How to Create Pro-Quality AI Videos With Gemini Omni

Before we jump in today, I've got some real big news. We just announced our first speakers for AI Business World 2027, the 2-day AI conference running inside of Social Media Marketing World dedicated entirely to helping marketers implement AI. Nearly 2 dozen speakers were announced so far across both conferences with dedicated AI sessions designed to take you from experimenting to actually implementing. Every speaker I personally recruited. Every session is pitch-free. Go see who's speaking at AI Business World dot live. That's AI Business World dot live.

Welcome to the AI Explored Podcast, helping you put AI to work. And now, here's your host, Michael Stelzner.

Hello, hello, hello! Thank you so much for joining me for the AI Explored Podcast, brought to you by Social Media. Examiner. I'm your host, Michael Stelzner. This is the podcast for marketers, creators, and business owners who want to know how to put AI to work. Today we'll explore creating professional AI videos with Gemini Omni. My special guest is an AI video strategist who helps small businesses create pro-quality video content. She runs the Video Spark membership where she provides ongoing training and support. Her tool is Spark Sheet 360, which includes Character DNA, Environment DNA, and Storybirds. Eve Whitaker, welcome back to the show. How you doing today?

I'm doing great. Thanks for having me. Love being here.

Super excited to have you here. We're gonna talk about something really exciting today, which is AI video. But first I wanna ask, what do you see as some of the misconceptions out there or the biggest misconception when it comes to AI video?

I think currently the biggest misconception is people think it's hard to prompt and AI video generators have really gotten great at understanding natural language. So you no longer have to use AI to reformat a prompt in JSON or any kind of coded verbiage. You can speak to it like it's your assistant, and it really does a great job of understanding.

Well, and that's important for some people who have never done any AI video at all. They may not even realize that. But it's fascinating 'cause I just had another guest on the show and I don't know which order this is gonna fall in. And we were talking about Midjourney and on Midjourney you had to know how to do all the things in order to get the output that you want. And now you can just talk to it and that's really exciting. All right, well, we're gonna talk about Gemini Omni. So when video is done well with Gemini Omni, and we'll get into what the heck Gemini Omni is in a minute, but when people do it well and they pay very close attention to what we're gonna talk about today, what is waiting for them on the other side? What are the benefits?

There are so many with this tool. First, it allows if you don't like being on camera, you can now create an amazing avatar of yourself in 10-second clips. So it does require a little editing, but you never have to be on camera again if you don't want to. You can put yourself in any environment. You can put yourself on a rooftop, on a mountain, on a beach, and it does a great job. It also allows you to edit video. And let me explain that because it is very different than any tool out there right now, where I can take video that I shot with my phone or that I generated with AI, and Omni can take objects out, can change outfits, can change me into a cartoon.

Wow.

And it does it with the most simple, basic prompts, which is incredible. And the use that I tell my students the most, which I'm using, is it allows you to create those wonderful things in the marketing world called hooks. I can now grab people's attention using my avatar doing really silly things that used to take a crew or a lot of time for me to go shoot. And then I can marry those, meaning I can bump them or edit them together with video of me talking. But I've grabbed your attention by doing something silly or eye-catching that people stop the scroll because they want to know why I just dumped Nickelodeon slime on my head. And now I've got your attention.

Very, very interesting. And we're going to get into a lot of this in detail. So what I'm hearing you say is that with this tool, Gemini Omni, which I will explain to everybody, Gemini is from the Google ecosystem, folks. So that means it's been built on their world models, which means it's been trained on all the content on YouTube. So that's kind of a big library if you think about it, right? They understand Everything you can possibly imagine, they've got it on YouTube. It's been trained on all these different environments and applications and people. That really is kind of the big upside, is it not?

100%. It understands movement, it understands environment, it understands people talking inflections like no other model does today, hands down.

When we were chatting, you said, ask you about Starbucks. There was some sort of story. Do you remember the story? Can you share that?

Yeah. So You can use Google's Omni using the Gemini app on your phone. And I was doing a presentation for a group of real estate agents, and I wanted to show them how simple it was to use it. And so while I was in line at Starbucks waiting for my flat white, I literally got on my phone, I called in my avatar, and I made a really quick hook. Just saying, "Hey, put me on a buffalo riding in front of this house," and I took a picture of a house from Zillow and say, "Hey, we have an open house this weekend. I'll be here; the buffalo won't." And within the time it took me to get my coffee, pay for it, it had rendered out and it was done. And I did that to show them this is what is such a game changer. Not only can I do it on my desktop and do it very sophisticated, I can. I can be getting my hair done, I can be in line at Starbucks, and I can be creating content, and it's so easy.

That's really cool. Let's start with what the heck Gemini Omni is. And you kind of mentioned a little bit of it, but let's just kind of lay out what it is because some people might be familiar with the fact that there was this thing called Veo, right? So what is Omni? Like, explain what it is and let's break it down. Sure.

