Loading summary
Transcript109 lines
- [00:01]
A
Hey there, Mike Stelzner. Before we get started with today's podcast, picture this. While your peers are struggling with basic chat GPT prompts, you're the marketer everyone turns to for AI solutions. You're automating tasks that used to take hours, creating stunning visuals in minutes and analyzing data like a pro. This, my friends, is not a fantasy. It's exactly what's happening to marketers. It's in the AI Business Society. When you join the AI Business Society, you get monthly live training from leading experts, real world examples you can implement immediately, and a community of innovators pushing the AI boundaries. Don't let this moment pass you by. Visit Social Media Examiner.com AI and start your AI transformation today. Before we get on to today's show, here is something that might surprise you. Every marketer using AI is at one of four distinct readiness levels. Here's what's scary. Most marketers, they have no idea which level they're actually at. While you're wondering what AI skill to tackle next, your competition is already a couple steps ahead of you. Because they have a clear roadmap, they know exactly what to focus on because they understand their AI readiness level. The difference they took our free AI readiness assessment. It doesn't just tell you where you stand. It gives you a personalized 30 day plan designed specifically for your level. Stop guessing, Start progressing. Find your AI readiness level in just a few minutes by visiting social media examiner.com Aiassessment Again, social media examiner.com pause this podcast. Take the assessment right now. Welcome to the AI Explored podcast, helping.
- [01:56]
B
You put AI to work.
- [01:59]
A
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 Stelzer, and this is the podcast for marketers, creators and business owners who want to know how to put AI to work. You've probably been hearing about AI video for a while and you're like, wow, that seems kind of complicated, but I would love to experiment with it if that's you. Today's episode is going to be amazing. We're going to explore Google v3, which is right now the world's leading AI video that has been used even in commercials because the quality is so good. And you're going to learn how to create professional quality AI video without needing any film crew, without needing any real background in doing video work. And I think you're going to love it. And I'm going to be joined by Leslie Samuel, who's going to break it all down for us. If you're new to this podcast, be sure to follow this show so you do not miss any of our future content. Let's now transition over to this week's interview with Leslie Samuel, helping you simplify your AI journey. Here is this week's expert guide. Today, I'm very excited to be joined by Leslie Samuel. If you don't know who Leslie is, he is an AI enthusiast and video creator. He's the founder of Interactive biology, a top YouTube channel for demystifying biology for students around the world. He's also a professional MC and and AI speaker who is also our regular main stage MC at Social Media Marketing World. Leslie, welcome for the first time to this show, but welcome back to one of my shows. How are you doing today?
- [03:47]
B
Oh, I'm better than supandas. Thank you so much for having me. I'm excited for what we're going to be talking about today.
- [03:52]
A
Me too. Today, Leslie and I are going to explore how to create professional quality video with Google v3. Now we're going to get into all the goods. But before we go there, Leslie, how in the world did you get into AI? Start wherever you want to start. What was your journey in?
- [04:08]
B
Okay, let's start way at the beginning. So my journey to getting into AI video actually started when I was born. All right, now just bear with me for a second. When I was born, I was born on the beautiful Caribbean island of Saint Martin, and I was born prematurely at four and a half pounds. Okay? So I am this extremely tiny baby. And I'm so tiny throughout my childhood. I want you to imagine this. At the age of five, when I first went to school, I was so tiny that my mom could put out her palm and I could sit in her palm. That's how tiny I was. Now, as tiny as I was, there was one part of me that wasn't tiny at all, and that was, unfortunately, my head. Now imagine tiny Leslie big head. Imagine me as a child. You know, I'm getting teased and I'm getting teased a lot. And I can remember going to my mom and saying something like, mommy, every time I go to school, my friends are teasing me and I feel so bad. And I remember my mom coming back and saying to me, leslie, don't worry about those kids. When they tease you about the size of your head, you tell them that's where your brains are. And for some reason, there's something about her telling me that about this brain that I have Inside my head. That was comforting. It gave me confidence, and it allowed me to face my bullies. Now I want to fast forward a bit. Now I'm in college. I'm in college and I'm taking this class because I'm a terrible student at this point. I'm not doing well in classes or anything of that sort. But I sat in one class. It was a physiology class. And the first day of the class, the professor is up there, and there's a picture of a neuron on the screen, and he is talking about this neuron. And there's something about the neuron that is fascinating me. He's talking about the neuron and how it works, and it's so complex, and it allows us to do all these amazing things. Then he gets to the nervous system, he gets to the brain, and my mind is blown because that same brain that my mom told me was. Was what was inside this big head was extremely fascinating to me. And learning about how the brain function did something to me. And it re energized me, and it transformed me from a mediocre student to one that was all in on something. I was all in on trying to