These AI Videos Has 21.5M Views — Here's the Exact System (Fable 5 + Higgsfield)

What if I told you there are now social media accounts with purely AI-generated content that are getting millions of views and hundreds of thousands of subscribers? They are making their creators tens of thousands of dollars every single month through sponsorships, affiliates, and product sales. So what we're going to do in this video is reverse engineer their tactics, and I will go through 3 different video styles that we can do that are getting a lot of views today. And then I will give you all of my resources as well as skills in the comment section below as well. All right, so let's dive into it. So the first creator that I found was POV Lab, and you can see their entire account here is filled with AI-generated content, and they have 8 million likes and almost half a million followers. So definitely something that we want to pay attention to. All right, the second account here I found was History Films, and all of these videos are made by AI. As you can see, they got almost a million followers on Instagram. The third one that I found was Chloe vs History. And this one is kind of unique because this person has kept the character consistent. And basically Chloe goes back in time where maybe she interviews a famous figure, like basically the creator can get her to do whatever they want and be able to monetize this fictional Chloe character that they made as well. Alright, so if you're new to my channel though, hi, my name is Helena. And on this channel, I teach you how to use the latest AI tactics in order to increase your productivity, generate more leads, and create better content. I also have a free course that you can join as well, where you can dive deeper into all of these topics. The link is going to be in the descriptions below. I would also really appreciate it if you can like and subscribe to this channel, because it will help with the YouTube algorithm so much as well. Alright, so the first thing that we're going to do when we are trying to reverse engineer it is to see what they have done, right? And then we will break down step by step what they're doing, then we can use the AI tools available to reverse engineer this. So let's just watch one of the videos in POV Lab. That's the one that I think we should start with. Let me play one of them for you now. This one's got 21.5 million views here. All right. So I'm just going to turn down the sound here so you can see. Basically, it's a POV of someone in a particular time era, and they're doing different things. All right, so this is POV of a kid in Egypt in 1250. All right, and it's showing the things that they're doing, and it's going to one scene after another scene after another scene. Okay, so these are just, um, AI-generated videos of one scene after another scene after another scene, and there is some text on top. Alright, so now that we know the style, we can begin to create this. What I'm going to use to create all of these videos is Claude plus Higgsfield. So Higgsfield is an all-in-one MCP. So basically, once you install the Higgsfield MCP, you get every AI video and image generation platform, you don't have to go sign up for like 100 different accounts, right. And the great thing about an MCP is that based on your prompt, it actually knows the right API endpoint to call it to. So it knows which model that is supposed to use based on your prompt, okay, and what you're trying to do. So first, what we're going to do is we have to connect Higgsfield over to our Claude account. And what that allows us to do is when we type in what we want to do inside of the Claude chat, Claude can improve upon our prompts and call the appropriate image or video APIs in Higgsfield and generate those images and videos for us. And this saves tons of time because what I was doing before was I was getting the prompt from Claude or ChatGPT, copying and pasting it into like Veo or Nano Banana and getting the results and assembling everything together in like a video editor. And now Claude can do the entire process for us when we connect it to Higgsfield MCP. This was a game changer and it saves so much time. Let me show you how to set everything up. So first, we're going to open up our Claude account. And we're going to go over to Customize, then we're going to go down to connectors. And here we will connect to the Higgsfield MCP. So you're going to go up here and click Add. And you will go to Add Custom Connectors. Now here is where we're going to put the MCP service URL for Higgsfield. And it is this right here, mcp.higgsfield.ai/mcp. I will include this link in the descriptions below in the resource section. And here, this is where you just put the name of the connector, which in this case is Hixfield. And that is it. So you're going to press Add. And what you'll see here is that Hixfield now shows up in your list of connectors. Now you're going to press Connect. And what you need to do is log in to your Hixfield account via their OAuth system. Okay, so I'm going to log in right here. And now it is connected. So after you have connected it though, click on Hixfield and make sure that you select Always allow on everything, or else Hixie was going to ask you, can I do this? Can I do that? I just say you can do everything, right? I don't need to limit what you can do. So I always go and I select always allow to get the creation process run even faster. Now we will be doing all of our prompting here inside of Claude. And I know there are times when I created a video like 3 days ago, and I don't want to scroll up infinitely to find that. So you can actually go to Hixfield and click on assets. And you will be able to see all of the videos, all of the images that you have ever generated using the Hixfield