The ONLY 3 Tools You Need to Create ANY AI Video

If you want to start making AI videos, it's easy to feel overwhelmed. There's hundreds of different AI tools out there, and unless if you're in the space every day, it's impossible to know which ones are actually worth using. Luckily, this is all I do, and I've tested pretty much all of them. It turns out that the best AI videos can be created using just three tools. So, today I'll show you what those are and how to use them in a complete workflow to create the best possible AI videos. >> [crying] >> Quick. Jump. >> The first tool is an AI language model called Claude AI. Now, Claude doesn't generate AI videos on its own, but we can use this language model as an automated way to write prompts that create the AI videos for us. Now, we could just directly ask Claude AI to help us start writing prompts, but what I found works better is to actually have a customized skill. And what I have here is a skill where I've trained the AI to write prompts exactly the way that I want. First, I taught it how to write a prompt for a design sheet. And the design sheet will lay the foundations for the visual concepts, including how the characters look, the color palettes, anything related to the general visual style. I've also taught the skill how to write a prompt for a storyboard, where we can take our design sheets and actually take those concepts and put them into a series of shots that tell a full story. And finally, I've also taught this Claude skill how to actually write prompts for the AI video generators to turn our storyboards into full AI scenes. I'm going to put a download link to the skill in the description of the video, so you can go and grab this for yourself. Once you have the skill downloaded, in order to use it inside of Claude, first you're going to find this customize feature. Then, inside here, go and create a new skill. I already had the skill installed inside of Claude, but what you're going to have to do is go and create a new skill, and from there, upload that skill file which you've downloaded. So, you'll just have to drag and drop that in there. It's telling me to upload and replace this skill since I already have it inside Claude, but you'll be able to just directly upload the skill. Once you have the Claude skill installed, the first step of actually using this skill is creating a design sheet for the visual concept of the AI video. So, inside Claude, I'm going to tell it to use the AI video prompt writer skill, which is the skill that we've just installed. And then I'm going to describe the general idea for the AI video I want. I'm going to create a short AI film about this apocalyptic event where a volcano erupted and now there's these two characters trying to escape from the city. And then if you run this, what Claude is going to do is use that skill to come up with a prompt for our design sheet. And inside here, it's going to include the character designs, the the visual style, what the environment looks like, everything that we need to make an AI video. Now, we can't actually use Claude to generate the AI images, which brings us to the second AI tool we need to use, which is an AI image model. And specifically, we're going to be using the GPT image 2 model, which can generate the widest variety of different scenes. So, I'm actually going to copy the design sheet prompt. And then to use the AI image model, I'm using this platform called OpenArt, which has all the different AI video and AI image models. So, the first step is just to create the images for the design sheet. I just need to make sure that I'm actually using the GPT image 2 model, and I already have that selected. And then inside the prompt bar, I'm just going to pasting that design sheet prompt. Inside the settings, I do advise you to at least use a 2K image resolution since there's going to be a lot going on in the design sheet. So, I'm just going to go ahead and generate a few of these. Here's what one of the results I got. It's got the two character designs for the dad and the daughter. It has some scenes of what the world is going to look like. It's a super great and ash-covered world after the volcano eruption. Some of the props that are going to be inside this AI video, and also a general idea of what the color palette and the lighting in the scene is going to look like. Now, what's cool about generating a design sheet first is that you can really customize it and locking on exactly the visual style that you're going for. Let's say you wanted something that looks more colorful. So, then you can go back into Claude and tell it how to change the visual style to match exactly what you want. The design sheet is a bit too gray and drear and dreary. I want some bolder, more vivid colors inside. And then Claude will make those adjustments for you. And it's always a good idea to make sure you locking on the exact visual style you're going for before actually generating the AI videos themselves. So, now that I have my design sheet down, the next step I need is to actually generate the actual shots that are going to occur inside my AI video. And in this stage, what we're going to do is make a storyboard. The reason why I'm creating a storyboard first before generating the videos is that generating images is way faster and cheaper than generating AI videos. And being able to prototype and visualize exactly what all the shots are going to look like is going to save a lot of time and money. So, to make the storyboard, what I'm going to do is actually take one of my design sheets that I was happy with and upload that into Claude. And then all I have to do is tell this AI prompt writer skill to create a storyboard for me. So, I'm just going to write a storyboard sequence that's going to show them getting out of the city and searching for gas. And what Claude is going to go ahead and do is look at our design sheet and use that to come up with an appropriate prompt that describes all of the characters, the scenes, as well as each of the individual panels and what's going on inside of them. Then all we have to do is copy this storyboard and then go and paste that prompt inside of the GPT image to model. Now, before we just go and run this prompt, what we want to do is actually add in the design sheet as a visual reference. So, here's the design sheet that I generated earlier and I'm also going to drag that inside of the prompt bar. And what this is going to allow us to do is to really hone in on that exact visual style we're looking for. Now, for the storyboard, I would definitely recommend creating them in 4K resolution because we're trying to fit so many different shots into the same image, having that higher resolution is going to give us the maximum quality. One thing about