[00:02] prompts, only to get decent results. But then I realized you can actually use AI to make this entire process way faster and better. So, I found a custom AI prompt generator that turns rough ideas into structured video prompts. And now I
[00:14] AI does it all for me, and the results are actually 10 times better. So, in this video I'm going to show you how it works and how you can use it yourself. in three different modes, and each one is designed for a different situation.
[00:28] step, but if you want to get full control over the results, you'll want to But for now, let's start with the first mode. I call it text only, and it's really simple. So, let's open up our AI prompt generator. For this, you can find
[00:42] below. Once you open it, you'll see this beginner-friendly interface, and it's really easy to use. Right in the middle, you have a text box where you can simply write the idea you have in your mind. So, let's say I want a dirty rally car
[00:55] drifting hard through a hairpin on an open desert track at midday, dust you have a few settings you can choose from. The first one is the model selection. So, if you press this button, you can select the model you want to use
[01:08] Kling, Seedance, and a few more. And the reason this is really important is that every AI video generator has a different way to respond to prompts. So, one prompt that works amazing for VO might not look as good inside Seedance. But if
[01:21] you select the model you're creating the prompt for, this video prompt generator will give you the best prompt structure for that AI model specifically. Once you choose from, and that's what you want to create with your prompt. So, it could be
[01:34] a cinematic shot, a motion graphic, a B-roll, or even a character, which I'm But for our example, it'll be a cinematic shot. Now, let's press generate and wait for the model to create our structured prompt. And as you
[01:46] can see on the right side, we got our optimized prompt for Seedance. It might normally used to, because this is in JSON format. Now, if you've never heard of this format before, it's actually how all the AI video generators read
[01:59] prompts. So, by giving it in this structure from the start, you actually show you what's inside. As you can see, it created a category where it describes [music] our subject, which in this case is the dirty rally car. Then the action,
[02:12] turn. And then it gives a few more details about the scene. But what I gives us the shot settings. So, let's see how the camera will look. As you can see, the type is wide with a low angle. mentioned that it's using a 35-mm
[02:27] you wouldn't even think about adding to your prompt if you wrote it by hand. Then it also created a few more sections like the lighting, the color palette, and the style. And besides the duration and the aspect ratio, it even added a
[02:40] negative prompt, which tells the AI what not to do. And that's considered one of high-quality prompt. And because writing prompts with other AIs usually means going back and forth between them and the video generator, this tool makes the
[02:53] as you've seen, we simply write a prompt. And once we get the structured and optimized version back, you can press this copy button at the bottom. which takes you straight to the AI video
[03:06] focus on getting results rather than switching between tools. Now, if you you can use the link in the description down below to create one. From here, we just pick the model we want to use, which in this case is SeedAnz 2.0. Then
[03:20] I'll paste the prompt we created earlier. Now, for the settings, I'll go with 10 seconds for the duration, the resolution to 1080p, and the aspect ratio to 16 by 9, just like we got from the prompt generator. Now, let's press
[03:32] generate. And here's the result we get back.
[03:45] how the rally car is drifting and creates that dust cloud behind it. And lighting both look really on point. So, just like that, we have a fully staged 10-second clip from one sentence and a few clicks. But, this text-only mode
[03:58] let's go to the second mode where things start to get a bit more advanced. [music] And that's because instead of letting the AI invent the whole scene from scratch, we bring our own starting frame to it. But, before we do that, we
[04:10] first need to create an image. And for this, I'll be using Higgs Field again. But, this time, I head into the Cinema Studio section and pick GPT image 2 as the image model. Now, Cinema Studio was trained only on real movies, which means
[04:22] that no matter what prompt you give it, the result comes out cinematic. So, let's paste in our prompt. The aspect ratio stays at 16 by 9, and for the hit generate. And let me show you the result because the quality is honestly
[04:35] insane. As you can see, the image is really clean, and our character actually looks like one from a Pixar movie. The lighting is perfect, and the colors are on point. So, now that we have the starting frame, we can finally go inside
[04:47] our AI prompt generator and create the video prompt. this one, I'm not going to start by describing our character or the situation. I'll simply say something will create a prompt for us that we can paste alongside our reference image
[05:00] inside the video generator and get the result we want. Now, before I press set to seed ants. And for the style option, I'll go with stylized, which is custom for non-realistic looks like anime, claymation, and cartoons. And
[05:13] I'll hit create. And here's the result it gives us back. It's the JSON details we want. And you can read through it. But, if for any reason you simply press this regenerate button at the top until you get what you want.
[05:27] Once it's done, just press the copy button or the Higgs Field button on the right. From here, we head to upload media to add our reference image. Once automatically check for eligibility. Just wait a few seconds until it shows
[05:40] as eligible. After that, we paste in the prompt the AI gave us. I'll keep the settings the same and generate. After a few seconds, here's the result.
[05:59] So, the prompt did a really good job describing the action. But, the most impressive thing is how the water flows. It respects the laws of physics, which is something you don't usually get from the first try. So, the prompt generator
[06:12] consistent results that go beyond just one shot, let me show you the third method. With this, we'll use pre-saved elements like a character and a location multiple clips. So, the first thing we need to do is head back to image
[06:26] generation inside Cinema Studio. I'll make sure GPT image 2 is selected and can see, I want to create an astronaut to be my main character, and I'm looking for something realistic. One small trick I use for that is adding the word
[06:40] not seem like a big deal, but it actually made a big difference when I was comparing it side by side with characters that didn't have it. So, make Now, with GPT image 2, we can also select the quality, which is quite
[06:54] different from the resolution. This quality feature is only available inside low to high. Where high is going to take a longer time to think, but the result generation. In my opinion, she looks extremely real. And even for a single
[07:08] shot, she has all the details you might think of. She even has the American flag her other hand. So, now let's use a similar process to create the location. For this, I'll use pretty much the exact
[07:20] that I'm going to change the prompt to one that describes an aircraft hanger. And here's the result we get back after a few seconds. It has all the details I plane back there on the left side. So, now I'll go back into our AI prompt
[07:34] going to describe all the details about my character and the location because those from the reference images. All I'll say is this, a returning astronaut in a sunlit hangar lifts her head and says quietly to camera, "I look forward
[07:48] that I was part of it." I'll choose cinematic for the option and press create. Now we have our Jason prompt that's been optimized for seed ends. So to the video section. Now let's press on create video and here I'll make sure
[08:02] seed ends 2.0 is selected as the model. The first thing I want to do is press on upload media and drop in my character and the location we created previously. paste what I got from my generator and I'll make sure to use the ad tag to tell
[08:16] character. Next, I'll make sure the audio is on. Let's hit generate. After about 30 seconds, here's the result we get back. >> I look forward to the day when I can look back and say that I was
[08:29] when I can look back and say that I was part of it. naturally looks at the camera and starts speaking. And as you can see, the location and the character stayed consistent throughout the entire video,
[08:44] but that's not only because we attached the reference images. It's also because we used a detailed prompt specifically created for this scene. And we were able generator. So if you want to get these types of results without spending hours
[08:57] mastering every prompt, go try it for free using the link in the description down below. Thanks for watching and [music] I'll see you in the next one.