Every shot tells a story. Peace versus chaos, scale versus focus, isolation or connection. The best AI filmmakers are masters at using the AI camera to direct cinematic scenes that tell a story. So, today I'll show you how to take control of your AI video generator and create every camera movement you need to make any AI film. The first camera shot we're going to talk about is actually the static shot. There's no motion from the camera, which is perfectly still. This is the most basic camera movement. Since there's no motion at all, the focus is entirely on the composition of the scene, the characters, and the atmosphere. When the camera doesn't move, every movement inside the frame matters so much more. Use the words static camera in your prompts to create this effect. While we keep the camera completely still, we can still create a sense of motion and movement in the scene by changing the focus between different elements. In this shot, we can start with the camera focused on the woman's face, but then shift the focus to the revolver on the table. This is called rack focus. When writing prompts using this effect, make sure you also describe exactly how the focus in the scene is shifting. We can shift the focus from the woman to the revolver on the table or go in the opposite direction where we shift the focus from the foreground to the background characters. The prompt I use here is rack focus. The camera switches focus to the people in the background. The man wearing the burgundy jacket in the background drinks a glass of whiskey. The next camera motion is a dolly shot where we physically move the camera forwards or backwards. This will physically bring us closer to the characters in the scene or you can dolly out to pull the camera physically backwards farther away from the character to show the scale of the environment create distance to make all the different camera movements. I'm using imagetovideo AI models where I first start with a reference image and then prompt for different camera movements. The platform I used is called Open Art which has a huge selection of all the different AI video models. I found the one that typically works the best for creating different camera movements is the new Clink 2.6 model. So, if I upload a reference image of my scene, I can enter a prompt that describes the camera movement that I want. One great way to control the camera movement is to describe whether you want the movement to happen quickly or slowly. So, for example, if I use the word fast in front of my camera motion, fast camera dolly in and for the camera to quickly move forwards towards the woman, it's going to create a much quicker dolly in effect compared to if we just asked for a regular camera dollying movement. You can also create this vertigo style camera effect where the size of the subject stays the same inside the scene, but the background moves forwards or backwards at a different speed. This creates a sense of disorientation. The best way I found to create this kind of effect is to prompt for a dolly zoom and then tell the AI that the background environment should get larger or smaller while the subject's size remains constant. Now, to create all of these different shots of my character, which I'm using to animate with these different camera motions, I'm using an AI image generator called Nano Banana Pro, which I think is the best AI image model out there right now. I'm going to go to the image creation tab in Open Art. I already have the Nano Banana Pro image model selected here. Uh, if you see something else, make sure you switch it to Nano Banana Pro. And I'll start by creating a full body photo of my Asian female outlaw character. She's tall. She's wearing ragged and torn clothing and a hat. And I'll adding a little bit of visual inspiration inspired by the Badlands inside a gritty sci-fi western film. Once you have a base image like this of your character, we can create all sorts of different shots of the same character in different scenes, environments, different poses, camera angles. What you'll need to do is upload that full body shot of the character into this image guidance. Then in the prompt, I'm going to describe a new scene for my character to be in. I can prompt for a lowo close-up shot of her sitting at a table inside the dimly lit salon. And this way, we can generate a ton of different images, which we can then turn into AI videos with all of our different camera movements additing. So, we've seen a static camera and how to move the camera forwards and backwards, but we can also start moving the camera side to side to create a horizontal movement. This is where motions like the camera pan where the camera rotates on a swivel and can follow your character walking around the scene or rotate around to show more of the environment that's in the background. Or you can use what's called a truck camera movement where the camera physically slides on the track left to right. So in this case, the camera is actually sliding along on the track as it follows the character walking. So the thing with AI is that uh it still doesn't quite understand the subtle differences between these different camera movements if you're just using the prompt. While the AI does understand the difference between left and right, it doesn't exactly know the difference between the panning movement where the camera's rotating or the truck movement where the camera's sliding on a track left to right. So, how the camera movement turns out will largely depend on the way the image is framed and the motion you described. If you have a side profile shot of your character and you use a prompt telling the camera to go left to right and follow your character, it'll typically end up actually doing a pretty good job of creating that truck left to right movement where it feels like the camera is sliding on that track and tracking the character walking. However, if you're starting with a reference photo which has a front view facing your character and you try to animate the camera moving left to right, typically the camera will rotate at least a little bit. But when you want it to move left to right, also if you use the keywords camera rotate, that's also a great way to create this rotational movement. We can also have the camera move vertically up and down, which is going to be the tilting camera movement where the camera is going to rotate upwards or downwards. In this scene I created for a Viking funeral, I told the camera to tilt up as the smoke from the ship rises into the air. You can also have the camera moving directly up and down using the pedestal up or the pedestal down motion. So here I had the camera directly pedestal up to show the entire monastery building. You could combine different camera movements where you're moving the camera around in different ways while rotating. for example, uh telling the camera to move directly upwards and also point downwards at the same time. Here's a secret you need to know. To create the most advanced camera movements, it's