MCP. Okay, so for me, this is a super effective way to find things easily. Alright, so let's go back to creating our POV video here. So again, first, what we want to do is get Claude to write us a script like this, right? So going back into a time period, having the POV of that person doing different things. And what we can do here is actually like just train a skill. So I usually train a skill inside of Claude when it's a repeatable prompt that I want to do over and over again. Alright, so I can say something like this, I want to create a skill that will allow me to create POV style videos for a specific character. Like I can go back in time and pretend I'm a poor person in London, or I can be a person living in 20 30, etc. And basically, I want you to write the script that I will need for these reels that I will post on social media. I've seen that they are getting a lot of attention lately, and basically every scene will be a POV of me doing a specific action, and I need you to write the script for me. So POV, I'm doing this first, POV, this person's doing this thing next, etc. Okay, so I can just type in something like this. And I can say at the end, give me this skill to save. Alright, so you can just talk to Claude and tell it exactly what you want it to do. And Fable-5 is really smart enough to already know your end objective. Now I do want to point out a major difference between prompting in Fable versus the other models. Now you may have heard of all of the prompting techniques like you need the role, you need the exact actions, you need examples, you need constraints, like in all your prompts. So you get like these very detailed prompts. But Fable actually works differently. When you use these standard types of prompting advice on Fable, you actually handcuff it and you limit what it can and can't do. So when you're prompting for Fable, it is better to just tell Fable what your goals are and let Fable figure out how it will accomplish the task that you want it to do, because model is actually smart enough to figure that out. Alright, so let's wait for our skills to finish writing here. Okay, so it's been about a minute later, and you can see that Claude has created our skill right here. And once we click on Save Skill, what happens is that this is now saved in our list of skills. And how we can access that is to go to Customize, then we go to our skills, And we can see the new POV scriptwriter skill that we have created. Okay. And basically what a skill in Claude is, is a repeatable set of prompts that every time you call this particular skill, it will run through this entire prompt right here. This saves you time because you don't have to prompt it over and over again. Claude will automatically remember your specific set of instructions. Now, How I would build this skill from here is test it. Like I'll say, you know, POV scriptwriter, poor kid in London, and see what it comes back with. What we want to do is test it out and see if it's giving us the results that we want. So I'm going to call the skill, and the way that I call the skill is I use backslash, and then I put in the name of the skill that we have just written, and it should turn blue like this. Okay, and I'm going to say like give me the POV of a poor kid in London. All right. And then we will scroll down and see what the results are here. Okay, and see if it's exactly what we want. And if it's not, then tell the AI exactly what it didn't do right. And if this is what you want, then what you want to do is add to the skill. So test it one step at a time, right? Once we nail down the scripts that we want the AI to write, next we want to do is connect it to Hixie Higgsfield MCP so that it can start generating these videos for us as well. So let's just do that part right here. Okay, now that we have nailed down the video script, what I want to do is connect this to Higgsfield MCP so that we can start sending the video prompts to Seed Dance for it to create these videos. I also want you to compile all of the videos back and stitch them together and add the caption text above the top third of the video. I'm going to attach an image for you to see what the style should look like. All right, so now I'm gonna go back to TikTok here and take a screencap of what the style looks like. And then I say, please update this skill now. Okay, so basically, this is the process that I have gone through to go back and forth with AI to nail down the skill. It took me about 2-ish hours to really get this down. And eventually, I trained it to do exactly what I wanted to do. Now after I spend all that time training the skill, I just want to show you what the end result now looks like. Okay, give me the final skill that I trained is called POV Reel Builder. And give me some ideas for the reel. Okay. And AI is going to give me a list of ideas. Here, I'm using Fable 5 for everything so that I get the best generations here. Okay. Um, okay, so it's given me 3 of its top ideas. Ooh, I love this one. I love the first one, POV, you are a gladiator in Rome, 80 AD. So I'm going to call this skill and the way I call it is I just press the backslash. And the skill that I trained for this particular reel style is called POV reel builder. And I'm just going to give it the idea now. And now because I have the detailed instructions here in the, in the prompt, like telling it it needs to call the Higgs field, it needs to know the topic, break down that topic into the different scenes, then create the videos on it, then stitch all of the videos up together and add the caption on top. And yes, Claude, like right in the Claude chat, you can stitch videos together, add text on it. Like Claude can now do basic video editing without you having to install anything new, without you having to go