ai vlogs are taking over the internet right now and in this video I'll show you exactly how you can create them and the exact prompting strategy you need to ensure every single clip is perfect hey guys today we are catching fish looks like we caught a big one all we got to do is wait now smells good already and that video you just watched was actually created all inside Google's new tool now I've already made a tutorial on how to use Google Flow but I'll show you guys exactly how you can recreate this specific tutorial using this specific platform so if you haven't already hit the link in the description head on over to Google Flow so you can get started so I'm actually going to open up this project here because this is going to give you guys the original prompts and show you guys the exact workflow that I used to generate just one of the series that you're seeing online that's going completely viral so one of the first videos the first one that you guys saw here this one of course is of Bigfoot in this forest situation and you can see that he's holding up a fish now I think you guys know what the videos are but the most important thing you guys have to understand is of course the prompting structure and so what we really want to pay attention to is this lower bit so I'm just going to click this button right here and you're going to see it opens up this smaller menu where I can see the entire prompt now what you're first probably going to be surprised by is the fact that there isn't some extensive prompt the main thing to realize is that VO does all of the heavy lifting for you including some of the minor imperfections in the speech that make it sound so realistic so essentially what we have here is first the subject and what they're doing so I first this is how I structure all my prompts i would first say Bigfoot is holding a fish and then of course I would add the camera angle now remarkably this actually took me a very long time to get down i tried many different things for me this is the only thing that will give you that POV style look the one that everyone is currently using and you want to have this selfie camera angle and then of course you want to have shot from extended arm perspective then of course lastly we want to have our environment because the first thing we want to do here this first section is going to be the image prompt so that section is going to be basically setting up how the first frame is now once that first frame is built you can see I've just put he says and then in quotation marks I put "Hey guys today we're catching a fish looks like we caught a big one." And that's how I get my first clip there now I would say remember to initially you know maybe test this through a much cheaper image generation prompt that way you can understand if you've got the perspective right and also that the model understands what it is that you're going to be doing remember any minor imperfection in your prompt will result in a poor or terrible generation and that means you're going to once again have to generate even more clips resulting in even more credits luckily for me I'm able to afford the ultra tier because I spend so much time in AI but for those of you who have much less credits you're on a much tighter credit budget meaning that you need to be a little bit more careful to ensure that you first have the subject then have the subject action then have the camera angle then of course have the environment and then of course have the last section be exactly what it is that they are doing so you can see right here that looks pretty epic now of course with this one I of course got another angle here i'm not even sure if I use this VO actually generated this one but you can see once again considering the fact that I've used the same prompt the only thing that I've changed is a few things so once again you can see I've changed the environment so I said Bigfoot is now sitting at a campfire with a fish in his hand once again this is the prompt structure you want to first have the subject then you want to have the action how are they doing this one was Bigfoot was standing i said he was holding a fish and of course this one I said he's sitting at a campfire with a fish in his hand i used the same exact camera angle the same exact environment because that is how you get the environment to have consistency and that's why your videos won't look weird so if you have a video where he's at a lake make sure it starts at a lake or near a lake so that you can have at least a little bit of consistency and considering it's a dense forest the reason I like this kind of environment is it's hard to pick out anything that is a little bit different i will say that in the background the trees are maybe a little bit different but in a forest it's believable that you move to an area where there are less dense trees so now once again I've just put in what are we going to do now is cook this bad boy so I actually generated two videos and I will say that sometimes what can happen is that sometimes your generations may not work so for example this generation right here this one didn't work but I do believe that this one was V2 so you can see right here it says V2 fast so with V3 the highest quality one that is going to be how you're going to get number one the audio number two the highest quality so you ideally want to be using V3 so you can even see here when I use VO2 even this Bigfoot style creature kind of looks like a different kind of monkey i'm not sure what they're called but they're the orange ones i really don't have the name i think it's an orangutan but you can see right here whilst yes this is remarkably impressive it doesn't look as good as this one right here so we can see right here this is Bigfoot singing at a campfire with this fish in hand and I did say the vlogging camera is there and he cut and remember guys this was actually a mistake and this is why you need to be super careful with your prompts considering I put vlogging camera right here it just plpped the vlogging camera