not enough to just write a prompt. If I want to have my Viking character dive into the water and they have the camera follow him, I can write a prompt that describes that movement. He jumps off the ship into the water and the camera follows him by diving into the water as well. However, when you're trying to make big movements like this, the AI is going to have a really, really hard time keeping the character looking consistent. And he turns into this blob at the end. The way to actually create more complicated shots like this is to use multiple image references as frames of the video. When you're starting with multiple image frames, it gives AI a lot more hints to work with as to what you want the final video to look like and it does a way better job of animating this scene of my character diving underwater. So to create this effect, first you need to create the different image references. So I started with this image of my Viking sitting on the boat and then I uploaded that image as a reference into Nano Banana Pro. Then I told that he had to generate a different shot of this character except now the Viking is swimming underneath the water after jumping in and the camera is completely submerged underwater. So now I can use these two image frames of my character to guide the AI video creation process. Now using the cling 2.6 model and I'll use both of those frames to guide the AI video creation. Most of the work is done at this point, but I still need to provide just a little bit more detail in the instructions for the AI video. So, I'll say that the Viking with red hair, he puts down his ax and dives into the water, and this is going to create a far more physically accurate shot of what's happening. Now, you're probably wondering, will these prompts work for all the different AI video generators? Maybe you're using Google Veil or a different model. Well, different AI video generators actually will produce different results. So, let's take a look at a comparison. Here's one I made using Cling 2.6, and I asked for the camera to tilt down and dolly in on the Maz boots. So, the camera movement, I would say, is accurate. It does the motion I prompted for, but it's a pretty subtle and slower movement. Now, if I use a different AI video generator like the C Dance 1.5 Pro model using the same exact prompt and image reference, it actually made a much bigger motion. The camera tilts down more and it also dollies in and moves forward faster. However, that doesn't mean C dance is going to be better in every instance. Here's another shot I created. This time with a new camera movement, the camera orbit where it's going to rotate around the character on the platform. Using Queen 2.6, it generates the camera movement accurately and all the characters in the scene look pretty reasonable throughout. However, if I used C dance 1.5, while it creates much more of a rotation movement, but if you look at the characters on the platform itself, there's some changes. So, it starts with a bunch of captives standing on the platform and somehow at the end when the cameras orbited around them, it turns into just two characters with the woman sitting down in the chair somehow. So, I would definitely recommend trying out the different AI video generators instead of just sticking to one. And that's one of the benefits of using platforms like Open Art which have a huge collection of AI video models is that you can switch between them to get exactly the camera movements you want. So, the orbit camera movement basically circling around the characters. The best way to control the way this movement works is not necessarily telling it to go left to right. Instead, you want to use the characters or objects in your video as landmarks to guide the AI camera. So, if I tell the character to rotate behind the Viking, the camera will move in that direction. >> Head is not clear, my son. However, if I tell the camera to rotate to the back of the woman, AI will actually direct the camera to orbit in the opposite direction. >> I see it. Now, >> you can also create a 360 orbit all the way around your character from front to front. This is one of those new camera movements that really wasn't possible until the newest AI video models we have right now. The way to make a shot like this is to start with two uh pretty identical shots of a front view of your character. All I did between them is in the image on the right, I slightly changed his face to be looking upwards so the AI knows that there should be a little bit of movement involved. Then I uploaded them using the same start and end frame feature where we're using multiple frames to guide the AI video. And I'll instruct the AI for the camera to orbit completely around the Viking as he looks up into the sky. So this 360 camera orbit around the character is definitely one of the hardest movements for the AI to animate. You will get some bloopers like this video here where it does rotate the camera around the character, but it kind of looks like uh I guess he splits into two people in the middle of the video. A more reliable camera movement to animate would be the 180 degree rotation. In this case, we're starting from the front and rotating around him to the back. And I found that this type of camera rotation tends to be a bit more consistent. Now, when you're telling an AI story, it's not just the camera movement that you add into the scene, but also the shot selection, the frame of reference of how you're seeing your characters. Whether it's a bird's eye view from above that shows the expanse of the environment, or a shot like this, which is uh basically a full body shot of your characters with the legs cut off. There's different shots like an over the shoulder shot where we're looking over the back of one of the characters at another one or a close-up shot which will bring us a lot closer to the characters themselves. All of these different shots were made using the Nano Banana Pro model where I started with an initial image and then I used the AI to change up the camera angle, zoom in, zoom out, and put the characters in different types of environments. If you're generating camera movements using prompts or multiple different image references, the more effort you put into describing exactly how you want the shot to turn out, the better and more consistent your results are going to be. Sometimes you will feel like the AI kind of distorts the world a little bit in this rotation shot. Something just doesn't quite feel right. almost as if the positioning of the camera like she's almost rotating around a little bit too much for it to look natural. So instead, I adjusted the prompt for the camera to orbit around the back of the robot and then dolly in to a close-up shot on the woman's face. and the world in general seems to be more consistent throughout the shop. Camera motions are an extremely important part of making AI videos. If you want to learn the only seven prompt styles you'll ever need to create any AI video, go watch this guide right here.