into Claude Code, do a million things. Like Claude is getting pretty damn powerful right now, if I should say so myself, right? Okay, so you can see this is the exact instructions that we are giving to the AI. And by the way, I will include this prompt in the description below. And as you can see here, it's giving me all the scenes and all the captions that it's going to include on top. It's asking me to approve this generation. So I'm going to do that. I'm going to tell it, yeah, please create the first scene using Seedance. So Seedance is currently my favorite AI video generation platform. If you want a cheaper one, Klang would also work. Now you can just say create the first scene to Higgsfield. If you want to specify the exact model that you want the AI to use, you can do what I did here as well, just tell it in natural English what model you wanted to use. And it's actually smart enough to figure that out. So I'm going to generate all the scenes, and I will be right back once that finishes. Okay, so it's been about 3 to 5 minutes-ish, our first clip has finished generating. So let's take a look. All right, so that's pretty good. I'm happy with this video. Now I'm going to tell Claude to generate the other 4 videos. I am happy with this video. Please go ahead and generate the other 4 videos using Seed Dance as well. Then I want you to stitch the 4 videos together and add the captions on top of them. So I'm just going to send that prompt. After I reviewed the first video, I'm happy with the generation. And you can see here, like in the prompt, Claude actually tells you how many credits each of the videos will take. If you use a less powerful image or video generator, it will cost you less credits. That's why by default, if you don't specify that you want to use a more powerful model for video generation, uh, Higgsfield is going to default to CLING, which is probably the best balanced model at the moment. Like, you are going to lose some on quality, but the cost is significantly less than Veo or SeedDance at the moment. I just love SeedDance at the moment, so I'm okay with spending a couple cents more to create that. So basically, we're just going to have to wait about 5-ish minutes for the remaining 4 videos to generate. And Claude is going to stitch those 4 videos here for us together and add the captions on top. This was all trained in the skill, and we just trained it with like everyday natural English, and Claude did all of that behind the scenes. Okay. Alright, so we're gonna wait a minute for all this to finish generating. And I will be right back once that's okay. So I think it's been about 7 to 10 minutes-ish. And our videos have generated, and I have downloaded all of them here. So one thing to note here is that Claude can't actually see the videos that Hixfield has generated, it can only get the links for you and display it here. So in order for me to get Claude to stitch everything up together and add the POVs, I actually had to add in all of the videos in the order that I want them to appear and give this over to Claude. So let me just give it the instructions here. Now, here are the 5 videos that Hixfield just generated for us. I want you to stitch the 5 videos together and burn in the captions and give me the final video. Okay, so again, I'm just describing in plain English what I wanted to do. I've uploaded all of these videos here. And because I have already trained the skill to be able to stitch the videos together and burn in the captions, now Claude actually knows exactly what it needs to do. Alright, so we probably had to wait another 5 to 10 minutes here. And you will see the final video in just a moment. Alright, so it's been about 10 minutes or so. And we have our final product done now. So actually, in the first version that Claude generated for me, when I watched it, what happened was the words that it put onto the screen, it was actually bleeding to the edges, and it was getting a little hard to read. So that kind of just made the video look cheap. And I didn't really like that. So what I did was, again, I just told it in natural English what I needed to do. So I told it to update the skill so that the captions that is burning into the videos won't bleed to the 2 sides of it. Okay. And it updated the skill. So in order for me to now save the skill, I will press on Update Skill. And this is how you can improve your AI generations over time by retraining your AI to do exactly what you want it to do. So whenever you look at the generation, tell it what you don't like, and it'll continuously improve. Okay, so now I'm just going to press on Upload and Replace. And now my skill is saved. And now in the future, when it's Burning in the captions, it won't bleed to the 2 sides anymore. And I have the final version here. Let me just play it for you so you can see what it looks like. Pretty cool, right guys? Like, we pretty much copied the Reel style on this AI content account, like to a T. And you know, they're getting like 400,000 followers. So there's no reason why we shouldn't do the same. And the cool thing is, once you finish this, you can actually schedule it on Cowork so that this skill is automatically generated. So how you do that is you start a new chat, And you will see the 2 buttons here at the bottom, chat or cowork. We want to select cowork. And we want to tell it to run the POV skill that we have just built, let's say every Monday at 9 AM. And that's how we schedule this in. And we can have a new reel all assembled together for us to review at the certain times that we want it to be done. Okay. So this is it. Now that you know how to create POV style videos using AI, it's now time to move on to the second style of AI videos that are absolutely blowing up. It's going viral like on every