right there and of course it says he cuts the fish open guts it and places it on top of a stone that is on top of the campfire in a dense forest and he says "We have to clean the guts out first of course." Now you can see right here this prompting strategy is remarkably effective okay it's able to give you the clips that you want but that isn't the only thing and trust me guys there was something that I realized that a lot of creators were struggling with that considering the fact that they didn't think about this was why their videos aren't as good as some of those top creators that I'm seeing on platforms like Tik Tok what you need to be able to do is once you've generated a scene that is somewhat continuous you want to be able to use one of those frames to generate the next scene so let's say for example I've got Bigfoot right here what I was able to do was I was able to add this to my scene so now that I've clicked add this to scene it's going to open up a scene builder a scene builder is a visual timeline where you can visualize your clips and you can crop them delete them and add multiple clips together so what I did in this instance was I played the clip and then towards the end of the video what I did was I basically used this final shot here and it's got this really cool button where you can save this frame as an asset so if I save this frame as an asset now what I'm doing is I'm creating continuity in my shots and I know I just said that wrong but we're continuing on with the video so you can see right here I've already saved this image right here but the image will automatically be saved into your entire environment what this allows you to do is if you want to create a longer shot it allows you to do that with a lot much more character consistency so once we've got this image right here what we'll need to do is we'll need to go to frames to video it's much easier if you do it here it's just how I think and if you click plus and then we add this then when you generate your prompt it's actually going to use this as the first frame remember we generated this video and of course we can use that last frame where he was you know doing whatever he was with the fish where he was laying it down for him to do something else with the fish for me the prompt that I used was that he squeezes some lemon on top of it so we can see right here it looks a little bit more consistent in terms of the consistency the only thing that happened here that really annoyed me was that it didn't actually get the text prompt which sometimes unfortunately does happen which means you're going to have to generate it again but of course it's much easier to add all of this to your scene builder so I'm going to add this one to my scene builder then I'm going to come back over to here remember when using Google Flow do not click all the way over to Flow because you will lose all your progress you won't lose your videos but if you're building something in the scene builder and you've edited it and you've cropped out the clips you will lose that progress and you will have to do it again so let me actually add the first clip right here what I'm going to do is add this first clip add to scene and so now I want to arrange these if I just click arrange I'm going to just go over here and then so now you can see that I've clicked done we can see that when I arrange these we'll see exactly what happens again so remember this is the video I showed you guys at the beginning we've got Bigfoot with his food saying "Guys here whatever yada yada yada." And then of course here he places the fish down which is really cool loving this clip right here and then of course as well what we do see is that he's then seasoning the fish with another clip as well so of course if I wanted to finish this off what I could do is I could add this clip so what I could now do now to even extend this is I could save this frame as an asset so now it's going to upload and then now it's going to just basically save that as an asset i do realize that on short form platforms they don't really want you to save frames it would probably be better to generate an entirely new scene what I could do here is I could say Bigfoot starts eating the cooked fish in the dense jungle i could say with his hands and then of course I could say right here i could say and then of course I will add he then says this is tasty nothing like fresh fish and so now what it's going to do is it's going to use that last frame and so this is more for you filmmakers if you want to really just have a long extended scene but you could just have this prompt on its own since Google recognizes this character and kind of just has a really consistent theme so this one I'm going to go once again making sure I double check and I click that this is V3 often times you will burn through your credits if you don't double check which one it is and always make sure you have one on because by default it's set to two so change it to one and then click enter now as well if you want to have a different style of prompt you can come back to this page and then if you ever want to use or reuse a style of prompt what you can do is you can basically just click this button right here so we can click this reuse button if we want to reuse a prompt so for example this one right here i can reuse this and I can then just edit these sections so I can say Bigfoot sitting at a campfire with a fish and he's eating it and of course I could say he eats the fish and then I could say he says and then I says this is how we do it so once again of course this is going to be probably a little bit different considering I'm not using a first frame once again I will just generate this for the video sake and so this is actually a clear example of why VO can sometimes be unpredictable and why you always want to use the same prompt and ideally the same image so with this one I used the actual frame and we can see here that we can see that of course he actually starts to eat the fish and then you can see of course this is