So Veo 3 was the first AI video generator to come from Google, and then they created Omni. And Omni to me is more of the people's generator. So, and I think most people are using it now more than Veo. And so it lives in a few places. I don't want to confuse anyone, so I'll keep it simple. It's OmniFlash is what it's called, and it exists on your desktop and on your phone using Gemini, the Gemini app. And then you can also use it if you have one of the many aggregators that uses all the different models. It's, it's available in all those, and then it's also in Google Labs. And basically it is just AI video generator that does more than any other AI video generator today. Now, you know, tomorrow that could change, but today its capabilities are beyond any other one that there is out there.

Okay, so let's dig in just a little bit more. So if I have like a paid Google account, like my company, we have a Workspace account, I just go to gemini.google.com. And I can go ahead and I can bring Omni in. How do I actually do that? Just out of curiosity, do I have to put a slash mark or something like that? Or—

So just like where you would go in your chat, you hit the little plus button and where when let's say Nano Banana first came out, you would hit image. Now there's a thing that says video.

Oh, I see. So if we select video, it's gonna be using Gemini Omni behind the scenes is what you're saying.

Exactly.

Now let's talk about this Labs thing. I think, and you correct me if I'm wrong, that's like a separate interface that allows you to do more sophisticated stuff. Is that true? If so, let's elaborate on that a little bit more.

Sure, it does. So if you want to really play with all the benefits and what it can do as far as correcting video, changing video, stuff like that, you're going to want to do it inside Google Labs. And it's labs.google. And it is almost like their own storefront of many tools. There is a lot under the hood. And OmniFlash is one of them along with Veo. So it's in the same world, you just have to choose the So you mentioned that there's a limit.

Explain a little bit about the duration limit and kind of the cost also.

Sure. So if you have your paid Google account or you pay for Gemini, which some people do separately as their LLM, you automatically get to use it. But there is a certain limit, and I don't know what that is today, but when I first started using it, you could generate probably 5 or 6 videos a day before it would say you've done too many, and it reset in a couple hours. So it wasn't like it just stopped you that day. It would literally stop you within an hour or 2 hours of using it and say, you can come back and do this, which a lot of people found frustrating when they didn't realize that it's almost like they're enticing you to play with it using the app or desktop. But once you see the benefits, they want you to go over to the Google Labs and use it there because that's where all the bells and whistles live.

What are we talking about as far as the cost if you get beyond The limits, so to speak.

So they've actually really made it a lot more affordable where you can do it in different tiers. So I'm on the highest ultra tier. I think I pay $200 a month and I use it all the time. And that gives you access to all their tools. But I think they have one as low as $20 a month. And again, it's— and I always tell people, start at the lowest because you don't know how much you're going to use it and you can always buy more credits. So that's what I would recommend. If you're really interested in learning and doing more with AI video, I also suggest using a third-party aggregator like OpenArt or Higgsfield. Omni lives in there, but it has a lot of other tools. And a lot of times you go to those platforms and you can get a great deal and you begin to play with the platforms or the models, and then you're also getting access to everything that's coming out at once. So it just depends on where you're at. The more professional people are the people that are really creating with AI video. They're always going to have an aggregator in their arsenal and then they They take whatever their favorite ones are, like mine is Omni. So I also have a subscription to the Google Labs.

Okay, perfect. So now that we got the technical stuff out of the way, let's get into the fun stuff. You already mentioned that you could create an avatar of yourself. So let's kind of dig in on this a little bit. First of all, let's talk about the old way that most people had to do it. And then let's talk about the new way and then how to actually do it.

Okay, sure. So let me just make a distinction between an avatar and a clone.

Okay.

Because a lot of people use and they think about, hey, Agen, and that is how you generate your AI clone. And to me, a clone is where you are going to be able to feed it a very long script and it will be that talking head you that does it. That is not what Google Omni does. It creates an avatar, or what I like to call your AI twin, and it is con— it is under the constraint of only being able to produce 10-second clips. Now that might change where it will be able to render out longer ones, but right now it will only take 10 seconds of you. And so you're really having to think about if I want to do a minute video, I'm going to really play around with the different settings of where I want to be and play around with that and put myself in different places to create it. But if you remember Sora, a lot of people created their AI twin in Sora, their avatar in Sora. It is the exact same process. You do it on your phone. Once you've made it on your phone, you can use it on any of the platforms. It will call it in. And so basically you get on your Gemini app on the phone and you hit the little plus button and you scroll down until it says avatar, and then it will walk you through almost like when you create it, like a Face ID when you first get a phone or something and it has you look left, look right, look up, look down, and then it even puts words on the screen it wants you to say because it records your voice. And then once you make it, you name it, and then you never have to use it on your phone again if you don't want to. I can be in Google Labs and I can call in my avatar just by hitting the and sign or the at sign.