platform. And these are these morphing videos like this channel does a really good job of it. And they have almost 1 million followers on Instagram now. So it's basically like one person morphing into another, morphing into another, morphing into another. I've also seen accounts where they create videos like this, but it explains the life of someone from birth to death, like usually famous people that they do this for. So when we're looking at these videos, and we're trying to reverse engineer them, right, we want to see like, how do they work? So it looks like they have a still shot of this person. And then they, and then the AI transforms them into another person. And then this person is then transformed to another person. And these are just like individual takes takes that are very smooth, like the transition is not even noticeable. So it feels like it's an entire video, right? Entire generation that's all done in one. But right now, we don't have the ability to generate videos over 15 seconds. Seedance actually just announced that their new version of Seedance 2.5 is going to allow you to generate AI videos up to 30 seconds. I'm really excited about that. But that's currently not available via API right now. And then so we know that they're just stitching these videos together, they're generating these images and then morphing the images one after another with some sort of AI generation tool like Seedance or Veo or Kling. And then they have these captions here. And this is really just a Python script that you can actually ask Claude to write for you. Okay, so now that we have reverse engineered what they did, we can start now building our own. And so we go back to Claude here. And then we want to train the skill again, just by describing in natural English exactly what we wanted to do. And what I would do here to train this particular skill is I would start like taking some screenshots of the still images one after another like this, and then putting into Claude so that it knows like how the morphing is supposed to look like so that it can write the most appropriate prompt here for us. So as this person moves, then I'm going to take another screenshot. And this is because Claude at the moment, it can't watch any videos, it can only see images. So I really wanted to see like what this kind of transformation or morphing looks like. And it helps when you give AI as much information as it needs. See, it's now transforming into a different person. Okay, so basically, I'm just going to take all these screenshots here. And really allow Claude to see how this morphing or this transformation of those people really look like. And then I'll put in my prompt here, I want to create a Reel or TikTok that showcases a particular historical time frame where the person transforms from one to another. I want to make these, uh, content really engaging so that the audience learns about history. Basically, we have one frame of the person in a particular time period, and through a series of natural movements, as you can see from the screenshots here, they morph into another person. And also, there is captions that is going to be at the bottom third of the reel. I want you to make sure that you leave margin on both sides of the reel. And the date, as you can see, is going up by increments of 1. And so you need to speed up the counter so that when the person morphs from person 1 to person 2, the dates are going to line up. All right, so we can start with this and get Claude working on it. I would just, uh, probably add here— actually, I would add something else here. We are first going to get Hixfield to make all of the image inspirations for us, and then we are going to agree upon a script, and then we will use either Veo or Seed Dance to generate all of the videos. Then I will feed all of the videos back to you. You will stitch them together and put the captions on the video. All right, that feels pretty complete. And I'll just fix some typos. And then we will send this and I'll be right back once we finish training this skill. Okay, not going to lie, it took me about half a day to train the skill properly because there was a lot of back and forth. I looked at the results that Claude was giving back to me. I didn't like it, had to go back and forth a lot of times, but I think I finally nailed it. Okay, so I call this skill Era Evolution Wheel. And let's get it to come up with some video concepts here for us. Now, this is a thing with training skills on Claude, like it's an iterative process. So you look at the results, you tell Claude what you don't like, and then you tell it to update the skills so that it knows for next time not to do the things that you don't like, right? And you can see here, what we're doing here in this skill is we're first researching the timeline, then we're going to generate all of the portraits first. And when we like all of the portraits and we like the timeline that we're going with, then we're going to create all of the morphing clips. And then I will feed Claude back all of the morphing clips because Claude can't actually see the videos that Higgsfield has actually generated, not in Claude Chat. Claude Code can do that. So I have to feed all of the MP4 files back into Claude. Like what Claude is doing is that it's just grabbing the link to the MP4 files from Hixfield, but it doesn't actually see that link. So I'm going to post it back into Claude, it will stitch everything together. And it will add in the caption with the time counter. And it's pretty insane that Claude can write like Python code that just does that. Okay, so after telling Claude to come up with some video concepts here for me, it's Its top pick is the evolution of Istanbul, by the way, beautiful city, amazing tea, and I love Turkish breakfast. Highly