the kind of consistency that we do want in the clips for some reason it generated some crazy looking creature here that's not what I really imagined and so overall sometimes they will say your words sometimes they will slow your words sometimes you may actually even get a human person holding a fish i'm not sure who this guy is but he looks happy with his fish so yeah this is the tutorial if you want to make these videos of course just go to the scene builder and for example for that last scene what I will do is I will just add this last scene to the video and so yeah you guys can see that this entire thing here this is how we do the clip we've got 30 seconds of usable footage and then if we want to download this all all we need to do is click this download button and it basically just starts exporting everything of course you'd have to crop this for Tik Tok but it doesn't give you the option to have 9x6 which is of course the aspect ratio and so yeah the tutorial is not over just yet i actually got a few more interesting things because there's a lot more formats that you guys do want to know so for example when dealing with multiple characters it's always best to distinguish the characters in this specific prompt format so I've said initially what the composition is i've said Bigfoot and a white yeti and then of course I always say what they're doing they are sitting and then of course I added the environment the environment is a campfire with a fish on top of the stone that is on top of the campfire in a dense forest very similar to what we said before and then of course when we introduce the dialogue we have to make sure we once again reference the first character and what they say and then referencing the second character and what they say because if we just say he says this and he says that it's not going to work at all and you're going to get messed up generations so what we first do is we will say Bigfoot says it's great to finally have a friend here and then we'll say the white yeti says "Yeah man it's getting lonely up in these icy mountains." And that's how we actually managed to get a clip that looks just as good as this one right here ah it's great to finally have a friend here yeah man it was getting lonely in them icy mountains and so the point is I'm going to go over this one more time so you understand it when dealing with two subjects or more we have both subjects being described then describe what they're doing are they walking are they running are they sat down then describe the environment that they're in then we describe the character one what they're saying then we describe character two and what they're saying so remember that when you're dealing with two subjects or more in a specific composition right now V3 works best with two characters maximum with three characters it starts to fail a little bit now remember guys it's not just Yeti vlogs that you can do you can do other characters as well so what I have done is I've actually added Albert Einstein here once again we follow the same structure and pattern we did the first strategy so who is the character Albert Einstein i of course added the camera POV and then I said it's handheld and he's sitting under a tree so of course you can see I've got the action I've got the character and I've got you know the location which is under a tree and then I says he says the only reason I can say he says here is because there's only one character in the scene so Vio will not get confused and I says so guess what i figured out something wild stay tuned for a vlog guys this is big and you guys can watch that clip now so guess what i just figured out something wild stay tuned for a vlog guys this is big and so you can see right there that when you use this prompting style you get really really nice results from V3 you won't get those messy generations and as long as you focus on the subject what they're doing and the environment and then you focus on what they're saying you'll easily be able to create tons of viral videos that you can share a lot now another example one that is also going viral is the GoPro selfie camera stick so what you can do here is once again we use the same structure this time we just have a different POV camera so for this one I have a plague doctor vlogging with a selfie stick okay that's what the character is doing it's a plague doctor vlogging with a selfie stick and here I input the camera POV i put selfie stick camera POV because of course he has a selfie stick but we want the camera placed on the selfie stick then I say he's walking remember we need to add the action around the murky streets of London and this is going to be the environment in which we place him and then of course I say the plague doctor then says yada yada yada this plague is really getting out of hand and that's how we structure our prompt to get this final clip wow this plague is really getting out of hand i got to get out of here now the most surprising thing for me when generating all of these clips is that Google's V3 will do a lot of the heavy lifting for you if you're confused about anything it will literally just add the breathing i'm not sure how but it does manage to add the context for you for example in this clip you can actually hear the fact that the breathing is inside the mask which means that Vio is quite likely looking and analyzing visually what's going on to be able to generate those sound effects so unless you really want some kind of specific emotion you don't need to say that unless you want that specific emotion even in these Bigfoot clips there was a lot of breathing going on which is exactly what you would expect so don't think you need some super crazy prompts you just need to once again have the character have what they're doing have the camera POV if necessary have the action the environment and then of course we have that there so with that being said I think that pretty much covers absolutely everything that you want as always if you want more scenes you can just literally add it to the scene