So, in today's online tutorial, I'm actually going to show you guys how you can really get the best out of VO3, even if you are a complete beginner. I've seen tons of online tutorials, and most of them don't cover how you can get the most out of this piece of software. So, of course, in order to access VO3, you do want to make sure that you are in Google Gemini. Now, usually, if you hover over the button, it will say either V3 or V2, and it also will tell you how many more videos you can generate. Depending on your current subscription, it will let you know how many you can generate. But if you are on the standard $20 a month subscription, you do get access to V3, but you do get limited access. So, just make sure you watch this video until the end because I don't want you wasting your V3 allocations because you don't get that many generations unless you're on the Google AI Ultra platform. So now it's pretty easy to generate these prompts, but let me show you guys a video that I created using video 3, and then I'm going to show you guys all of the different prompting strategies that I use that are specific to VO3 that even as a beginner will help you. Package for Mrs. Lee. Please sign. So hopefully you have seen the FedEx ad that I've just made. What I'm going to show you now is seven key things that I consider when prompting in VO3 that allows me to achieve a much higher quality of video. I've tested this for several hours. I've got the Mac subscription and these are the seven points that matter the most. So, one of the things is of course the character. So, in the initial prompt, this is one that, you know, I don't want to say it was a hallucination, but it didn't go that well, is that the character design actually failed. And one of the things, the entire prompt for this was simply an F-35 jet screeches to a halt in a quiet culde-sac, canopy pops, pilot hands neighbor small package, and the pilot says, "Sign for Mrs. Lee." I thought this prompt was enough, but you're about to see that this initial prompt right here, it actually doesn't really capture everything. And you can see the prompt fails. And I'll show you guys how I managed to fix that. So there we can see the character getting out. And it's really funny because the person that is actually handling the package actually looks like the person that's supposed to be receiving the package. So, how I would fix this is that when you're prompting VO3, make sure that you include a full description of the character. Right here, you can see I have the prompt that says a man in full military pilot outfit complete with oxygen mask, tactical gear, and a flight helmet. So, I decided to go all out because I wanted to ensure that when he was coming out of the jet that he would still have his entire outfit on. If you aren't specific enough with your prompts, the problem with Google V3 is that it will just heavily assume something about the character you are trying to design. So, I would say that if you are trying to design a character, please make sure that you describe the main character completely. What is their age? What is their gender? What is their race? And what is their outfit style? Those are probably the four key factors for me that work the best when looking to design a character. Of course, you could always use an AI tool to help you do this. Maybe create an image in Google's image effects or in midjourney and then of course have chat GBT describe that character to you in terms of their age, gender so that you can actually get the video done because VO3 doesn't currently support in Gemini natively image to video. But what we can do is we can just completely describe the character in a universal kind of outfit. That way we can achieve some kind of consistency. And so with this one, I think it's really important that you know you design the character because as I was saying, Google will essentially just design your character if you don't. So think about all of those factors I just mentioned. And of course, make sure you design both characters as well because if you don't design both characters that are going to be in the scene, Google will just heavily imply what those characters are going to be wearing or what they're going to be doing. Now, the second one here, before I show you guys this clip, let me just show you the main prompt. And the second thing you need to focus on, even if you're a beginner, is the kind of action that's happening in the clip. So, essentially, when I created this, I wanted to have a F-35 fighter jet landing and screeching to a halt in a quiet culdeac. The canopy pops, the pilot throws a parcel from the cockpit. Now, I thought that this setting the scene would have been enough, but you'll see that VO3 is basically something that does exactly what you say and sometimes doesn't infer the best thing. So, if there's something that's in your mind, you need to be super specific with as to how you prompt it. Because when you see the video right here, the jet screeches to a halt and it does land. Okay, it does land in a culde-sac. And then, of course, the canopy pops, which is the second part. And then the pilot does throw a parcel from the cockpit, but that is about it. One of the things I forgot to mention was that the parcel should land on a doorstep. Someone should be there. And then even weirdly, we do get this strange hallucination from V3. The reason that we get this hallucination from V3 and I would say this is something you need to be very careful of is because hallucinations can occur. Now, the hallucinations aren't that big, but in some scenarios like this, we can see that the driver