recommended if it's not already on your travel list. And it's drafted these timelines for me, which looks amazing to me. So the next step is to generate all of the images. Alright, after telling Claude to generate images of all these characters, here is what we got here. This is the first man who is the Greek colonist fisherman, and he is going to morph into a Roman citizen, and he is going to morph into this merchant woman, and she is going to morph into this soldier, and then he is going to morph into the Ottoman, and then he is going to morph into someone from the Turkish Republic, and then finally he is going to morph into a young woman from Istanbul in 2026. Okay, so these are our characters. And I literally just told Claude, like, go ahead and create those characters. And I do find that when you tell AI to like, make the background a relatively like same hue, just from my testing, the results from the morph will look more natural because like the colors of the background isn't always changing. It's just the characters that's changing, it makes it look more real and more in sync. Okay, so our next step at this point is to create our first morph. And this is just like totally a Helena thing. I like to create the first video first because videos take up a lot more credits. So I want to make sure that the AI is doing it right every single time, rather than having it generate all of the videos. For me, I like to be involved every step of the way. I like more human in the loop. So I am reviewing after it gives me the script, I'm reviewing after it generates all the image, I'm reviewing after it generates the first video, and then I tell it to generate the rest of the videos. Alright, so this is the prompt I'm going to put in, literally just describing in natural English, like there's no trickery here. I did— I'm just telling Claude, can you create a transformation video from the first first character to the second character to see if it looks right. So I'm going to enter this and I'll be right back to show you the results. Okay, so it's been about 5 minutes or so and our video is generated. And let's play it so we can see what it looks like. So we see the fisherman, and he is morphing very naturally, you see, into the second person. So I'm very happy actually with this video generation. And you can see it is pretty close in terms of the quality and the style of the History AI Films Instagram channel that we were just taking a look at. Now, if you are not happy with your video for whatever reason, just go over here and click Create, and then add in, you know, your prompt for what it is that you are not happy with. So you can recreate things right inside of Claude right here, which is super duper useful. You can also just inside of Plot, once you have your Higgsfield MCP connected, tell it to edit photos, and it will know to use the exact API endpoint for the AI tools that can edit photos to be able to like Photoshop things, remove things. Like, it's super powerful. It's like an all-in-one tool for anything you need for AI content creation. So I was really happy with the first video. Now I'm going to generate all of the videos, and I'll be back to show you what those look like. Okay, so it's been about 10 minutes or so, and all of our videos here are done, as you can see. Now the next thing we need to do is to upload all of these videos in the order that they're supposed to be in our Instagram Reel and get Claude to stitch them up together and burn in the captions. Okay, so you can see here I uploaded all of my individual clips, and each of these are 8 seconds long of one person morphing into another. And now I'm just telling Claude here to please stitch the videos up together and burn in the captions. And I told it to refer to the skill that we have worked on together earlier, right? I think a lot of times when you're watching YouTube and other tutorials, like, you always see the polished demo, but you never see the mistakes. And I just want to be here to tell you Like when you're training these skills, a lot of things go wrong and you just have to kind of like roll with the punches and retrain Claude and retrain Claude. So in this case, what happened was Claude didn't actually make the counter. So it literally like from one frame to another, it was saying like, you know, 600 BC and that went on for 8 seconds and then it started to say like year, I don't know, 500 CE and then so forth, right? So the counter wasn't working. So when I looked at the results, I told it what it didn't do right. And also the captions were bleeding on both sides. So like some words were getting cut out on the left side and the right side. So like it really just wasn't adding up anyway. So I gave it the feedback and they fixed it and they rewrote the stitch year ticker in Python, as you can see here. And this version did work very well. And then it gave me the file to update. my skill so that it knows for next time, right? I already saved it, so I'm not gonna save it again. And finally, it gave me the final MP4 that I can download, and this was amazing. All right, let me play the video for you now. I mean, heck, if I should say so myself, this looks like the style of the History AI films. Like, this is exactly how they are doing it, you know. And in fact, I'm so inspired by this, I'm going to start my own AI history, uh, channel, and I'm going to report back here and let you guys know in a couple months how many subscribers I end up getting. Um, because I just, I just love creating AI content. I mean, like this morphism and transformation, like, you see how smooth it is and all the, all the people going from one to another. I love it. Chef's kiss. I think I nailed this one. Okay, I hope you guys are getting massive value out of this video so