understand the brain, the nervous system, and how all of that functions to the point where the other students in the class were coming to me to understand this stuff. And then I started teaching them and helping them, and I was at the top of the class. And. And when I was finished with that class, I decided, I want to get more into this. I was going to do a master's in neurobiology so that I could study the brain even more. So now I'm getting my master's in neurobiology. I'm doing research between the US And Germany, and I am learning about the neurons, learning about the brain and how it functions at the deepest level as I can, because I'm so into it now. When I finished with my master's, I wanted to become a university professor, but I didn't have the PhD. So I decided I want to teach people about this stuff whether I have a PhD or not. So I went to YouTube and I made my first video. And my first video was, what is a neuron? And I started breaking down how the neuron works and how does the nervous system and the human brain works. And I'm getting into these things and sharing it, because the concept is I want to take these complex topics and break them down so that the average person can understand it. So now I'm creating these videos, and now the channel is over 370,000 subscribers. And it's growing. People are engaging with my content. And then something happens. In 2022, a lot of us know about this, but ChatGPT came on the scene. And when ChatGPT came on the scene, it turned my world upside down. Because now it seems like these machines are thinking and processing in ways that are similar to the human brain. Now you know me as the brain guy. This is fascinating me, not just in terms of how it's enabling me to be more productive, but just in terms of learning how the processing is happening. So I started leveraging ChatGPT, started learning about automation, and I started using those things in my business to transform how I do what I do. And then Google VO comes out, and then Google VO3 comes out, which allows you to create video and audio at the same time. And now it's like the limitations that were on me before were no longer there. So now I can leverage AI, which functions in similar ways to the human brain, to take my creativity to the next level. And anytime you combine how the brain works and how the processing happens with the ability to create its own, something that just captures my heart. So this, that's kind of like a snapshot of my journey to how I got here.
- [09:19]
A
Love it. I know a lot about Leslie's journey because we've been in a mastermind group for like a decade together every week. So he's going to reveal throughout the interview some of the things that he's done through his video creative process. But he is creating YouTube videos, he's creating shorts, he's creating short form, long form video, all the things. And AI is assisting him, and it's pretty powerful. What do you want to say to marketers and creators right now who are, I don't know about AI video, Not sold on it. Said another way, if they employ what we're about to talk about today, what's the upside? What's the possibilities here with this AI video stuff we're about to talk about?
- [09:57]
B
Okay, so here's the upside and here's what you need to understand. Up until this point, what I'm able to create in terms of video is limited by my ability to write scripts, my ability to storyboard and all of that stuff. But not just that sometimes you have to film. You got to recruit talent, and then you need professional editors. And if you want to do some animations and so on, you need to hire someone that can do that. There are so many aspects to this video creation process and so many what some would call limitations, right If I am a solo creator here in my office just trying to create content, I can only do so much AI video and a platform like Google vo, what it does is it removes those limitations. So as someone as simple as me, simple creator here in my home office, I can make high quality productions, which in the past were only available to Hollywood studios and people that had huge budgets. But now I can do that. We're going to talk about some of the types of videos that I've been able to create. But the main thing is this, the limitations are no longer there. Your imagination is the limitation. And if you can imagine it, if you're thinking about product videos or videos where you're telling stories and animations and so on, you are no longer limited because the tool can help you to do that in very simple ways. So the main thing is this limitations are a thing of the past. Now the only limitation is your imagination.
- [11:32]
A
Now, Leslie, you have been on television shows as an expert commentator. I believe you've even been on the news.
- [11:38]
B
Yep.
- [11:39]
A
How crazy was it to see the production that was necessary for that kind of stuff? Like, just tell people what it's like.
- [11:44]
B
Yeah, I remember. So I'm on a show on the Science Channel and I remember they have a entire crew. They have script writers, they have editors, they have videographers and so on. And you come into this room and then there's a team of people around you that are involved in even the simplest or the seemingly simplest of productions. This is a huge staff, there's huge budgets. I'm talking tens, if not hundreds of thousands of dollars to make one simple video. But now with a tool like Google Veo, I can be all of that right here in my office and I can create some amazing things. So to. To be able to go from needing tens and hundreds of thousands of dollars to just doing it here with a few bucks. It is such an amazing transformation.
- [12:31]
A
Love it. Okay, before we get into AI video, let's talk about, like any kind of mindset shifts or anything we need to be thinking about before we get into AI video just because, like, this sounds daunting, you know. So, anything you want to say there?
