--- title: 'Create Consistent AI Videos with ChatGPT Image 2.0 and Seedance 2.0: The Storyboard Method' source: 'https://youtube.com/watch?v=tbG7ccbRn2o' video_id: 'tbG7ccbRn2o' date: 2026-07-24 duration_sec: 660 channel: 'Arca Artificial by Lordwind Enrique ' --- # Create Consistent AI Videos with ChatGPT Image 2.0 and Seedance 2.0: The Storyboard Method > Source: [Create Consistent AI Videos with ChatGPT Image 2.0 and Seedance 2.0: The Storyboard Method](https://youtube.com/watch?v=tbG7ccbRn2o) ## Summary This video presents a storyboard method for creating consistent AI-generated videos using ChatGPT Image 2.0 and Seedance 2.0. The creator demonstrates how to maintain character and environment consistency across video frames, making it suitable for social media, ads, and e-commerce. ### Key Points - **The Consistency Problem** [00:02] — AI video generation often struggles with maintaining consistent images over time, including faces and environments. - **Storyboard Method Overview** [00:44] — The method uses film and advertising techniques adapted for AI, leveraging ChatGPT Image 2.0 for consistency and Seedance 2.0 for movement. - **Prompt Structure for Storyboard** [00:58] — The prompt includes: 1) General structure (e.g., 12-panel storyboard), 2) Character description or reference image, 3) Setting, 4) Story of actions, 5) Movement and transitions, 6) Visual consistency (clothes, lighting, etc.). - **Using ChatGPT in Thinking Mode** [03:12] — For better results, use ChatGPT in 'thinking' mode (not instance mode) to generate the storyboard prompt. - **Generating the Storyboard** [03:42] — Upload a reference image and provide a prompt; ChatGPT creates a storyboard with sequential panels showing the character from different angles. - **Transferring to Seedance 2.0** [04:38] — Define whether the video is a continuous shot or multiple takes. Use the storyboard images as start frames and references for movement. - **Tagging Images in Seedance** [06:01] — Upload storyboard images to Seedance, tag them as image1, image2, etc., and specify image1 as the start frame. - **Two Approaches for Video Generation** [08:05] — Option 1: Use storyboard images exactly as frames (results in cuts). Option 2: Use storyboard as reference only (more fluid movement). - **Applications** [08:47] — The method works for social media, ads, e-commerce, product sequences, and storybooks, maintaining consistency throughout. ### Conclusion Combining ChatGPT Image 2.0 with Seedance 2.0 via a storyboard prompt enables consistent AI video generation, useful for various commercial and creative applications. ## Transcript always has a constant problem, which is creating consistent images that never change as time goes on, the face never changes , the environment never changes. That's why today I want to show you the storyboard method you can use with Image 2.0, taking advantage of the good consistency of this model and then bringing it to Sinans 2.0 Zero to create very good videos with good consistency and very good movement, and thus create videos like the ones you are the channel, I am Lorwin Enrique, captain of the artificial ark, a channel specializing of the artificial ark, a channel specializing in images or videos. Let's begin. techniques that are applied in film, advertising, everything is being brought into AI and the results are very good. But to take advantage of it, you have to understand the prom. And as always, I have the best promo that at least works for me, and it's the following. The first thing you need is a general structure where you will tell CHDPT to create a 12-panel storyboard for you. In this case you can change it, you can make an eight-panel storyboard, you can call it a 3x3, you can call it a 4x4, whatever you want. You can vary this structure to suit your needs and add any formatting you want. It can be 16916 or whatever you're looking for. The second thing is the character, the vehicle, the object, whatever will be the protagonist of your video. And you're going to put, in this case, if you want to add an image, you can say that it's based on this image that I'm uploading. If you want to create it directly without a main image, don't say it's based on the image, but rather, "Hey, create a character like this for me." Describe the age, body, hair, skin, all the descriptions you want of your character, you can put them. Third, the setting. Where is the is the object? the whole atmosphere. Describe it, describe it as best you can so that you become a dept. The story of the actions describes the story you want to happen, don't write it step by step, but look, I want this person to be in place A and get to place B, but while they are getting there, such and such happens to them. The story goes that Chas Dept will be in charge of creating the best images. Fifth, movement and transition. Each image should advance the story. I remember that one of the first things I improvisational theater is that for a story to move forward, there have to be actions. And finally, visual consistency. You have to emphasize to him that in the 12 panels or in the eight or wherever you want, in whatever quantity you want, he has to maintain the same clothes, the same character, the same environment, the same place or the same lighting. You must add all of this. If you come to GPT chat, always remember to bookmark and create an image. In this case, I'm going to upload a main image which is of a girl standing in a blank space. What I want is for her to have consistency because we're going to show the garment she's wearing. It's going to be like a kind of model without Remember that the other thing I've always explained in the CHPT Image explained in the CHPT Image 2.0 videos is that you should put it in thinking, not in instance, but in thinking. Important