far. If you learned something new, and if you like this content so far, please like and subscribe. That's going to help out this channel so much. I've also got a free course that will teach you how to build AI agents and take your AI content game to the next level. It's absolutely free to sign up. Make sure you check the descriptions below as well. Alright, so now let's move on to our 3rd Instagram Reel style that is absolutely going viral. This content is also 100% AI generated. So basically what's happening is that the creator is creating an AI persona, an AI person, and that person stays consistent throughout each of the posts. And that person is doing various different things just the way that, you know, a human does. This person has a character, has personality. And I don't think anyone does it as well as Chloe vs History. I mean, she's got almost like 650,000 followers at this point. And basically, the whole premise is that like she is this Gen Z video blogger, and she's going back in time to various different events and she's interviewing historical figures, she's blogging herself, and it's quite entertaining to watch. Like, I'll put on one of them here. Like, she's spending a day with Maria Antoinette here and it's flashing to Maria Antoinette and she's basically explaining to people like what life was like in that particular time period in a very engaging way. Way so that people are learning what's going on in that time period without reading a boring textbook. This is quite engaging and fun to watch, right? And that's why the style of video is blowing up. So first, what we need to do is to create a consistent character. So you can see from her videos, like she looks like the same person in every single video, right? And now this was a task that used to be pretty hard to do. Because AI, like every time you tell her, tell them to like generate a vlogger who's brunette, blue eyes, like they end up looking like someone new every single time. Hixfield does fix this problem. And you can create a consistent character. And then as soon as you reference that person's name in your Claude account, Claude actually knows to create images of that particular character for you. So we're going to go back into Claude here, and again, say something like, I want to create a new character. She is going to be a Gen Z video blogger, and I want her to have brunette hair, and she is going to be in her 20s. Please design a couple versions of her that I can choose from. All right, so that is literally all we need to do. Now, while this is loading, what's happening is that Fable 5 is actually prompting for us and learning what it is that we want, and then finding the best way to connect with the right API in Higgsfield in order to produce the content that we are actually looking for, right? So you can see here that Hixfield actually has their own image generator. And it's generated a couple of different versions of our vlogger here. I've actually created one earlier, and her name is Zoe. And in order to call that person up, you just reference that person up here, please pull up her from my characters in Higgsfield. Okay, so you can create a new character, give her a name, choose the one that you want to pick. And here is what's cool. So once we have designed our characters, they are actually saved in Higgsfield. So if I want to create another video with Zoe in it, then you can see her reference image is already here. And I can just press Use. And you can see that her ID now comes up. So if I want to generate image of Zoe skiing down a mountaintop in snow pants and a jacket, okay, I spelled Zoe wrong here. And now because we referenced this character here, we can now get it to create any kind of image or video that we want. And so after we trained our character, and once we are happy with the results, now we can start writing the script and the scenes for what this character is going to do in this particular short clip, right? So you can see here, yep, there's Zoe, skiing down the slopes, right? And whenever you see this, like, you know that it's referencing this particular character, you will see the name become like almost like a clickable button here, right? So and you can see here, like, I just gave it a basic prompt. And now Fable 5 took my idea and turned it into a much more detailed prompt. And this saves me time because now I don't have to copy and paste this, you know, into Klang or into Veo and get it to generate the image or video that I want, and then use a video editing tool to stitch everything up together, right? Total game changer here. Alright, so once we nail down our character, the next thing we have to do is to write our script. So what I decided to do after brainstorming with Claude Fable 5 for a bit is to put Zoe back in time and get her to spend a morning with Cleopatra. And here is the 9 scenes of what's happening. So I always like to review my script first because video generation APIs are expensive. So what I'm noticing here is what is she wearing, because I want her costume to evolve as the video evolves so that like she goes in with her Gen Z outfits and then she's changing into something more Egyptian, more in line with what people would have worn in that time period. I'm also noticing like what she's saying, and I want to make sure that the scenes from one to another is not disjointed. I'm also noticing like what kind of actions are they taking and does it make sense for that particular scene, right? So I'm not leaving everything up to the AI. And this is the final version. I was happy with this. So let's go ahead again to generate the first scene. And once we like that, we're going to generate all of the rest. Alright, so after 5 minutes or so, here is the first clip that was generated. I'll play it for you now.