using these AI models is that sometimes there's a disconnect between the language models and the AI image generators. Claude might have an idea of what's going to look good, but sometimes the AI image model isn't going to be able to render everything perfectly. So, if we take a look at this third shot where the girl's pointing her finger, it looks a little weird like she's pointing behind the map. And also, if we look at this shot at the bottom where the girl's looking at the photograph, it's a little weird that the image on the photograph is actually facing us. So, what I did is I actually went back into the prompt and make some minor adjustments. So, in this line here where it says the daughter is looking at the worn photograph she has taken from her chest pocket, I'm going to add inside a line here that says only the back of the photograph is visible. That way we don't have that weird effect where we can actually see the image of the photograph even though it's facing away from us. Here's what some of the resulting images look like. This time when she's looking down at the photograph, it's clear that we're only seeing the back and the front isn't visible. One of the cool things about these AI image generators is that you can even put yourself into the AI shots as well. So, here's an example of a design sheet prompt, and I've actually attached it this reference image, which is going to be a grid of photos I've taken of myself. And if I attach this as a reference along with the design sheet prompt, it's actually going to be able to put me inside of the design sheets and also the storyboards later on. Now that we have the storyboard, the next step is to actually generate the AI videos. The tool that we're going to use to create the AI videos is called C Dance 2.0, which is the best AI video generator by far that's out there right now. So, what I'm going to do instead of OpenArt is go to this video creation, and inside here I'm going to actually select the C Dance 2.0 video model. Then I'm going to hit text with reference. From here, we can animate that storyboard. So, the first step is to actually just upload that storyboard into the video model. I'm going to use a very simple prompt that just describes the sequence of events that's that are happening. Now, I do recommend also adding this line at the bottom that says no subtitles and no music to all of your AI video generations, since these are things that I can add in myself later. And I also need to help the AI video generator a little bit by telling it what the storyboard image actually is. So, I'm going to type at, and it's going to let me tag that image storyboard right here. And finally, to make sure the AI actually understands exactly what the character and the visual style looks like, I'm also going to attach that design sheet. And down here, you're going to have some more options for how long the video is going to be. When you're trying to generate a 12-panel storyboard all in one shot, you want to maximize the duration at 15 seconds. I'm also going to show you how to actually extend the videos to be longer than 15 seconds later on as well. So, let's take a look at what the results look like. So, not bad. It actually did go through our entire storyboard sequence. However, the problem with this method is that we're trying to fit 12 separate scenes all into that same 15-second window, which can work for some applications, but 15 seconds really isn't long enough to animate an entire storyboard. And a lot of the times when you try to animate 12 shots inside the same 15-second window, it's not going to animate all of the shots accurately. It might skip a few of the shots here and there, but if you want to create something quick and dynamic like a fight scene, for example, this can actually work pretty well. So, instead, here's a different method I would recommend if you want the maximum control. So, instead of animating the entire 12-panel storyboard at the same time, what I've done is actually cropped out each row of the storyboard into a sequence of four panels. Now, I did this inside Photoshop, but you can use a bunch of free tools like Google Slides to do the same exact thing. And the AI should do a much better job of the animating all the different shots and also keeping the scene looking exactly the way that I want it. Now, when we wrote that prompt for the storyboard earlier, the Claude skill actually also wrote down AI video prompts at the same time that go along with each of the individual shots. So, here's what the AI video prompts look like, and they've been written for each of the 12 shots inside the storyboard. Now, we're only going to be animating four of the shots at a time. So, I'm just going to copy the prompt from the first four shots. Then, I'm going to upload that prompt storyboard panel, which has one row with four different shots inside of it. And then, I'm just going to pasting that prompt from earlier. Uh let's actually divide this a little bit so it looks neater. Now, I also do need to give the AI video generator a little bit more context. So, at the top, I'm going to adding a line that says, "Generate a scene using shots in the uploaded film storyboard sequence." And then, leave a tag for image one. Now, let's see what happens when we just generate these four panels at the same time. It's a much slower-paced video. It's still going to make uh some subtle errors. So, for example, up here at the start of the second shot, you see the car actually moving forward a little bit. It should be static, but you can just cut that out pretty easily. I also went ahead and used the same method to generate the other shots inside the storyboard. Now, I also thought that it would be a good idea to adding a little bit of dialogue to some of these shots. So, back inside Claude, I'm going to tell it to add some short dialogue in shots 9 through 12. And it's going to writing some dialogue for those last four shots. >> Manual pump. Tanks underground. Ash can't touch it. >> It's working. Don't stop. >> We've got enough to clear the city. Maybe reach the ridge by dark. >> Do you think anyone made it out before us? >> Yeah. Someone made it. >> So, there's one scene here that doesn't really make sense of him pumping gas into the can. It looks like he's just I'm not really sure what's going on here. But, the rest of the shots turned out pretty well. And now I have three separate 15-second video sequences, which makes an entire 45-second scene. And you can keep generating more shots in the same way as long as you want. So, if you follow this workflow and use these three AI tools, you'll be able to generate just about any AI video you want. Now, if you want to learn how to make AI videos better than 99% of people, go watch this guide right here.