if you don't know how to stand out or write because you say, "My pronunciation is very weak." What you have to do is open another tab in CHDPT and tell it, based on this prom, two points, you place the prom, create a prom for my character. You upload the photo based on my character because I want to make a storyboard of her showing the dress she's wearing. And when you click create, it will analyze the prom. This is what you'll get because it's in Thinking, and as you can see, it will generate the complete promo, adapting it to what you explain to it. I only gave him something simple, but you can explain many more things. This is an example. After I click create, chatpt creates the storyboard where the girl is facing forwards, is from the side, and shows me eight steps, which is what the girl should do in the video. As you can see, creating the storyboard is simple; it only requires an image and a prompt, and it creates the steps according to what I asked for. In this case there were eight, in others it may be 12, as you wish, but once you are here you have to take it to video. And what are you going to use? Sedans 2.0, but you also need a prom. I'll show you. For this type of video, the first thing you have to do is define whether it's a continuous shot or if it's going to have different takes, but you have to define that in the first part of the promo. Then you're going to talk about character, okay? If you use image one, image two, exactly the same, I'll show you so you can take that image as a reference, not only as a reference, but as a start frame, that is, as an initial frame that will start from there. Then there's the character's movement, which is what they're going to do, and that's where the storyboard comes in. You're going to use the second image that I'm going to put together for you, you're going to label it so that it creates the movement and follows that sequence of actions. Four, you're going to add certain addresses, but with the promo that I'm going to give you, which you'll also have on the website to download along with the image promo. So, what you have to do is give it to chatt or the LM you want so they can adapt it to what you're looking for. Five, define the tone of the movement and the feeling that the video will generate. And finally, you can add special effects, music, all of this you will add in the last part of the prompt. If I come and Hitfield, you can do it on whatever platform you want, Freep, whatever you want, you're going to do it because you can also tag it, okay? So, if you look here, it already tells me image 1. So, what I do is upload the images to uploadat. In this case, I already have it loaded. Select the first one storyboard. In this case, image 1 of my prom, I write @ and select image one. If you keep pressing, notice image one. If you keep pressing, notice that it will say image 2. So @ image 2. Here I'm going to tag it again. I'm going down a bit because here it's going to tell me down a bit because here it's going to tell me that image 1 is the start frame. I need to label it so Sidan knows where the video starts and what the video starts with . When it's like this, all I have to do is click generate, and while it's generating, I'll show you other examples I made. Yes. we made of the garment has been generated, see how it looks. generates the video as the images we place on it are displayed. As you can see, it's her facing forwards, she's here, then she's to the side, and that's what's happening in the video. Note, there are two ways to do this: you can do it this way or simply have Sidan use the storyboard as a reference, and you do that through the prom. You don't tell him to be exact, but to take it as a be exact, but to take it as a reference and the result can be like this. completely different. One is more fluid and the other has some cuts, but both work. But let's look at this from a different perspective, beyond just "yeah, it looks cool." This can work for social media, for ads, for e-commerce, for product sequences, for storybooks, like the one I just showed you of the girl, what's up? This can be a consistency is maintained, and that's something that many people, including you, are surely looking for in their videos. The truth is that Chat DPT Image 2.0 came out quite powerful, Chat DPT Image 2.0 came out quite powerful, and mixing it with Sidan 2.0, which I showed you in the video where I talk about how to create a complete branding, and by the way I tell you what works and what doesn't work without lies, without mincing words. I'm telling you that too, and you can see it in this video here because the truth this video here because the truth is you can get a lot out of it. In other words, there are many ways to take advantage of it, and this is one of them. I'd like to know if you enjoyed this video. Please leave a comment, like it, share it, and recommend the channel. That way, I can reach more people who don't know about and want to learn how to use artificial intelligence in this way. watching the videos for a while but have n't yet decided to join the Telegram community, I'll leave the link in the description because it's completely free. I'm always posting information there, we're asking questions, and that's how we can stay connected. On the other hand, if you have a project and don't know how to manage it and need advice, I offer one-on-one consultations, where we sit down for 75 minutes to talk and work on your project. You'll find that information in the description. You can write to me if you have any questions and I'll gladly answer them. Remember that this is the artificial ark. Here's another video that's really great for creating videos with artificial intelligence, and here's my face so you can subscribe directly. See you in the next video.