- [12:46]
B
Yeah, absolutely. The first thing I want to say, especially when it comes to limitations, is don't limit your thinking. Before, when I wanted to create a video, I had to think about, okay, what do I know how to do? If I want to do an animation, I might have to figure out how to build something. And for me, I would build it in Photoshop and try to get things to move in a certain way. And that was a very complicated process. Some people use motion graphics software and all of these kinds of things. And because I didn't know how to do that, I couldn't do that. The first shift in thinking is don't limit your thinking. Like if you can imagine it, you can do it. And once again, we're going to share some specific examples in a little bit, but that's number one and number two, it's similar to kind of what you got to think when you're using a platform like ChatGPT. A lot of people go to ChatGPT and they give it a prompt and it doesn't give them what they want. And they say, you see this chatgpt, it's all hype, it doesn't work. But it's a process. It's a back and forth creative process. It's the same way with AI video. It's the same way with Google VO3. You're going to start, you're going to give it a prompt, or you might give it an image and you have it create something and it might not look exactly the way you want it to look, but it's a creative process. Think about it as someone on your team that is really good at all of the video creation stuff and you are working with them hand in hand to create something awesome.
- [14:15]
A
Okay, this is not part of our script, Leslie, but I know you can answer with great comfort. Before we get into the Google VO3 details, can you just give an example of what you've been able to create with this? Just so people in their mind can wrap their head around what's possible here.
- [14:27]
B
Okay, so one that I was very excited. Mike mentioned earlier that I'm the main stage emcee for Social media marketing world and I've been emceeing at different events. I wanted to create a video to promote my emceeing. So what I decided to do was go to Google VO3 and I used ChatGPT and Gemini to help me to come up with something that was really cool. And what I ended up coming up with was a video where it is 2040 or 2050, we're way into the future and everybody's saying that AI can do everything. So they hired an MC bot to host their event. So now you have this futuristic robot that comes on and he's hosting the event and he's trying his best, but he fails. And then he wants to learn how to do better. So he then goes and studies, quote unquote, the greatest MC of all time. He watches my videos and he comes back stronger and so on. I was able to make this Sci Fi production, which I can't tell you how to do something like that without AI. But because now I have Google vo, I thought about it. I went back and forth with ChatGPT and with Gemini and we came up with this amazing concept and I was able to execute it within a few days.
- [15:43]
A
Yeah, and you had a consistent character throughout the entire thing, which is also intriguing. We're going to get into that. What about your work with the shorter form stuff? Do you want to talk about that at all?
- [15:51]
B
Oh, yeah. So I've been making a lot of short form videos on my biology channel where I'm just teaching about different things in biology. And what I've always wanted to do is create like an animated cartoon, like a Pixar animation, but I don't know how to do that. Fortunately, Google v3 knows how to do that. So I've been able to use Google VO3 to make animations, Pixar style animations. I'm doing a series where there are different invaders that come into the body and the body then issues an attack. And it's kind of like one of those adventure cartoons.
- [16:25]
A
Like Inside out the movie.
- [16:26]
B
Exactly. So I'm able to do that. And I didn't know how to do that just a little while ago, but because now the only limitation is my imagination and my creativity, I can create those kinds of things.
- [16:38]
A
Very cool. Well, I thought it was useful for everyone to kind of understand. And the applications here are way beyond this. Right. Obviously you could use this to create news, you could use this to create all sorts of things. So, okay, let's get into VO3, spelled V E O and the number three, all one word. Right. So what is it exactly? And let's break down kind of the things that it can do in a little bit more detail.
- [17:00]
B
Absolutely. Okay, so Google vo is Google's AI video creation models. And they have different versions, Google VO2 and Google V3. What it allows you to do is create eight second video clips. And those video clips can be anything. It can be animations, it can be product videos, it can be telling a story, it could be people doing things, it could be video ads that you're creating, but you're creating them eight seconds at a time. The interesting part about that is because you're creating them eight seconds at a time. If you prompted to create something and it creates it and you needed to do something else or the next step in the process. There's an issue with character consistency and we'll talk a little bit about that in a little bit. But you can create these 8 second clips and string them together to tell a bigger story and to make a bigger video. So whether you're talking about product videos, you're talking about ads, you're talking about educational videos, you're talking about news videos or whatever the case might be, you can create just about anything. The beauty of the latest model, which is Google VO3, which we're talking about now, is the visuals are much better, number one. But number two, it also allows you to do audio. So you can have a person that is talking, you can have an animal that is talking, you can have music going in the background, you can have sound effects. It's very advanced and as of Model 3, it is very realistic. I remember some of the earlier videos that I've seen. AI videos, maybe you've seen like the Will Smith eating a bowl of spaghetti and so on. It was interesting, but it was mediocre at best. It's come a long way and their models have been very much refined and it does an amazing job.
- [18:45]
A
Yeah, and there's been some that have actually been made into commercials. They're so photorealistic, you can't really tell the difference. There's been a whole series about the Old Testament. I think there was like a Moses version of it or whatever where he was out there.
- [18:57]
B
Oh, those are the best.
- [18:58]
A
Yeah. And they look super cool. So, okay, so eight seconds at a time, let's talk a little bit about how to set up the account you need to use as of right now and the credits and all that stuff.