Yeah, this is really fascinating. Okay, so what I heard you say is you got to download the Google Gemini app on your phone, and then you look for the avatar, and then it will train you and it'll tell you to talk and move your head and all that stuff so it understands what your profile and all that stuff looks like. Now, there are some people that want me to ask you this. I'm going to ask you this. Does this mean that my identity now is up in the Google Cloud and other people can access me? What's your thoughts on that?

It probably is up in the Google Cloud, but right now they cannot access you.

Okay. You're not feeding the machine is what I'm trying to say, right?

Well, you could be to an extent, right? Anytime— I mean, just putting us on this podcast on YouTube is feeding the machine. So it's kind of like, But with Sora, you could actually share your avatar and other people like big name people, Hollywood people would share it. And then I could make a little video with Gary Vee in it, right? You cannot do that in Omni right now. The avatar is only for you. Only you can call it up because a lot of people ask that because they want to do one with their client or something. And I'm like, well, then you have to log into their account. Like that's how you would have to do it. They would have to create it and let you log into their account.

Can you create more than one avatar? Just Out of curiosity.

You can, but not— let me be specific about that. You can't have 2 avatars at the same time right now, but if you create one and you don't like it, you can redo it and start over. Because to create a really good avatar, you want to be in really solid light. Uh, I say not a lot of shadows, almost like if you were taking a photograph of yourself, you want to be in the best light so you can capture everything. And you want to make sure you're in an environment that has There's not a lot of background sound, so it really picks up your voice because if it doesn't get something, the AI is going to make it up. So if I don't show and don't turn and don't do everything that, that it asks me to do, or if the video isn't good or you have a spot on your camera, then the AI isn't going to tell you like redo it. It's just going to take it and then you might not look exactly like you or you might not sound exactly like you. But if you do it right, It's the best, in my opinion, for making an AI twin looking and sounding like me.

Let's talk a little bit about this because is this background setting— does it matter? Does it strip you out? So for example, would it be better to be up against like a white wall or like in your case, you've got a bunch of furniture behind you? Let's start with that question. And then let's also talk about what kind of light, you know, because as you know, you have a background in television. There's all sorts of kind— there's natural light, there's all sorts of light. So what are your thoughts on this?

I would say natural light. Stand in front of a window or to the side of a window. You don't want to be, uh, you don't want to be filming with a window behind you, right? Because that's going to create a silhouette. It's not that sophisticated. I did mine in my living room, and if I prompt it and I don't give it enough— and we'll get into prompting— but if I don't give it enough information, it will pull from what's in my background. But But this is where Omni's awesome, because if I say, call my avatar in and put me on a beach in Mexico walking, it does that too.

So it knows to knock out the background, basically.

It just knocks it out. It's incredible.

Okay. What about clothing?

Whatever you're wearing is what you will be wearing unless you prompt it otherwise. So I wore a red shirt very similar to this. I had this necklace on. If I don't prompt it to be, if I don't say, put me in a pink suit or put me in this, I will be wearing a version of that shirt. Sometimes it makes it a dress because I didn't do my full body. And I will always have this necklace on. Now, if I prompt it to take it off—

What about— you're wearing a cowboy hat for those that don't see it. What about hats and glasses and earrings and stuff?

Yeah, if you are wearing hats or you are wearing glasses, it will do that. And that's a good thing because— or that's a good question because I can prompt it to take off my hat, but it doesn't have any data on what it looks like Under your hat. Under my hat. So AI will make it up. So I would say don't wear the hat. And if you want to prompt it with a hat and you want to be— have a specific hat, just upload a picture of the specific hat to put on your avatar that you recorded without the hat, and it will do a fantastic job of doing that.

Fascinating. What about the audio side of it? Because you already mentioned, like, if you're in an echoey room, that's not going to be good. You know, I'm a podcaster. You have a professional podcasting mic. Obviously, it's not going to catch that if you're actually just using your phone. Any tips on how we should speak? Because anybody that's got any background in social media knows there's different ways that you speak. There's your natural way of speaking. There's your animated way of speaking. You know, there's all that stuff. Do you have any— what do you recommend?

So I think they do this on purpose is what they have you read makes absolutely zero sense. Like what they had me read was like the octopus walked straight up the mountain, turned around, Yellow is actually blue. So it made it so I just spoke naturally.

Okay.

Because that's how you want to speak. If you're not speaking naturally, if you're animated, it's going to get you animated. And it's harder to prompt Omni to have you not be animated if your avatar is always animated versus if I just speak in my natural voice. But in the video that I'm making with my AI twin avatar, if I want to be yelling or I want to be animated, I would just put that in the prompt and then it would just do it.

