Top Tier AI Video is Finally Affordable - Hailuo AI

I recently tested one of the best video models that I've ever used. It's called Hyo 2 and it's ranked number two on the artificial analysis texttovideo arena leaderboard right now. And as you could see, it outperforms some of the top models that I've covered before like Google's Veo 3 and Cling 2. 0. Now, this comes from a company called Miniax that I partnered with for this video. And if you go to their product page, they actually have all kinds of different AI products. So, they have a chat platform similar to chat GPT. They also recently rolled out an agent platform that's really good, too. Worth a try. But for this video, I'm going to focus on their video model. This is Hilu and I'll link it in the description below this video. So, you could try it for yourself. Now, I'm going to show you a ton of examples. And I come from a video production background. So, I have a lot of experience making commercials. I'm going to show you a lot of commercials and business applications for this. And I also come from a filmmaking background. So, I'll show you a ton of cinematic options as well. couple of things that really set this apart. One of them is stronger prompt following capabilities. This actually achieves a 85% success rate when it comes to complex and nuanced instructions. This is also extremely good at more realistic and natural physical motion. So for this example, extreme physics, right? If you're doing any type of acrobatics or if you have any intricate movements, I haven't seen any model be able to do it as close as this model can. And this also has improved character consistency. So it maintains stable character identity across different scenes and different angles. And the general visual aesthetics of this is a big improvement over the previous models and compared to other video models out there. So you're going to get richer lighting and it has a cinematic quality to it. And when it comes to these type of models, that is a place where I really want good cinematic quality out of my clips. Okay, so let's dive into this platform. Again, link in the description. And the place I like to start is just under the video tab over here. And if you look on top, you're going to have three different ways to generate a video. You could do image to image. So starting with an image here. And they do have an image generation platform here on this tab as well. Now they also have text to video where you could just start with a text prompt. And typically before I start using this, I usually never use text to video. I always generated an image in any video platform and then use that image as a starting point. And I still like to do that from time to time, but the text video option here is actually really, really good. So, I'm going to show you a ton of different examples with that. And then you also have subject reference where you could actually give it a reference and then it will generate a video based on that. Okay, let me start here with text to video to show you some examples. The model I'm going to show you is the O2 model over here. Now, you have two different resolutions. So, if you're just testing out prompts, I would always start with this one to get what you're looking for. And then when you are happy with that, you could upgrade to the 1080p resolution. This is credit based, so depending on what you're using it for, it will use up credits to generate. And then you could do 10-second clips or six-second clips, which most of the shots that I do are going to be six-second clips over here. And then you could choose how many quantity output you want from a single text prompt. Now, I'll start with some commercial options. Since I used to create a lot of high-end commercials, so I really want to see how far I could push this model to give me shots that would come directly from a real commercial that you would see on TV. So create a slick slow motion video of a luxury perfume bottle on a rotating pedestal with micro closeup of glass, liquid motion, and gold details. Okay, so that's going to be my prompt here. And you'll see a couple of different icons down here that are interesting. The preset option is interesting because you could choose from all kinds of different presets. Like golden magic hour is going to give a unique look that usually happens during magic hour in a day. So if you select any of these, it's going to overwrite your prompt, but it's going to create that environment and then you could add to that prompt. But this gives you a nice starting point if you want. And then what's really nice is if I paste that prompt back in, you'll also have camera control and it gives you a little preview of what those moves are going to be. So again, if you don't know anything about cinematography, well, this is really nice and easy to be able to scroll over each one and get a little preview. So upward tilt, right? That's going to be hard to explain if you don't know anything about how cameras move, but then you could just click it and you could see push in pedestal up. That's the camera movement that it's going to add to the end of your prompt here. And then this option right here, I usually leave this on. So when this is enabled, your prompt gets refined to enhance the generation quality. I do recommend this to be on for pretty much any time you generate with a text prompt. Okay, here's the first option I got. Wow, that looks fantastic. Let's see what the second one gave us. Oh yeah, I like this even more. Now, I didn't describe the shape of the bottle, so it will follow your prompt very closely. That's one of the big upgrades that came with this O2 model. So, make sure you give it as much