- [19:08]
B
Yeah, when you're setting up an account, you want to use your personal account. You can use Gmail accounts for some of the AI features that Google has. You can use your Workspace account. But I found that in order to get access to Google v3 and the most advanced models, you want to use your personal account.
- [19:25]
A
As of today, that could change.
- [19:26]
B
Of course, as of today, yes, I'm pretty sure that's going to change. And it's possible that by the time they're listening, listening to this, it might have been changed by then. But you use your personal account and once you do that, Google has different plans. There's the AI Pro plan and the AI Ultra plan. If you sign up for the Pro plan, you get the first month free and then it costs 1999amonth. And for the Ultra plan, it is significantly more expensive.249amonth. And you get the first three months at 50% off. But with Google AI Pro, that's the $20 a month plan, you get a thousand credits per month. And the credits are what you use in order to make the individual clips. And there are different amounts depending on the type of clip you're making and the model that you're using.
- [20:13]
A
Talk to me about how much credits a video typically would take.
- [20:17]
B
Yeah. So if you're using Google VO3, which is the latest model, there are two ways that you can use it. You can use the fast model, which isn't as high quality, but it's really, really good. It's the one that I use and that takes about 20 credits. So at 20 credits, if you get a thousand a month, you can make 50 video clips a month. And if you are on the Ultra plan, you have 12,000 credits. So you could make significantly more than that. All right. Now, if you're using the quality model, that takes 100 credits per 8 second video. Really and truly, I never use that model because it's much more expensive and I've tried it and the difference in quality that I see is not that significant. So I stick with the fast model and I spend 20 credits per video clip that I'm generating.
- [21:06]
A
Okay, I love this. How about the resolution, the quality of the output? Here's what we know. We know that there's high quality. It looks visual.
- [21:12]
B
Yes.
- [21:12]
A
It looks photorealistic. And we know that the AI Pro is a thousand credits at about $20 a month. And AI Ultra is literally almost like 14, 15 times as more expensive at $249. But you get 12,000 credits.
- [21:26]
B
Yes.
- [21:27]
A
And you previously told me you're using the AI Ultra plan. What's the resolution of the video?
- [21:32]
B
Yeah. So when you make it, depending on how you make it. And we'll talk about the different ways you can use Google VO3 within Gemini, that you can just go into Gemini and prompt it to make a video. But when you do it that way, it's 720p. Now, Google also has a platform called Google Flow. And when you access the model from within Google Flow, that's their video creation suite, you make it at 720p, but then you can upscale it to 1080p. So generally speaking, I will do all of my video creation within Google Flow. It gives me a separate platform that I could use to create the videos and keep it all in one place and organize them in projects and so on. It's pretty good. So I make them in there and the final output is 1080p.
- [22:17]
A
We're going to get back to both those in a little bit. Well, if you haven't been tracking, there have been some major AI updates over the last month and the changes just keep coming. I know this feels overwhelming, but what if I told you you can actually thrive in this rapidly evolving AI frontier? The marketers succeeding aren't the ones with the most AI tools, they're the ones with the right guidance. That's exactly what you'll find inside the AI Business Society. Our team cuts through all the noise so you can focus on what actually moves the needle in your marketing. As member Shannon Caldwell told us, quote, these trainings are filling the gap between knowing what AI can do and knowing how to do it. Are you ready to thrive with AI? Join me inside the AI Business Society by visiting socialmediaexaminer.com AI so here's what we know so far. VO3 is not free. That's the first thing we need to understand. It is not free. You're gonna have to put some money into this and obviously not every video is going to turn out perfectly because you already mentioned it's a creative process. So why don't you give us some tips on how to like, increase the likelihood that we get what we want in the first go around. Let's talk about that a little bit.
- [23:33]
B
First thing to mention really quick, just in passing is it is not free. But if you do the AI Pro, you can get a month free trial and then you pay the 20. So if you just wanted to try it out, you can try it out for a month. Okay, so some tips for getting good quality. Number one, what I do, I go to create the actual video as the last step. Okay. And here's what I mean by that. I will try to work out the scenes first before I bring them into Google VO3. And there are different ways that you can do this. So one of the ways that I can do this is Google actually has another platform called Google Whisk. And you can prompt in Google Whisk, that's an image generation platform. You can prompt it there and it uses the same image generation part image generation features that you use in Google View, but on a visual platform that is much cheaper. So when I come up with my prompt, I will go into Google Whisk and prompt it there and see how does it look. And if it looks the way that I want it to look, then I say, okay, great. Now I'm going to go and do it in Google v3.
- [24:41]
A
Google whisk is spelled W H I S K, just in case anybody's wondering. So give us some tips on using Google Whisk. You said it's an image generator, so it sounds like it's something similar to a ChatGPT or mid journey. Could be. But you also mentioned that you storyboard, you work out the scene. So like slow down a little bit and like talk to us a little bit, give us some tips on that process, you know what I mean?