So I'm in Alexandria, 41 BC, and Cleopatra agreed to let me follow her actual morning routine.

That's pretty good, right? So now I'm going to go ahead and generate the rest of these clips, and I'll be back in 10-15 minutes when all of that is done. So as Hixfield was generating the remaining videos, and by the way, I'm using Seed Dance 2.0 to generate all of these videos. My 2 favorite AI video generation platforms at the moment is Veo and Seed Dance, but I'm using Seed Dance for all of this tutorial here today. So as it was generating and going through the script that I gave it, it noticed that I needed 2 more characters in here. So So Seedance here generated Cleopatra for us, and it also generated Iris. So Iris is one of the servants who serve Cleopatra. It asked me to confirm that this is what I want all 3 characters to look like. And after I confirmed that, it went ahead and generated the rest of the videos. And after about 10-15 minutes, and this is what we got. I'll play it on silent for now just just so that you can see what is happening. So the next scene, Iris is applying sunscreen and makeup onto Zoe so that she's prepared to meet Cleopatra. But you can see Zoe is staying consistent throughout every single clip. This was a major issue with AI-generated clips before, was that the characters shifted from one frame to another. And because AI-generated video is usually 8 seconds to 15 seconds. That just looked really weird, right? And then this is the other one that she is doing. And you can see now she is meeting Cleopatra. And so once we're happy with all of our videos, and if you're not, just click Recreate here, put in your prompt for what you don't like. And Claude and Higgsfield is going to regenerate this for you. You can even tell it to regenerate it from like a different platform like Cling or Veo. Okay, so you can see the rest of these videos here. Now, there was probably like 1 or 2 videos I had to regenerate, but overall, I was pretty happy with this. Now I'm going to download all of these videos and upload them into Claude in the order that they're supposed to appear and get it all stitched up together. And I'll be back in a second to show you the results. Okay, you can see here I have uploaded all of my videos, all nine clips, and here is the final video. So let me play it for you.

So I'm in Alexandria, 41 BC, and Cleopatra agreed to let me follow her actual morning routine. They're taking me to get presentable, the implication being I'm currently not. Okay, this is Eres. She runs the entire beauty operation. What's the eyeliner made of? Ground stiblite. It keeps the sun from her eyes and the flies from her lashes. So the most famous look in history is basically sunscreen? Protection first. Beauty is what survives. She's putting it on me now? I didn't ask questions. Now you are fit to meet her. Iris did well. Her braids move as one piece. That's a wig. Beeswax and human hair. A queen does not sweat in public. The wig sweats for her. So now it's work. Grain reports in Egyptian. Your family ruled Egypt without speaking Egyptian? They ruled it. I wanted to be understood by it. Lunch on the river. Lunch is on the river. Apparently the Nile is the only room in the palace without listeners. Okay, one last question. 100 years from now, what do they say about Whatever Rome decides to write, I have read their drafts. I am prettier in person and better at mathematics.

Okay, so you now know how to use AI to create the 3 most viral types of video: the POV, the morphism, and consistent characters. So I hope you learned a lot from this video. All of the resources will be linked below. Also, don't forget to sign up for my free course. And I also want to shout out Hixfield for sponsoring this video. Love you guys all so much. I will see you soon in my next video. Bye for now!