Hey, I want to tell you something that's coming that I think you're going to be really excited about. We just announced our first speakers for AI Business World 2027. That's our 2-day AI conference running inside of Social Media Marketing World, focused entirely on helping marketers and business owners put AI to work. Every speaker I personally recruited because they're getting real results with AI. Right now, and they're going to show you exactly how they're doing it. Quote, the AI speakers were amazing. Every bit of information they shared was completely new and valuable to me, said Sarah Heinbrotner. The beginning lineup is now available at aibusinessworld.live. Tickets are on sale at discounted prices. And here's something worth knowing. If you grab a virtual ticket, you get access to the content from AI Business World and Social Media Marketing World. 2 conferences, 1 ticket. Go check the speakers out at aibusinessworld.live. That's aibusinessworld.live. Okay, so we've spent a lot of energy here specifically talking about how to actually create an avatar. Let's talk about how in the world to use the avatar. So you've already mentioned that you can tag it somehow. So let's elaborate a little bit more on What in the world we do once we have an avatar. And by the way, how long does it take? Like 60 seconds, 2 minutes? What are we talking about here?

Yeah, I'm talking 5 minutes max from the time you decide you're going to do it to where it's done. 5 minutes. I will give you a guide that you can put in the link just because listening sometimes hard to grab it, but it's not hard, but I'll give it so people, it's a step-by-step, whether you're on phone or whether you're on Android or on iPhone, and it will walk you through it and it walks you through how to redo it if, if you don't like the way it turned out.

Okay. So let's talk about how we use it now that we have it.

So the use cases for me the most are I don't like filming myself on camera. So I know as a social media content creator that a lot of times we film ourselves and there's like lots of different methods, right? But one that's really popular that I think really gets people's attention is if I'm like walking up and I say something and then I turn the corner and I say something and I'm getting my message out, but I'm filming myself in all these different locations, you know, looking in bucket, doing whatever, just to kind of change the normal just talking head. And I've tried it and it works. I get more engagement, but it's exhausting. It's exhausting doing that. So now I can just literally prompt my avatar to do it. And what's fun about that too is that's really where you can get that scroll-stopping stuff in there. For example, I told you that I did one where I dropped the Nickelodeon green goo And then was just talking and being like, I don't know why, did I get your attention? I did one where I'm standing in like an evening dress and I'm like ready to speak, and all of a sudden someone comes and just shoves cake in my face. And literally I'm like, did that get your attention? And you don't even have to be that crazy. I mean, I've done ones that like I'm in a park with a bunch of dogs and I just have tennis balls falling from the sky like rain. You know, just, you, you can really think of Anything that might stop and get their attention. And just like you would do it in like where you're filming yourself in different locations, same thing, but you just prompt it.

Okay, cool. I want to get into the hook stuff in just a few minutes, but I want to back up just a little bit because not everybody understands how to actually creatively apply their AI version of themselves or avatar into an environment. So let's talk about kind of the basics that you teach, you know what I mean? Like there's action and all this other kind of stuff. So give us some of the basics for anybody who's never done anything like this before?

The most simple formula I say is your subject, whatever's happening, the action. So what are they doing? What's the environment? Where are they at? What are we doing? And what is the camera doing? So if you kind of think of it like subject, some people say setting versus environment, but you just follow those 4 things. So a prompt that I would do If you're going to use your avatar, you call it in by using the and sign, right? Or the at sign. I keep saying and. At sign. And you just hit it and it just brings it up. And so I just say— so mine's called Eve Whitaker. So it'd be @EveWhitaker. She is dancing in the street. It is raining tennis balls. A bunch of dogs are all walking around her. And she walks up to the camera and says— Like whatever you picture, and the most simple way to do it is try it. Try the prompt, and if it doesn't do exactly what you want it to do, a lot of times you can adjust by just looking at what it gave you, what output it gave you, and you're like, oh, I I didn't realize that I just I didn't tell her to walk up to the camera. I pictured her being closer to the camera. And so then you would just say it again and just now say, okay, she walks up to the camera, or she's standing close to the camera. So it's almost like if you took a photograph of yourself and you look at the environment and everything around what you want to have happen, and then you just tell Omni what you want to do.

Okay, so we mentioned action, right? Which is what is the subject doing? In this case, the subject is us because we're the avatar. We talked about the environment. In your case, you said it was in the street. And then you also said the camera. So this is something a lot of people don't understand, like how to think of where to place the camera and stuff. Do you have any thoughts or tips on that side of it?