- [25:02]
B
Okay, so let me walk you through my process for creating these videos. The first thing I do is I try to figure out where do I want this to go? Like, what is the goal of this video? Is it to promote an event? Is it to promote a service that I offer? Or is it just to entertain people or educate people? Once I know what that is, I will then storyboard the entire thing out. I will create the plot, I will write the script, and I will just kind of lay that out and I will go through that process until I have a script that I like, until I have a flow that I like. So I go through all of that. Rather than going into VO3 and try to make the first scene and then make the second scene, I try to get the full picture, I try to create the entire storyline and the storyboard and then I start seeing, okay, in order for this to be a video, I know that it's going to be produced in 8 second increments. So I will actually break it out and say, okay, the first scene is going to be like this and that's my first eight second clip. The next one is going to be like this. And then I will go out and plan out each individual scene. And once I have that, in my case, I'm using Gemini all the way through to help me plan this out. But once I have that planned out, I am then going to come up with my prompts for each individual scene.
- [26:24]
A
So you're using Google Gemini, which is equivalent to ChatGPT. It's gemini.google.com and it does come with your workspace account and I think you can get free access. I'm not 100% sure.
- [26:35]
B
Yes.
- [26:35]
A
How do you even figure out what the heck a plot and script is? I mean, like, are you asking it? Because like, I don't know, most people don't even understand what the heck a plot, plot and storyline as like, is it helping you along the way?
- [26:45]
B
Yeah. So first of all, I've been making videos for a very long time, so I'm very familiar with that process. And Even so, I am actually generating the bulk of it using Gemini. I'm going to it. And I'm saying from the beginning, I'm like, hey, I have this idea. I want to do an animated series for my biology YouTube channel. Can you help me come up with a few different ideas that can work? And then it comes up with a few different ideas, and I say, hey, this idea that you have about kind of like this immune system squad that's scouring the body and trying to figure out what these foreign invaders are, can you give me some examples of topics that would work well there? And then once I have the topics, everything that I just said, can you give me an idea of what that story would look like? What would a script look like? And I'm going back and forth with the AI, with ChatGPT or Gemini in this case with, to be able to help me come up with a script that's going to work to accomplish the goal that I have. So all along the way, I'm leveraging a large language model. I'm leveraging something like ChatGPT or Gemini to help me come up with all of these different elements. And I'm not doing it on my own. So even if you have no idea how to do it, you can go and have it help you figure out how to do it.
- [28:02]
A
Okay, so presumably with Gemini, you're walking away with a script and maybe some descriptive visuals, but just text, Is that right?
- [28:09]
B
I'm walking away with a script, and I'm walking away when it's all said and done, with prompts for each individual scene.
- [28:17]
A
I see. Okay, got it. Where does Google Whisk play into all this?
- [28:21]
B
Okay, so now that I have the prompt for the scene, I'm actually going to take that prompt and I'm having a dialogue with Gemini and I'm saying, hey, in this process, first I'm going to design the scene in Whisk just to make sure it's working well. And once we have it looking good, we're going to come over and make it in Google VO3. So I'm taking the prompt that I get from Gemini. I'm going over to Google Whisk because I know it uses a similar image generation system. And I'm putting that prompt in there and I'm seeing what it comes up with. If it looks perfect, great. I've got what I need. I know that I'm going along the right lines. If it doesn't, what I do is I take the image that it made and I come back into Gemini. And I'm going to say, hey, this is what it made. I don't like it because the blood cell that it created looks a little weird. Can you tweak the prompt so that it can look more like the way that I want it to look? And once it tweaks that prompt and gives that to me, I'm going back to Whisk and I'm going to try it again. And after doing that a few times, eventually I get to exactly what I want it to be.
- [29:29]
A
Now, is Whisk free or does that cost money?
- [29:31]
B
Also, Whisk is part of the whole, like once you're using Google AI Suite, so you're on AI Pro or anything of that sort, you have access to flow, you have access to whisk, you have access to music effects and some other AI things that Google has made for creators.
- [29:47]
A
Okay, I feel like we have covered so far everything except for the audio side of it, the music and the spoken word and the ambient sounds and all that kind of stuff. Let's talk about that a little bit.
- [29:57]
B
There are two parts that we actually haven't covered in detail. One is the character consistency and how you get that. And two, there's the audio. Now, here's the thing with the audio. When you give a prompt to Google v3, you can describe as well what you want it to sound like. You can describe what you want the characters to say and all of that, and it will do that for you. Now, there's a key thing here when it comes to. And we can cover character consistency here.
- [30:25]
A
Before we get into character consistency, can we talk about the audio a little bit? Because if each audio clip or you.
