here's 30 tips to help you master creating AI videos using Google Veo3 whether you're a total beginner or an advanced user I guarantee that you'll find something in this tutorial to help you level up your AI videos tip number one you can create these vlog style videos of a talking character such as a white yeti simply by using the prompt self camera angle shot from an extended arm then we need to fill in the rest of the details so I added that it's a Yeti with wife for holding up a fish he's in an icy landscape below that I also gave it some dialogue for the character to say like "Welcome back to Yeti Adventures today we're learning how to go ice fishing." By the way this is all done inside the text video generator welcome back to Yeti Adventures and today we're learning how to go ice fishing now for a complete vlog video we also need some different camera angles so we can try a different prompt of holding up a vlog cam on a selfie stick he is talking to the vlog cam i've been out here since sunrise just me no internet trolls asking if I'm related to Bigfoot we can also use the camera angle in the prompt shot from a vlog camera this will give us a full wide angle perspective of the yeti doing whatever activities he's doing in this case she's smashing a hole through an icy lake of course we can use all sorts of different characters for example a stormtrooper in our vlogs morning trooper squad just polished the armor got to keep that Imperial shine tip number two create man on the street style interviews simply by using the phrase man on the street i just start my prompt with a man on the street interview inside Gotham City a young female reporter askked a young man what he thinks about Batman the young man is unhappy with how many problems Batman causes what are your thoughts on Batman honestly he causes more problems than he solves this city's a mess because of him now in Sodom Prompts we can also specifically tell each character the exact dialogue we want them to say for this example where I'm having a Stormtrooper reporter interviewing a Sith I specifically want a script for each character to say for the Stormtrooper reporter I tell him to say "Are the rumors about the war in the outer rim true?" And for the male Sith I'll specifically tell him to say "The Outer Rim is ripe for conquest the Empire always expands." Are the rumors about war in the Outer Rim true the Outer Rim is ripe for conquest the Empire always expands it's important that you're clear in specifically what part of the dialogue is attributed to each character in the shot tip number three removing the subtitles you've probably noticed that in the videos I've generated there are subtitles at the bottom that pop up frequently this is kind of a problem in Google Veil right now i've seen different pieces of advice of how to get rid of the subtitles for example I've heard people say in the prompt don't use quotation marks when describing what you want them to say i've also heard people say add phrases like no subtitles to your prompts but I've tried these all none of them work consistently you're frequently still going to get these subtitle captions at the bottom of your videos there's a couple different ways of doing this although none of them are ideal so one option is you can go inside a video editor and crop out the part where the subtitles appear so in this video the subtitles appear at the bottom i can zoom out increase the size of the video and then hide the part with the subtitles that imperial shine remember cleanliness is next to loyalty another option is to use an AI subtitle remover i tried a couple different ones v-make has a decent one so if we go and use the video subtitle remover and let's go ahead and upload the video with the subtitles that we want to get rid of and you can see these subtitle captions appearing at the bottom it'll take a minute to process the video and clean up the subtitles for us now if you're using the free version you do get to preview a 5second video morning trooper squad just polished the armor got to keep that Imperial shine and you'll see that at the bottom subtitles have been cleanly removed now if you're going to be using their free version you might need to chalk your video up into different parts to get the subtitles removed in all of it um but I just paid for whatever the subscription was tip number four you can add in accents for your characters when they're speaking adding in the accents isn't as simple as just writing it into the prompt you need to consider whether it's plausible in the environment for that character to have the accent for example if I try to generate a Star Wars character with a Hispanic accent here's what it turns out to sound like mercy is a lie the weak tell themselves power speaks that doesn't sound like the character has an accent at all the reason is because inside a Star Wars movie there's never really any characters with accents like this however if we change the environment for example a medieval monastery we can easily introduce a British accent into there since that's very plausible based on movies and films they don't just take gold they take souls burn the churches mock the cross laugh as we scream the location of where you generate your video contributes to whether or not the character has an accent in the video or not i generated a video of a Brazilian man inside a store in New York City he's just going to sound like a regular