I would try to go simple at first. So what I try to explain to people is there are different levels of, uh, where the camera is going to take a photo of you. And in very simplistic terms, it's like an extreme wide shot, and that you're going to see the entire environment And your subject, your avatar or whatever. And that's usually what we use in the film world and TV room to establish what's happening. The closer you get to the subject, the more intimate it gets. So if I want someone to hear my message, I'm not going to have it be an extreme wide shot. So think of it in just how close the camera is to the subject. And that includes like if I want the camera or my subject to look larger than life, Then they call that the hero shot. Now, you're not going to know that, but you could say, I want them to look larger than life. If you want them to look smaller, you would have the camera up a little higher. Most women know that, right? When we get our photograph taken, we're always like, up, up, up. Because if you shoot me from up high, I look a little, a little thinner and a little taller versus if you're filming me from down low, not my best look. So what I tell people to do is start watching television. And when you're watching television, if something stops your brain, right? Like there's something that you're feeling, an emotion, just kind of take a screen grab in your head. What was happening with the camera? When you begin to look at that stuff, you don't unsee it. And you begin to get ideas because there's— directors do different things, creators do different things to get you to feel an emotion. If I want you to feel uncomfortable, I'm going to get very close to you, Michael. So I would be up very close, and then you're paying attention because there's something going on that doesn't feel right. So that's kind of just a trick. If I want extreme close-up, why am I looking so— or why is she so close? So that's kind of the camera stuff. It's one of those things that I tell people, just watch TV. And or a movie. And when you see that, what's going on? Why do they go from wide to medium to close-up?

Let's talk about voice and script. You know, sometimes when you're up close, you could whisper, you know what I mean? Or you could do something different with your voice. So let's talk about any kind of vocal variety stuff you might recommend. And then also scripting, because I would imagine if you don't give it a script, it's just going to make up a script, right? It is.

So it's really going to do— in fact, if you give it a very simple prompt, sometimes it literally makes stuff up. And you're like, oh, that's interesting. And sometimes it's a happy accident where I like it. So yeah, I would definitely give it a script. And again, if you're like, well, Eve, I don't know how to do a script, it's in your prompt. And I say, let AI help you. So I would go into ChatGPT or Claude or Gemini, whatever LLM you like, and I would say, hey, this is what I'm doing. I want to create— And you give it the length. Now, right now, Omni can only do 10 seconds. So if I want to make a 30-second commercial, I will say to ChatGPT, I want to make a 30-second commercial in Omni, and it can only generate 10-second clips. Help me break down the script. This is what I want to see happen. And I just talk. I use Whisper Flow, and I just like vomit my ideas. You know, that's how unsophisticated my prompts are. Like, if I show them to people, it's like, You hear my likes, my ums, my, you know, do it kind of like this, you know, yada, yada, yada. That's literally in my prompts. And then I just say to ChatGPT or Claude or whatever I'm using, ask me whatever clarifying questions you have to help me get the best outcome. And then it will ask me like, what is this for? Is this for education? Is this for promotion? Is this for blah, blah, blah? What is your niche? And it asks me its questions and then I just brainstorm with it. And then at the end of the day it will say, okay, Here's your— here's the idea. Do you like it? Yes. And I say, okay, now turn it into a prompt for Omni. Every LLM now knows how to create a prompt for all the major models.

I love it. Okay, perfect. Thanks for going down that rabbit trail with me. Let's get back to the hooks. So a lot of us are creating short-form video across TikTok, Instagram, YouTube Shorts, all the things, right? And the hook is the very important— most important part, right? The first couple seconds is what stops the scroll and gets them to watch the video. So what I heard you So what you say earlier is, hey, you can record your yapping video or direct-to-camera video the way you would normally do it, but maybe let Omni create this incredible hook to kind of get someone to watch the thing in the first place. So let's talk about some tips on how to actually create a hook for a video that you've already kind of created, which is not going to be an AI video. And presumably you're going to sew this together in CapCut or, or Instagram's edits app or something along those lines.

Exactly. So if I wanted to marry Hooks to a talking video, let's say I shot my talking video in this. Even if I didn't create my avatar in this outfit, I would take a screen grab and I would put it in and I would say, this is what I want my avatar to be wearing. So it's not jarring that I have an avatar and then the talking, right? So you wanna make sure you're dressed in the same and you just do that by giving it a screen grab.

Okay. So your existing avatar will match whatever you're wearing in the video. That's what I'm hearing you say.

Okay.

Okay.

Yep, exactly. 100%. And then again, I would go to ChatGPT, Claude, or whatever LLM you like to use, and I would say, hey, this is what I'm doing. Help me come up with a bunch of creative hooks. Because we all know that it is tough. I mean, if it was easy to come up with hooks, we would all be using hooks all the time and everybody would be grabbing everyone's attention in the first 3 seconds. So So for example, when I was doing the thing for the realtors, I'm not a real estate agent, so I just asked ChatGPT, what are some ideas of hooks that I could give them to sell the house? It came up with the one of me riding a cow. I changed it to a buffalo. I'm from Montana. I'd just been there. Um, it gave me this great idea of making me a little 6-inch avatar, and I'm walking on a countertop and I say, the only thing small about this kitchen is me. Wait till you see this place in person. It did one. It gave me the idea of having my avatar parachute in onto a lawn with a for sale sign, and again, I could give it the actual photo of the house and tell it to use that reference. And I came in for sale sign the house, and I could say there's an open house here this weekend. Click the link below, and I'll DM you the you know the info.

