and I have connected the assistant and the image node with the video node. Both the reference I have adjusted with the assistant also. Now let's type a simple prompt that this is what I wanted generate with this hit generate and it just give me a specific prompt for my video generation. Now with this, I'm using Gemini so that I get confident with my prompt that I will get what I wanted. Why? Because we have to save our credits because they are really expensive in video generation, right? So we have to make sure that whatever we generate, we generate our video in just one go. So this is what I wanted. I have simply type what I wanted, and you can check that this is just a simple language. You can also write it down and ask Jamna to give you a prompt for it. Now a second by second, Jamini has given me a prompt. Let's copy this prompt and paste it into our text area. Let's copy and paste it into our assistant. Also, it is just an overview of A, we can choose the prompt, the type of from that we wanted, and we can also teach Gemini that this type of prompt I wanted, and it will give us according to our prompt example that we have shared with the Gemini or any chatbard if you have in mind. So I'm using OpenAt and you can say there once you login and try to open any of the Imago generation or video generation, they just give you they show you the prompt, what prompt you wanted. Also, so there is one more website that is Higgs field, and you can copy this prompt, the complete blog, the URL of any blog on the Internet, you can upload this and share it with the Gemini and Gemini will give you the prompt according to what you have given as input. Now what I have done, I have tried to add a text and working as a tag. I have added a text node here and try to tag my images, but the tag system is not working in text mode. I have used, again, our assistant node for tag that with this image, I wanted the change, and I simply copy what I get from the Gemini. That's all I have done it. And tag all the images wherever they mention my image. Now, I simply hit the Run button, and this is the example of good prompt. Let's check it out. Also, I will show you in the upcoming lecture, the good example of bad prompt. 19. Good example of Bad prompt: So for this sequence, I'm thinking of changing a new frame that is we have already created from our first filame to second frame, that is from two D to three D and get back to Army of Toby. Now, the Army of Toby will get back to our vintage computer scene that we have already created. For this one, I'm going to connect both the node just like this is the same exact same step that I'm following here. And what I wanted, I simply type in our assistant box that I want this transution like this and connected my both images. That is the toby and the Vintage computer again here connected with the assistant and assistant node will connect it to the video generator. An assistant has given me this again, same step, and I have to tag them once again. The reference images, I simply have to tag them. If if there is any change I wanted, I simply have to change that and connect it with the video node. Right now, I'm trying to use my prompt without any assistant help and without any gemini help. And with this one, so this is what I got. So this is the good example of a bad prompt. I simply tie what I have in mind, and the AI has not understand what I trying to say, and it has generated this. And this is really bad generation. 20. Good video generation output using Gemini Prompt help: The next generation, I have uploaded both images in the gemini. You can see the variation in one frame, it just looking into the camera and in the second frame, it's looking into the computer. And the shape that we have used and generated is in the frame. I ask Gemini. I want it very detailed prompt how each elements work and video work. With a good example, I have asked Gemini and Gemini given me a perfect prompt for it and it has generated the result. You can say it is really good. 21. When to use simple prompt in video generation: So for the next sequence, I simply again, ask Gemini to help me. And this is the video generator. I'm adding again and connect it with the new sequence that is Toby and the computer and the vintage computer. What I wanted, I simply type it into the text prompt, and this is what I wanted looking onto the skin like someone is talking in monitor can feel calm and happy flower on his a glowing and clapping natural movement. Simple text I have added again with a simple image and the reference image, you can actually create a video. I have not used any start and end frame here. I simply use the reference image here. Let me tell you reference image will work perfectly with sed tens, but with the cling, the reference image will not work that way. For this one, seed tense is the best. I love you so much and I'm so proud of you. 22. Use many variation and reference in 1 video generation: I hope you have remembered that we have used a variation node in one of our image and see the different angle of the image. The nine variation we have already created, we are going to expand them into a separate image so that we can simply use all the images for the reference. For this one, you simply have to click on the arrow. A simple arrow you can see on the right hand side, upper right hand side. This is the arrow. Once you click it and it will expand into the space automatically. If you want to download it, you can download it and upload it. So both are the option you can apply, and it's depending on your need. Move forward. Like this, it will just expand it at a separate image. Now, whatever image you like, you simply have to expand them one by one. Or either way, you can simply select the image that you have generated and you can select them, and it will automatically appear on the front end of your spaces. Select the image, and it will automatically appear and connect it to your video generator. So all these references that I have connected with the Sed tense video generator, and I like this frame. There are so many images that you can connect with the sed tens. And right now, what I'm looking at the animation, a better animation that will work for my generation. But right now, what I have added, I have added, create complete sequence using all the Japanese images, shapes coming out from the monitor. And I hope this will work. A sim transition images if I want to add, tag the image sequence 10 seconds. Right now, I'm selecting the 10 seconds sequence here, and once I select it, you can see the credits. It's very expensive. Meanwhile, it's generating the video. Let's download the video that we have already generated. This is how you can download the video. So I have downloaded all the videos here that we have generated so far. The step is same. So now the video has been generated 23. Perfect transition video using simple prompt: Video has been created, exact same step I have used before for reference images, put it into C dens, tag the image, ask in the prom that I want to transition from this to this to this. That's all. And it has created this, and this is the C dens result. I have used Google Veo three, and this is the result of Google View three. I like the result of CDns although it's really expensive to use. 24. Kling Video Generation for sequence transtion Start and End frame: This is the killing animation, and it is really good. I asked you from start frame to end frame, and this is the prompt. You can see on the left hand side, right hand side, the image I have used and below is the animation. The animation is good. I'm satisfied. 25. Wan Google Veo 3 and Kling Video generation difference: By reference to the main character, Toby, I have created image of it eating popcorn in the cinema, and it has created this, and it is really good. You can see I have used Nano Banana. After that, I have created a video out of it by using reference, and I have used first, I have tried an 2.7, and it created this. The generation is not good at all. Right. So after that, I have used. After that, I have used Google Veo 3.1, and Generation is also not good. Now I have used Kling with the Kling three point oh, mini I got the animation, and the results it's fabulous. This is expected with the Kling and the result, I'm satisfied with this also. 26. Most advance feature complete storytelling using 1 click Seedance: Last, I'm sharing very advanced feature that I have discovered within the CDNs and using GPT two. I have asked GPT to create a storyline or storyboard for the ten to 15 second animation. You can see this is the storyboard with the frame by frame creation, simple prompt I have used, and it has generated this story line. After that, I have connected this story line just like any reference image, I have collected this story line with the seed tense because it is specifically Wawa seed dense, so I recommend you to Wawa see dense also. Then you have to add it with seed tens, and this is the result I got. 27. Project: That brings us to the end of this fool. We have covered a lot of ground from setting up our initial nodes and workspace in Magnific all the way to engineering high level prompt with gemini to save our valuable rendering credits and nailing that perfect video generation. But your journey doesn't stop here. Now it's your turn to take action. Right below this video, you will see the project and resources tab. I highly encourage you to take the video sequence or the storyboard frame you created during this class and upload them to the project gallery. Don't worry if it's not absolutely perfect. The Skillshare community is incredibly supportive, and I will personally be jumping into the gallery to take a look at your work and give you feedback, seeing what you build is honestly the most rewarding part of teaching for me. If you found this class helpful and feel like you have level up your AI creative workflow today, please take just 10 seconds to leave a review. Your feedback helps other creative students discover this class and helps me create even better content for you in the future. Don't forget to hit the Follow button next to my profile name, so you get instantly notified the moment I drop my next creative course. Thank you so much for learning with me. Keep experimenting, keep creating, and I will see you in the next class.