American guy good morning can I get a ham and cheese on a roll toasted and make it with a smile like yours however if I generate a video of a Brazilian man on a beach in Rio de Janeiro listen to what this sounds like [Music] it gave him so much of an accent that it actually change the language that he's speaking in tip number five you can control the tone of voice that the characters speak with for example I can ask for a Viking commander to aggressively shout and command his troops warriors to the monastery take it down we can also go the opposite way and generate a character that's nervous stuttering and unsure of himself the they're the Vikings they're they're coming i I saw them hundreds now when it comes to controlling the tone of voice I found that this works much better if you don't have a specific script that you want your character to follow and instead just describe the general idea of what they should be talking about here I've got this old man whispering about what it was like to fight back in the day fighting alongside Lord Nounaga those were days of fire and steel we carved a path through chaos tip number six adding ambient sounds into the background here I got a woman at the beach in Brazil and in the prank I specifically told it to generate the sound of waves at the beach oh you guys are too funny come on let's go for a swim this is another example of her walking through a forest with a sound of a creek running by and here's an example of the Brazilian man again except this time in the prompt I specifically asked for there to be the sounds of wind rustling and cars honking in the background calling by adding in these ambient sounds you create a video that's more immersive by the way if there's any Brazilians watching tell me down in the comments what this guy is saying i'm curious to find out tip number seven in addition to ambient sounds you can also specifically prompt for the types of music that you want to hear in the background you can prompt for suspenseful and thrilling music we can also prompt for a more relaxed piano playing in the background be specific in exactly the instruments that you described in this prompt I specifically told there to be a guing playing in the background which is a traditional Chinese instrument not something that you'd usually expect an AI video to be able to generate tip number eight upscaling your videos to a full HD resolution is free if you have a subscription to Google Veil 3 all you have to do is go to this download button and then select the upscaled version and Veil 3 will go ahead and upscale the video from 720 which is what it generates the videos in up to 1080p this takes a couple minutes now the upscaled video quality is okay it's not great it's not the best upscaler in the world if you look at it closely there's a little bit of graininess in the bamboo in the background for example her face also isn't that detailed but it's cool that it's a bonus feature that we get inside video 3 tip number eight use Flow TV to find inspiration for your prompts inside the Flow interface we can go to watch Flow TV up here at the top and here Google V3 has a bunch of videos that's recommended to us now below we can use show prompt and it'll actually show us the exact prompt used to generate these videos these were done using Google video 2 uh so it's not the most recent Google video 3 model but you can still get inspiration from this if we go to the grid view we can see the full assortment of videos but these are some pretty interesting looking videos you can also flip through the different channels and this is a really really cool way to get inspiration for your prompts tip number 10 can we make vertical videos inside Google Veil 3 i'm sure a lot of you shorts creators are curious about this if we look inside the Veil 3 prompting interface and scroll through the settings it doesn't really look like there's a place for us to make vertical videos but we can actually use the frames to video which lets us attach a reference image to create the vertical videos this is actually really really cool what I've done here is I've taken a originally vertical image of this uh character riding on a horse and I actually rotated it 90° so that now it's a full widescreen image i'll upload this widescreen image that's rotated 90° into Google Veo 3 so let's go to the image uploader and we can upload that rotated vertical image of him let's put that in there and then the prompt just describe it roughly it rides the horse now if we generate this image Voil 3 can actually animate a 90° rotated image believe it or not so using this method you can actually create good resolution vertical videos all you have to do is rotate the images generate the videos and then rotate the videos back into a vertical video i created a bunch of them for this character so how do we get videos of a consistent character like I'm showing you here well there's a couple of different methods to do it in video 3 i'll show you four different ways tip number 11 generating consistent characters using text video for this method we're simply going to be using the text prompt to generate a consistent character the key to this is we'll need a very very detailed description of our character's appearance as details as you can get actually in this case I asked for a female warrior with pale skin and long white braided hair braided into two thick plates that hang over her shoulders she