detail as you want, including shapes, size, color, lighting, and again, you could use some of those presets. Now, I'm going to show you more examples of text to video, but I wanted to see what happens if I upload an image of an actual perfume bottle. So, I have this image of a perfume bottle and I'm going to tell it to use that same exact prompt. Okay. The only thing is I'm going to move gold detail. I don't want this to be in gold detail. But the rest of this prompt is exactly from the previous generation. I'm going to generate this one. Okay. Let's see what we get here. Wow, that looks incredible. And look at that. It zoomed into the bottle and it gave us that macro closeup of glass and liquid motion. exactly what we asked. The text remained on screen the whole time and it looks like that exact bottle the whole time. It doesn't morph or anything. So for product e-commerce commercial applications, this is wow. This is really excellent. Okay. Now I wanted to take this to the next level. So I said put it in a forest and have sun rays hitting the back of the bottle and add the brand gold on top elegantly embossed on the bottle. Let me show you this one. Wow. I mean, that looks really, really good. Now, this one's fantastic for an ad and it has text on top of it. So, this is a model spinning in a dress with bold text reading 50% off. And this is a great use case for commercial applications. Now, I wanted to see how far I could push this. Something that would require a team of animators typically to put together. So, a tiny bustling town made of entirely of fashion, accessories, miniature citizens. Let's take a look at this one. I mean, that's pretty cool. That looks great. Okay, here's a complicated one. A lone glass bottle of milk stands in the middle of a golden wheat field at sunrise. And then a bunch of white birds are going to come grab the bottle and then take it to a city and put it in front of a child at a breakfast table. Okay, let's see if we could get this one. Let me go full screen with this one. Okay, that's my favorite so far. Okay, here's something you might see in a commercial. For example, a B-roll shot. Two strangers sharing umbrella during unexpected rain in Tokyo at night. Neon reflection. Right. It followed a lot of that. Exactly. Right. I tried more of an animation prompt here. stylized animated intro for enchanted forest awakening at sunrise. Again, it's following a lot of the details of that prompt. Light beams through trees. It's doing exactly that. This one could be from like a hiking commercial. Group of hikers reaching mountain summit, gearing up with backpacks. That looks pretty good as well. Now, here is an interesting prompt. I wanted to see how it would handle a montage. and it decided to kind of do this picture and picture effect on the very first prompt I gave it. And it's pretty good. I mean, look at the details on every single clip. The person looks good, the animals look good, the movement looks good, and the transition looks good, too. Okay, I wanted to see how it handles more complex movement. So, this is warriors in the mountain summit kind of having a sword fight. And for the most part, this looks pretty good. The movement makes sense. the snow. You want to look at the snow to see how he moves based on the movement of these warriors. And yeah, I don't really see any major issues with this one either. I try to run it again. Okay, let's see how it came out the second time around. Again, pretty good. I wanted to see what happens if I start with just one word prompt. So, I just typed in apple and he gave us a really cool cinematic looking apple from what you might see in some kind of an ad. So, that looks pretty good. Now, this is interesting. The prompt said dreamlike scene of a child floating above a glowing forest at night. And I did not give it a specific set of instructions to make it any style. And this is the style it chose by default. Based on the prompt, it kind of makes sense that this will not be a real life and more of a cartoon style look. But I mean, everything looks great about this, too. Now, the few times I saw issues, this one actually works out almost perfectly. And I just rolled this one time just so I'm not re-rolling to get exactly what I'm looking for. Typically in the second or third roll, I got what I wanted almost every time. But the prompt says, "Depict a real estate agent preparing a house for open house. " And the only weird thing is because it's a modern kitchen, she tries to like open the door with some kind of touch screen that is actually not there. But other than that, it's pretty good. Here's one from a sci-fi movie here. Spaceship in the cloud. smashing into a frozen lake. Okay, so it got a lot of that glass