just disappears. This just snaps shut. These hallucinations will occur if you aren't super specific. So, you ideally don't want to prompt them too long, but if you're going to be using VO3, make sure that whatever clip you are doing, you at least have a decent amount of control before you write out the scene because V3 credits are relatively expensive. And I don't want you guys to make the same mistake that I did, which is why I'm making this video. So, make sure that whatever action you have in the scene, ensure that it is the right description. If it's going to be landing, if it's going to be moving forwards, backwards, make sure that that entire description is good. Otherwise, you will get some hallucinations if it isn't described naturally. Another thing that you want to make sure you're doing as well is, of course, make sure that you describe the environment. Another thing you want to make sure you're doing as well is, of course, design the environment. So, right here, you can see I've got the environment is a sunny setting neighborhood with manicured lawns and large beige two-story houses. If you're in V3 and you're trying to generate any kind of consistency, it makes so much sense to be able to accurately get a description of where you're designing things. So, just make sure whatever you do is completely consistent. And that way, if you're trying to create some kind of consistency, having that design where it's like somewhat consistent with the background, it's going to help you guys create whatever it is that you do want to create. Of course, sometimes we don't have the luxury of image to video. Of course, if you're using flow, it's a completely different story. But in this scenario, having a really specific background design was something that was super useful for me that allowed me to essentially achieve more so consistency when dealing with the randomness of VO3. So, this essentially was another angle that I didn't use in practice, but it was good because I was able to generate multiple different scene angles because I had my location. So you basically want to make sure you describe the location, the sunlight, the things in the background, indoor, is it outdoor, are there shadows? So I would just basically think of simple things like the time of day, the place, is it rural, urban, is it like in a city? And then of course the lighting, because I think the lighting impacts things the most. The next big thing, which is point number four, is that if you want to have vocals, make sure you have the character say what it is in these quotation marks, or else it can be confusing for the model. So, ensure that you have these quotation marks. So, for this one, I said, "Package for Mrs. Lee. Please sign." And you can see right here that it said, "Package for Mrs. Lee. Please sign." And so, one of the things that I want here for you guys to understand is that please make sure as well when you're doing your vocals, you don't have any typos because it's not like an LLM that can infer what you're trying to say. Those typos will mess up. So, double and triple check before you enter the prompt. I know this seems like something super basic, but there were a few times that I didn't check for certain typos and it did result in some pretty hilarious hallucinations. I will show you that one right now. So, for example, for this one, what I was trying to do was add some emotion. With a V3 characters, you can add emotion in those characters. So, you can say the person is shouting, they are screaming, the person is upset, and it will actually do this. The only reason it didn't work in my specific scenario was because I generated something and one of the key features is this mask that's on the character's face. So, of course, it's quite hard to get that emotion out of them. But like I said before, remember that you need to double check your spelling because you can see right here that I wrote this really quickly and I actually wrote why will nobody sign man package and not my package. So, this actually resulted in a hilarious scene that I couldn't use. Why will nobody sign man package? And so from that example, you can see why it's really important that you double check your spelling and ensure that you know if you do want a character to have a specific emotion or emphasis on how they're saying something, it's important to vocalize that right before they say it. So with this one right here, it did kind of hallucinate a little bit because it didn't say everything at once. But with V3, there are still some hallucinations that are currently getting worked out. So please bear in mind everything is not perfect. A lot of the you know videos and final footage that you see online, you are often burning through some credits to get there. So that is an important thing when using this tool. Another thing that you need to understand about V3 as well is that you can generate multiple different styles. So I found this on Twitter and the prompt was 1990s VHS footage of a music television feature about a reptilian rockstar from Alpha Centuri. And essentially it will initially apply this, you know, low VHS quality filter that basically makes your footage look from that time zone which is super super interesting. And I think that if you're trying to achieve a very specific look, do not forget to add that to your prompt because VO is very good at doing that style of footage. So this was something for me when I found this out. I was like, "Okay, that's pretty cool. You can get really creative with this." And it means you don't need to apply any post-processing. you can just simply do it all in one prompt.