wears a dark red cloak with loose white trousers and strap leather gear around her waist she carries a short straight sword so that's going to be the character description notice how I talked about the physical appearance her hair her skin uh the clothing that she wears and then below that we can describe what's going on in that specific scene so here it's snowing and I want a close-up shot on her face as she quietly whispers about trying to find her sister again i will find my sister i have to find her again now if I want to generate more videos of the same character I need to start by pasting in the same exact character description I used earlier except this time I'll just describe the different scene I want her to appear in she's climbing up the stairs to a dark tower a look of intensity on her face clenched teeth it's cold and snowing [Music] so it didn't follow the prompt exactly she's not climbing up the stairs she's kind of looks like she's actually coming down a well but the character appearance stays fairly similar now this isn't going to be a perfect method there's going to be some inconsistencies from video to video if you compare these two videos side by side you'll see that the clothing isn't exactly the same the facial structure is slightly different and her hair braid comes over a different shoulder but the benefit of using text to video for your consistent characters is that it can use all the features of Google V3 including making your character talk but there are other methods to get consistent characters tip number 12 using the green screen hack this method came from Martin Nebolon make sure you go and check out his Twitter but in this case what we'll need is an image of our character with a green screen behind her what we're going to do is actually go to frames to video let me get rid of this prompt then we'll upload in that image of our character with a green screen behind her then in the prompt start it with instantly jump/cut to on frame one this is what's going to give us a consistent character after that just putting what you want to happen in the scene uh she walks next to a huge white direwolf in an icy landscape it is snowing and if we take a look at some of the results I've gotten using this method you'll see that the first frame starts with a character and a green screen behind her if we go ahead and play this video it'll cut to a new scene of that character just like how we described it in the prompt there are some other examples I made for her [Music] i mean the quality of these videos aren't going to be as high as using just the text prompt you can see for some of these the contrast is extra extra high in it but it will do a decent job of mapping our character into the scene right now in Google Vo 3 there's no way to generate dialogue for the character when you're using a reference image so these shots won't include the character speaking tip number 13 using Veil 3 ingredients to create a consistent character this is going to be the third way of creating a consistent character so below frames to video we have ingredients to video underneath here we can upload a bunch of different images that we want to combine into the same video so for example I can upload an image of my character this time I'm using a white background let's upload a different character also um how about this orc and I also uploaded an image for the landscape I want them to be walking on and let's say the female warrior walks next to the orc on the rocky landscape then if we go and generate this video you'll see at the top it says switching to a compatible model for this feature that's because this ingredients to video feature only works with the older version of Google VO2 it won't technically work with the newest Google V3 model but let's try it anyways so the results don't necessarily look amazing it looks like at the end it kind of swaps to a different scene of the female character walking by herself so this isn't necessarily the best way to get a consistent character either and it's also using the older Google V2 model so you don't get all the benefits like the AI generated sounds that go along with your video so what's actually the best way to get videos of a consistent character that look the same from scene to scene well the way to do this is to use image to video where we create the images of our consistent character first and then animate them into a video which is tip number 14 so I'll start with this picture of our female character again and what I'll do is upload that image into the Flux context image generator there she is and now all I'll need to do is to directly tell the AI chatbot how to create a new image of this consistent character so inside the chat box I'll tell the AI image generator create an image of this female character in a medieval tab her appearance should stay the same cinematic film shot on a Sony camera and it does a great job of putting out consistent character into a new image let's do a before and after comparison now the shading in that photo is a little darker which is kind of what I wanted anyways for a more cinematic type of image but let's see how close the physical appearance is starting with the hair and the face area the braids that she's got over her back and over her and over her shoulder are copied into the image her face also looks pretty much the same uh if we go down and look through her accessories and clothing her belt the