cracks and system sparks. So glass cracked. I guess the glass completely shattered because it disappears. But for the most part, I mean, it follows the physics of things pretty well. I don't really see a major issue except the glass disappearing like that and no residue of it. But again, it could just get lost in this water here. Now, next, let me show you the image to video because a lot of us actually are really used to that workflow when it comes to video generation with AI. So, I'm going to go to the image tab and we're going to generate some images as our starting point and then we could use those in the video tab here. And you would just again type in a prompt. Again, you could use things like chat GPT. They do have those presets available too. They won't have the camera motion because that comes only with video, but you will have these type of options available here. So, I typed in that text prompt and again I made sure this option is also turned on so it enhances my prompt. Usually get better results that way. And I'm going to show you character reference in a minute because with character reference you could get consistent characters in different angles that again you could pull into your video. And this is going to be in 16 by9. But if you're making anything vertical, you could obviously choose that option or square. And I'm going to get four here for just four credits. Okay. This is one of the images I got for the watch. And you'll see this option. It says video it right here inside of the image tab. I don't have to actually switch over and when I click it, it just brings me directly to image to video and it imports that for me automatically. So that's really nice. You don't have to jump between copy paste, download to your computer or anything like that. And then down here you could actually add any type of motion that you want. So I'm going to delete this one. And if you want to learn more about camera movement, they do have this guide. And this gives you all kinds of different things when it comes to character movement. They also have a director's mode too that you could learn about on this page. But what's nice is you could literally give chat GPT a link to this website or copy and paste the text and it will know how to prompt it based on their set of instructions. So I sometimes use a couple of different AI tools just to get better prompts that way. In this case, I'm just going to use the camera tab here to get all kinds of different camera controls and then I'll hover over each one to see what that would look like. And I'm going to go to the free selection tab. These work a little better for product type of shots. So, I'm going to see if we have a push in. A push-in would be just a slow zoom in into the shot. So, that's perfect for this. And that's all I need right here. And I could go ahead and generate that. Okay, let's take a look at the watch. Yeah, push in. Slowly pushing in. That's the only direction that it has. And then the prompt is an image prompt. So, it's using that exact image. And nothing is falling apart, right? The image is staying consistent. And the camera move is what I want. And at the end, it pushes in into the screen, which I did not say, but it actually makes sense for some kind of a transition into a next shot if this was for an ad. Now, I'm going to show you this subject reference option. So, we have a complete tutorial of this platform. This is not yet available with their new model, but it still works pretty good. But what you do is add a reference character and then your prompt, so your story could have a consistent character. Really easy to do. All you do is go to the image tab here. Type in a prompt. I'm going to use this one as an example. So, this is the prompt that I use for this one. And all I have to do is download this image. I'm going to download it without a watermark to my computer. And then I'll go to the video tab and I'll use the subject reference here. I'm going to upload that character here. So, I'm going to press confirm. So, that's going to be our reference character. Now, typically it might be better if you do it as a close-up shot. I just had this kind of medium shot of this person. But if you kind of create their face, it will also be able to do it that way. But this does zoom in and takes care of it that way. And then I'm going to type in a few different prompts here one at a time. So this is going to be my establishing shot, white shot, and eventually dolly in towards the face. So I'm going to go ahead and generate that. And then I have chat GPT just create an entire scene. So there's different shots and different scenes actually with different prompts with the same consistent character. And I'll just generate a few different clips to show you what that looks like. Now when it comes to price, this is the lowest cost in the top tier models, right? So if you compare it to Veo 3, very expensive or cling Hilu, this O2 model actually offers the lowest per second or per video cost in this class. Now if you want to try it for yourself, I have a link in the description below this video. And if you want to really dive into video creation with AI, Kevin, my partner from Futureedia, he actually created a course that we recently launched on our platform that shows you everything there is to know about generating high quality videos with AI from text prompts, from image prompts, really everything there is to know about it. It's an hour crash course, but there

is a lot there that you could use for yourself. I'll put a link in the description. You could try it for free as well. Thanks for watching this video and I'll see you on the next