So.

I try to just use AI to help me with those hooks. I've just seen a lot of people give so many trainings on different ways to create hooks, and they are great, like when you're filming yourself and like the different angles and that pattern disruption and, you know, all that stuff besides just the verbal ones. But where Omni comes in and really helps out is it takes the filming out of the equation. And you can do it. So I've even tested— I told you on when we were talking earlier that I took the same video and I tested 7 different hooks to see what people like, to see which one performed best.

Oh, that kind of trials reel or something like that?

Exactly.

Yeah.

And then that's the one I put the money on.

Okay, so I have a question for you. Um, can it do obviously vertical video? I'm assuming it can do vertical and horizontal. And then also, what about the resolution? Like some people film in 4K HD. I mean, I'm getting a little technical here, but thoughts on like the resolution and the dimensions? Because obviously if you're doing like Shorts, you're generally doing vertical video. Is that something Omni can accommodate?

Yep, it absolutely can. So you just tell it what you want it, whether you want it 16:9, 9:16, uh, resolution. I believe right now it goes up to 1080. I know it does 1080. I'm sure it will go up to 4K at some point just because that's where they all eventually go. And it allows you to choose the length The resolution and the size. And the higher the resolution and the longer the length of the video, the more credits you're gonna use. So a lot of times I will test it out. If I'm just putting it on social media, I stay in the 720 range because there's no need to go any higher. If I'm making a video that's gonna be on TV or that I'm gonna be presenting at a conference on a big screen, I'll go up to the 1080 and the 4K.

And folks, if you listen to the prior episode that we just recorded, with Ross Simons, you can upscale these videos. So there are tools out there where you can take your video, like your 720 video, and you can upscale it to double it and double it again so you can get that higher resolution video if you want. And my understanding is those upscalers are pretty good. How easy is it? Like, do you usually find that the first draft is good, or are you doing multiple iterations when you're making, you know, these kinds of hooks? Or, you know, because like some people that have had experience with AI video in the past are like, oh my gosh, it's not, it's not Like, what's your thoughts on this?

This is where the prompting does come in because again, the least amount of information or the less information you give it, the more it is going to do what it wants. So one of the most common mistakes is people give it more dialogue than 10 seconds will allow. Now, AI is not gonna stop you talking mid-sentence. What it does is it tries to cram all that dialogue into those 10 seconds. So you usually get like jumbly garbage towards the end that makes no sense. So I think that's really important. And, you know, I hate saying play with it, but when you begin to make these little mistakes, then you begin to learn, okay, I went too long, right? Because AI video is just not at the place where it is going to do exactly what you want on the first try. Sometimes it does and it's like, yes, Most of the time you are iterating on it. And that is one thing that Omni does that I mentioned earlier. But this is a good thing to bring back, is if it gives me something that I don't like or I just don't like a part of it, you can go back in and say, keep everything else exactly the same, but she needs to be wearing a red shirt, not the blue shirt.

Oh.

And it will redo it.

Okay, that's really cool. So that's like the big advantage is that instead of having to start from scratch, you can specifically say, hey, just— and ChatGPT does a great job with this with their image generator as well. You can be very precise and just change that one little thing. Now, you also mentioned this Captain Obvious thing when we were prepping. Can you explain what the heck that is?

It's one of my favorite tips to give prompts. So whenever I'm using ChatGPT or any of the LLMs to give me ideas, It will go to that Captain Obvious. And that's like a term we used to use in television when the producer would pitch us something and we're like, that's Captain Obvious. Like, if you can Google it, then everyone can, right? So I began to say that to the LLMs when I was looking for interesting hooks or I wanted to help get an idea for an ad or a commercial, because it will give us the Captain Obvious. If I'm trying to do something that talks about AI overwhelm, Hands down, 9 times out of 10, it gives me the outline of a woman or a man sitting at a computer with lots of tabs up. And I'll say, okay, that's Captain Obvious. Let's think outside the box. Let's use metaphors or analogies. That's when you begin to get the gold. That's when it begins to say, oh, okay, let's try again to something that the first person isn't going to decide.

Do you find that the newest, greatest models are the best creatively, or do you recommend experimenting with them? Because as you know, there's always new versions of ChatGPT and Claude and Gemini, and some seem to be more creative than others. What's your thoughts?