- [30:30]
B
Feel like they're connected, this is connected. And that's why I'm bringing it up right now. Because when I prompt and I say, hey, I want the character to say this in this scene, and then in the next scene, I say, I want the character to say this. The problem is it is starting again, right? Each individual clip is a new clip and you want it to sound the same as well as you want it to look the same. So this is a very important point, especially if you're using the text to video prompting. If you are describing what it's supposed to do, you need to give it a very detailed description of what the character should look like, what the scene should look like, and what the character should sound like. Okay? This is going to be extremely important because the more details you give it, the more likely it's going to look the way you want. It to look from one scene to the next and sound the way you want it to sound from one scene to the next. So here's how I do that. I will ask Gemini specifically, because Gemini knows Google's platforms. I will ask it when I have that character. I will ask it, can you create a very detailed description of what this person or this animal or this character looks like and also create a very detailed description of what this character sounds like so that when I bring it into Google VO3, it is going to look very close to each individual scene. So each scene will look similar to each other scene. So that's the important step. You want to make sure that you have a very detailed description of the audio and a very detailed description of the video.
- [32:20]
A
Perfect. Okay, so let's say you got that prompt and you're about to generate the video.
- [32:25]
B
Yes.
- [32:26]
A
Like, you've, you've done all this work and we're going to get into a really advanced way that we have not yet addressed about even cooler ways to have character consistency in just a minute. But let's say now we're, we're ready. We've done all this prep work. We're now ready to start using VO3. Give us some guidance on, like, what do we do when we're in. They're actually using VO3.
- [32:45]
B
Okay. So number one, one of the first ways that you can use VO3 is using the text to video feature. So you're giving it a prompt and it's generating a video. Now your, your prompt is going to contain a few things. Number one, you had Gemini come up with a detailed description of the character. You had Gemini come up with a detailed description of the audio. And then you also had Gemini come up with a detailed description of the scene. You're gonna prompt it with all of that. Here is the scene, here is the character description. I call it the character DNA because I'm a biology nerd. And here is the voice description. And here is what I want it to say. Once I've done that, I usually, within one to three prompts, I get exactly what I want.
- [33:34]
A
Okay.
- [33:35]
B
I don't think I've gone more than three because of the process that I've used. Because I'm using Google Whisk to test out my prompts before. And because I have this detailed description of the character and of the voice, I get what I want. Now, one of the things that I always do is I know that yes, it can create sound effects, it can, you know, add background music. But I know that because I'm using them and I'm creating them in 8 second increments. There can be like an abrupt cut from the sound effect in this previous scene to the sound effect in the next scene. So I will actually prompt it to not include any background audio, any sound effects, any music, or any of that stuff. Because I would rather it just create what is happening with what is being said and add the music. Add the sound effects in post production.
- [34:28]
A
Okay. That's really important for people that do not understand post production because they don't have experience in the video world. Basically, like we have a process here at Social Media Examiner. We will bring someone in studio and film and then we have an editor that adds animation, sound effects, music, all that stuff.
- [34:43]
B
Exactly.
- [34:44]
A
Now this is a topic for another day. I'm sure there's AI post production tools that we don't even know about that are coming. But, you know, this is something that's probably best left for another tool is really what I'm hearing you say.
- [34:54]
B
Exactly. Leave all the additional sounds for something else. Add that in if you're using Final cut or imovie or what. I don't care what you're using. Cap cut. If you're keeping it simple, premiere, whatever. Add those things in post, focus on the visuals and what anyone or any character is saying, and then leave the rest for later.
- [35:16]
A
Yeah. And just so we can say this, this is how the pros do it. Right? Pros always add the sounds afterwards. They have.
- [35:22]
B
Exactly.
- [35:22]
A
Specialists. Like when you watch a movie or TV show and you hear crinkling paper, that's not the actual sound of the crinkling paper. They add that in. Okay. They do it all in post production. Okay, so you didn't say it, but there's two other ways We've talked about text to video when you're inside of VO3, and that's the process that we've been discussing right now. But there is a more exciting option.
- [35:44]
B
There is.
- [35:44]
A
Which we need to explore a little bit, and it's called Frames to Video. And explain what the heck this is, because this is going to excite any of my perfectionist friends out there who are like, I care about quality. Boom. Like, this is going to unlock a lot of things for people. Talk to me a little bit about this.
- [36:00]
B
Absolutely. So we are working within the Google Flow platform. That is what gives you access to V3 and all of the advanced things that we're talking about right now. So when you go to create a video clip, you have by default selected text to Video. And by default it's going to select one of the models. It might be Google VO2, it might be Google VEO2 quality. Google VO3 fast. Google VO3 quality. I will always go with the Google VO3 quality. But then you can select an option called Frames to Video.
- [36:34]
A
You said you always go for Google VO3 fast earlier. Do you go for quality or fast?
- [36:38]
B
Oh, fast. Sorry, did I say quality just now?
- [36:40]
A
Yeah, yeah, yeah, yeah.
- [36:41]
B
Okay, so I meant Google VO3 fast. Because it's 20 credits, I could get a whole lot more done. Okay, so with Frames to video, what that allows you to do is bring in an image and then animate that image, turn that image basically into a video and add audio and all of that kind of stuff.
- [36:59]
A
Clarify. You said animate. So we're not talking animation. It could be photorealistic images, right?
- [37:04]
B
Yeah, it could be photorealistic. So I can have an image that's generated of a person that's talking to another person, and then I can take that and bring it into Google Flow using the Frames to Video feature, and then have that person talk to the other person. Okay, so you're basically turning an image into video. Here's why. It gets very fascinating to me, because using the Google Whisk platform, I can build out the individual characters and then I can say, take this character, put it with this character. I can design a background scene. So wherever they are, the scenario, I can design that. And I could say take this person, put them with this person in this scene and make it look this way. Now, the beauty of that is you get full control over what it looks like and you can go back and forth until you have exactly what the scene is supposed to look like. And then you bring it into Google Flow and use Google VO3 fast and it transforms it into a video. And that is, in my opinion, the ultimate when it comes to character consistency based on what we have at this point. So I just design it first in Whisk, bring it into Google Flow, and I use Google VO3 and I'm able to create exactly what I want looking exactly the way that I want it to look.
- [38:24]
A
Okay, for those that are image freaks out there, there's also flux and there's Mid Journey. And I think Mid Journey is perhaps probably state of the art. Don't you agree? As far as like, character consistency and.
- [38:35]
B
Stuff like that, they are top of the line.
- [38:37]
A
So do you use Midjourney and if so, any tips on how that would work?
- [38:40]
B
Also, I have used Midjourney. I am focused Primarily on Whisk right now because it's within Google's system and it uses the same image generation features.
- [38:50]
A
And you'll know that if it's generated with wix, it's going to look exactly the same. Is that generally what I'm hearing you say?
- [38:56]
B
It's going to look very close. You can never say exactly the same because with each clip it could be slightly different, but it's as close to what it's going to be. If you're using Google Flow that I found.
- [39:06]
A
Okay, crazy couple thoughts. Could you take actual photographs of yourself and feed it in there and would it model you? Have you tried that?
- [39:14]
B
Absolutely. Oh, I have tried that. So you can take a picture of you, put it into Google Flow and turn that into a video. You can add your voice, but it's not going to sound exactly like you. So what I've started testing out doing is I will do that. I'll have a picture of me in a certain environment and have it turn it into a video and then I will use 11 labs to clone my voice. And the beauty of that is you can take the audio that's generated from VO3, put it into 11 labs and switch out your voice for it. And now you have yourself that's talking right there using Google VO3.
- [39:50]
A
Oh, that's fascinating. But if it's really you, you could technically just re record over the top of the darn thing. I don't know with that or is that too hard?
- [39:59]
B
You could do that, but then you have to synchronize what's happening on the camera with what you say. And that can get very complicated. But if you've trained 11 labs on your voice, you can basically just have it swap it out.
- [40:09]
A
So how do you get audio in there? Like, we didn't really talk about getting the audio into VO3.
- [40:14]
B
So once you've prompted it, like my prompt will say, hey, this is the scene that I want it set. These are the characters. Because I'm using mostly frames to video, this is what I want them to see. And I will have it within quotation marks. Say this and this is the style of voice that I want you to use. And once you do that, it creates it with the audio in there. It's not something that's separate. The video is created with the audio.
- [40:39]
A
No, I mean, like when you're actually using your voice, like how do you get your voice into there?
- [40:44]
B
You can't get your voice directly into VO3. You would have it create the video version of you saying Something it will not sound like you. But then that's where I was saying, I will take the audio that it generated and bring it into 11 labs, swap it out and then I have to, you know, do it in a different program, add the new audio back in.
- [41:03]
A
Oh, I see what you're saying. So you can pull the audio out and then you could train up 11 labs version of you to speak it the exactly same way, same cadence timing, all that stuff. Is that what I'm hearing you say?
- [41:16]
B
Exactly. And they have a, I think they call it a voice swap feature. So you take the audio that's generated from VO3, bring it into 11 labs and say, okay, instead of this voice, use this voice. And that voice is going to be your voice. I've seen some really cool videos that have been done that way and it actually works out really well.
- [41:36]
A
Okay. I know people are saying like, mind blown. I'm sure people are like, their minds are blown. While you're driving, keep your eyes on the road. So wait, there's a third option. Ingredients to video. We didn't really talk. Give us the quick flyover on what the heck that is.