You know, it's like a race. And at one point ChatGPT is in the lead and then pretty soon it's Claude.

I'm talking about the same version, though. Let's say you're a ChatGPT person. There's a bazillion options to choose from. Like, do you find that the thinking one is more creative than the non-thinking? Do you understand where I'm going with this?

Oh, I see what you're saying. Yeah, yeah, I do. So it's like I've learned, like, in image generation, if I'm using image generation and it's giving me any kind of, um, issues, go to the most basic, Instant, like just the basic one. If I'm using Claude and I'm really trying to do something fantastic, I will use Fable. Like everyone was asking me, like, when do you choose between Fable and Opus? Um, it's again, it's you're playing because These things are updating at lightning speed. They're changing every day. And so if I'm worried about how many credits I'm using, I'll go down to a lesser version. If I'm working on something for a client, I'll bring it up. You know, one of the things that I laugh about is when people say how expensive all this stuff is. And I've said, and I said this today, I am spending more a month paying off my student loans than I'm paying on subscriptions right now. If that gives you any idea of what AI is giving and putting at our doorstep, you know, laying these tools at our feet, it's a lot less to play with the AI and learn than it was for me to go to school and get a master's degree.

Amen to that. And I'm go— we're gonna talk about how to create longer videos here in a minute. And just a quick tip, folks, I'm a big Claude user if you've been a listener for a long time. And I find that Fable 5 Low is way more creative with like hooks and titles and stuff like that. It's the most sophisticated model. And sometimes I'll use, you know, the old model, my favorite model, right? Which is Opus 4.6. And I'll open up another tab, another thread, and I'll do the exact same thing, but I'll choose Table 5 and I'll be like, oh, that's what I should have done, right? Because when you're doing something as important, as important as a hook, you want the most creative idea possible, right? So that's sometimes where the latest model is worth an experiment. So I always like experimenting. Okay, so let's get back to the longer video stuff. Let's say we want to create a 5-minute video or a 90-second video, and this is where it can get a little complicated, obviously. So now we're talking about the scene potentially carrying beyond the 10 seconds. So let's talk about like how all that stuff can work together.

Yeah. So when you're using Omni, like I said, you have 10 seconds. So what you need to do is pre-production, plan it out and know, looking at your script, what's going to happen, what is the last image in one of your 10-second clips, and what is it going to cut into in the next 10-second clip? So if I'm having a conversation with you and I, and I'm going to be cutting back and forth, that's an easy one because I can just cut back to me nodding my head as you're saying something, right? But if it's more like a linear story, you need to think about if it's just me, You don't want to cut to me and then all of a sudden cut to me somewhere else unless that's part of the storytelling, like having a jump cut, which is when you're changing time and place without a dissolve. So it gets a little bit technical. So what I tell people, again, go back to your favorite LLM and explain to it that you are going to be— and I use the word marrying— you're editing together. So if I want to make a 90-second clip, I am going to be taking 9 10-second clips and I'm stringing them together. I will tell the AI that's what I'm doing. And so I need to make sure that they're going to cut together. And so it will know. And I'll say, I don't want any jump cuts. So it will say, if I'm doing something where I recently did something with my avatar where I'm Gilligan on an island and I'm talking about how I've been here for 62 years, no rescue. And then I cut back to me and it's saying, Actually, all of this I just made up this morning. This is what some people call AI slop. I say it's blah, blah, blah, right? So I took those 2 clips and then I'm like, okay, AI, what can I do that's funny in between? And it's like, oh, do a montage of life on the island since you've been there. And it started giving me these like, and the montage was in 2 20 or 2 10-second clips. And then that's where you do have to have some editing. I say use CapCut, use whatever you want. But you have to be able to take these clips into a third-party editor and put them together and cut out the little mistakes, cut out the little things that don't work, and be able to put them together to create your entire clip.

The OmniLabs thing, can it do more of this sophisticated stuff? Do you recommend going there when we're doing these multi-scene longer videos, or can it be just as easily done through APIs or through You know, Gemini in general?

Both of those. I think they all work if you're brand new. And because what— let me first go back to the Google Labs. Google Labs does, and what they're trying to do is make it so you can prompt your next prompt off of the clip you just generated.

Oh, that's cool.

It is cool, but it doesn't always do it great. And so what I have found is because I know how to edit, it's easier for me just to generate the next clip and know that I'm going to marry them together. In CapCut. But you can play with it and try it. I think I personally found it a little frustrating. And again, this can totally change tomorrow and they will have mastered it a little bit better and it will be able to read that last clip a little bit more and do it seamlessly. Right now it's not there. So again, I go to generate your clips. You always have the option whether you're using a third-party aggregator, because like what you said, using an MCP, I do a lot of stuff in Claude, and I have it connected to a couple of my aggregator OpenArt and Higgsfield. And I love that because then I just spend all my time in Claude and I do it. That is a little bit more sophisticated than the person that's just beginning to do this. And so that's where I say, if you're brand new, just play in the Gemini app either on your desktop or on your phone because you can do it with the $20 subscription and you can begin to see what it does. Once you get there and get get comfortable and you see all the fun stuff it does, that's when I would either get Higgsfield or OpenArt or go to Google Labs and connect it and be able to just, uh, to really begin to play. The one thing I do wanna mention is if you wanna change the video, you have to do it in Google Labs or in one of the aggregate ones. You can't do it on your app.