- [41:50]
B
Yeah, so Ingredients to video is one that I've decided that I'm not going to use much. Here's why. Ingredients to video. Basically you can say, hey, this character and this character within Google Flow, you can say, put them together and have it do this. Unfortunately, as of right now, it is only available with the Google VO2 model and you can't use the Google VO3 model to do it. And when you're using ingredients to video, it's 100 credits. So you're going to get a lesser quality video, you won't be able to have any audio with it. But if you are just simply looking to take the individual ingredients, the individual characters and individual elements that you're using in your video, you can put them together using the older model, but without audio. The reason I don't use it is because I can do the same thing in whisky and then bring it over into Google Flow and use the latest models with audio to create what I'm looking to create. But that's the concept. You take the ingredients, you put them together and build a scene.
- [42:49]
A
Okay, I love this. Now I'm imagining pretty soon it's going to go beyond 8 seconds. Maybe you'll have a choice, 8, 10, 15 or whatever seconds. I am also imagining a time pretty soon where you're going to be able to generate some sort of a unique identifier that might identify A character character or a voice or something like that. But as far as the voices go, I've watched enough of these videos, and I noticed the voices kind of sound alike. Have you found that there's, like a variety of voices, or is it just a couple of voices that are in the arsenal right now?
- [43:18]
B
I find that the better a job you do at describing the voice that you want, the more unique a voice that you can get. Generally speaking, just like you. I've seen that a lot of the videos have a similar voice. The men sound a certain way, the women sound a certain way, but you can actually, with your prompting, get it to sound different. So I. I think it's something that is going to get better. I mean, this is the worst that it's going to be, and it's going to get better over time. And like you said, I do see a time where you can say, okay, this is the character that I want to use throughout. I want it to be a longer video, I want this voice, and so on. But as of right now, it all comes down to how detailed your prompt is.
- [44:00]
A
Now, you said it generally generates at 720p, unless you upgrade to something to get 1080p, you upscale. Do you know what the frame rate is? Is it 24 or 30 frames per second?
- [44:11]
B
That's a good question. I'm not sure what the frame rate is. No.
- [44:14]
A
Okay. Are you aware of other tools that will also allow upscaling? And is that kind of a mad science? When it upscales it, does it change it or does it actually look exactly the same?
- [44:22]
B
It looks exactly the same when you upscale it, but the limitation is you can only upscale it to 1080p. I've only used up to 1080p. I know there are some other AI tools will allow you to upscale even to 4K, and they're basically just algorithmically analyzing the pixels and then filling in the blanks to make it a higher resolution. But with VO3 itself, using Google Flow, the most you can do is 1080p.
- [44:46]
A
You talked about Ultra. Google Ultra, which is the $249 plan a month, is the only advantage to Ultra right now that you get more credits or are there other features that are exclusive to Ultra?
- [44:56]
B
So you get more credits. And the other tools that you get are similar to what you get on the Pro plan. You just get more of it. So with Gemini, you can use it more than you would you run into limitations on the Pro before you would on the Ultra plan.
- [45:11]
A
Leslie Samuel. People are going to want to actually see some of these cool AI videos that you've created. So my first question is, where can they find you and that video on the socials? And then if they want to check out other things that you've got going on, where do you want to send them?
- [45:26]
B
Absolutely. They can go to iam leslie samuel.com and I created a slash SME and I'm going to have a guide there to teach people how to get started with some tutorials to show you how to get started. It's going to have links to my videos on YouTube, so you can check that out there. So that's going to be I am Leslie samuel.com/SME. Everything that you need will be right there.
- [45:49]
A
And if they want to connect with you on the socials, do you have a preferred platform you want to send them to?
- [45:53]
B
Yeah, they can go to Instagram, Facebook. I am Leslie Samuel on the platform so they can find me anywhere, pretty much.
- [46:00]
A
Leslie Samuel, thank you so much for giving us this deep dive into Google v03. We're way better because of it.
- [46:05]
B
Awesome. Thank you so much for having me.
- [46:07]
A
Hey, if you missed anything, we took all the notes for you over@social mediaexaminer.com a68. And be sure to follow this show on whatever podcast you're listening to so you don't miss any of the amazing content we've got coming your way. And if you've been a listener for a little while, I would love a review. I would also love it if you would share this with your friends. You can tag me on Facebook, LinkedIn or X. And do check out our other shows, the Social Media Marketing Podcast and the Social Media Marketing Talk Show. This brings us to the end of the AI Explored Podcast. I'm your host, Michael Stelzner. I'll be back with you next week. I hope you make the best fast out of your day and may AI help you become more successful. The AI Explored Podcast is a production of Social Media Examiner. Just a quick reminder before you go. If you're ready to become indispensable in the age of AI, the AI Business Society is your solution. Join now and secure your discounted membership by visiting socialmediaexaminer.com AI I can't wait to see you inside the AI Business Society.