Let's talk about that. Let's say we wanna add text or motion graphics and stuff like that. Talk to me a little bit about like what, What it can do inside of the Labs with an existing video that you've already, you know, either uploaded that's less than 10 seconds, that's real or AI-generated, you know, what kind of edits can we do? Let's talk about that a little bit.

Yeah, this is where it blows my mind. So let's say you filmed a cooking show and you filmed yourself cooking a steak and you wanted to just add some motion graphics of maybe some what looks like, like slates that say like ribeye steak and something that says salt and pepper that's kind of coming out of your real video. You can prompt it to do that very natural language prompt, and it will look like a motion graphics artist went and added these graphics to your video. It is incredible. Absolutely incredible. On a very basic example, let's say I'm walking in a park and it's summer and I film my feet, and then I take the camera and I show the trees. I can bring that into Omni and I can say, make it winter, and it will keep everything the same. My shoes, my pants, everything exactly the same. And as I pan up, and I tested this, it took the lovely trees with all the leaves and it made them leafless and with snow on top of it. And it looks real. Like it does an amazing job at that.

Okay. But it's limited to 10 seconds, right? So does that mean you have to crop your video to 10 seconds in order for this to work? Because what if you have a longer than 10-second video? Any tips?

Yeah, you do. That is, that is the big drawback today is you have to be thinking in 10-second chunks. But having worked in television, unless you're doing this long dialogue, most of your shots you are going to use less than 10 seconds. So if you're putting something together, it really is that. Now, if you want to do something fun with a talking video, because I could change my background, I could take this And I could be like, I don't really like my background. I want to keep Eve talking just like I'm saying everything to you with my hands and everything, but put me on a lawn chair on the beach having this conversation. Or maybe I want more of a studio look like the news or something and like text coming up. I can get Omni to do that, but you are restricted to the 10 seconds.

If it's longer than 10 seconds, does it give you a cropping tool or do you have to like— Yeah, it does.

Yep.

Okay, cool.

Cool.

So if you upload something, you just select the 10 seconds and it will just import that 10 seconds, export it back out.

It's very easy.

Eve, we have just barely scratched the surface of this absolutely spectacular new tool that you just mentioned called Gemini Omni. You did mention earlier that you were going to provide some sort of something to everybody, so go ahead and tell everybody where they can find that. But also tell them if, you know, they want to connect with you on the socials, where should they go? And if they're interested in checking out whatever you've got going Yeah, I love that.

So I have a school community that is the best place if you want access to me, and it's called Video Spark, and it's a very fun interactive community. I am on all the socials under Eve Whitaker AI. I spend most of my time on Facebook, so if you want to message me, that's where you'll find me. But I'm also on LinkedIn and Instagram also. And there was another question and I forgot it. Oh yeah, you, you said you have some sort of resource that you were going to Yeah, I can give it to you and you can put it in the link, or people can message me and I'll send it to them. So either or. And it's just a PDF that goes into very step by step how you create your Gemini avatar.

Well, let's have everybody message you because why not, right? So folks, message Eve Whitaker, ideally on Messenger, and just tell her that they were AI Explored.

Well, when I did this last time, they just said, heard you on the podcast. Can I get that PDF? Yes.

Okay, perfect. Awesome.

Yeah.

Eve, thank you so much for sharing your insights with us today.

Oh, I had a blast. I always enjoy this.

Hey, if you missed anything, we took all the notes for you over at socialmediaexaminer.com/a124. Be sure to follow this show on your favorite podcasting app. And if you've been a longtime listener, we would love a review on whatever platform you're listening on. And also let your friends know about this show. You can tag me on Facebook, LinkedIn, Instagram, X, and TikTok. And do check out my other show, The Social Media Marketing Podcast. This brings us to the end of the AI Explored Podcast. I am your host, Michael Stelzner. I'll be back with you next week. I hope you make the best out of your day and may AI help you become more successful.

The AI Explored Podcast is a production of Social Media Examiner.

One more thing. We just announced the first speakers for AI Business World 2027. Dedicated AI sessions, personally recruited speakers, pitch-free training designed for implementation, and virtual tickets cover both AI Business World and Social Media Marketing World. See the lineup at aibusinessworld.live.