trousers that she's wearing the cloak the sword everything matches up i also create a couple more images of this character i'll tell the chatbot create a close-up shot on this female character's face in the same scene i mean in the original image you can't really see the face of the character that clearly so the AI had to use its imagination a little bit more to fill in what's missing but now that we have these images generated we can turn them into AI videos back inside Google Vo 3 make sure you have frames to video selected which will let us put in a reference image of our character there's one of her full body shot and in the prompt I'll just tell her to walk forward and raise her sword i generated a couple more shots of this character we can see her looking around the empty town now as a bonus tip you can also use frame interpolation what that means is we can upload the first frame we want so for example this close-up shot of the woman's face and let's also upload a last frame for the generated video this time a bird's eye view shot of her and then in the prompt if we put in something like the camera pose back what the video generator will do was generate a video that starts with the first frame of a character and ends with this last image frame that we put into here this is a really great way to control the camera movement in your scenes now the main issue with this again is that technically this feature only works with the older Google V2 model so it won't be able to generate any sound effects or anything like that now if you want to try out the Flux context model to create the images of the consistent character for yourself the way I'm doing it is using this image platform called Open Art uh starting from the Open Art homepage go to create image from there you'll see this chat toedit feature click on that and then inside we have the option to use a bunch of different image generators the new flux context max model is what I used i'll put the links in the description if you want to go try it out tip number 15 adding lip sync to your characters to generate dialogue the main limitation of using this method to create a consistent character is that when you're using a reference image in Google VO3 it can't actually generate character dialogue for you so if we actually want our character to be speaking we'll have to use an external AA app to make that happen there's some different websites where you can do this from one place that I like to use for lip syncing videos is a website called Design inside there they have this lip sync if you try this we can upload an image of our character i've uploaded this uh video of the female character that we generated in Veo 3 and now let's adding a voice so I'll upload my own audio file uh since I already generated that all right let's see what it comes up with it's finished generating so let's take a look we call them monsters but when I looked into the eyes of the last one I slayed I saw sorrow okay so obviously this is not the ideal solution it would be much better if we could just directly do this inside Veo 3 but for now this is a workaround we can use tip number 16 creating videos of consistent objects or products inside Google Veo 3 just like how we generated the images of a consistent character first we can also generate images of a consistent product or an object or a location again I'm using the Flux context AI image generator and I want to generate more images of this specific product i can simply go and tell the AI create a side profile photo of a young Canadian man wearing the blue headphones the gold patterns on the headphones should remain the same on a street in Montreal shot on a Sony camera and the AI bots able to put the exact headphones onto a model for us do a quick inspection the gold and blue colors look the same but I want to look a little bit closer at how refined the details are so this main logo looks very very very similar these little blotches of gold these mini patterns that go across the headphones also look the same and when we have the images of a consistent product we can simply animate that into a video inside Google Veil 3 [Music] tip number 17 creating fight scenes google V3 is incredible at animating live character animations and it's gotten pretty good at creating fight scenes also it's as easy as putting the words uh kung fu fight scene inside your prompt for example a lot of these videos start off kind of slow but as the video goes on you can see it turns into a much more dynamic action scene here's a lightsaber d I created that has the same thing going on in the beginning it starts off kind of slow and then turns into a very very dynamic action scene again if you want to try to put a specific character inside the fight scenes in the prompt describe the exact appearance of that character for here I described the appearance of the female warrior that I showed earlier and then in the prompt I describe what the rest of the fight scene should look like the movements aren't going to be exact but the fights have come a really really long way tip number 18 using fast mode there's a new fast mode we can use inside video 3 right here that says fast text to video now this only works with the text prompts it's only one of the cost as using a regular high quality model so I animated some fight scenes using this fast mode model the motions aren't quite as dynamic as using a regular model but the quality is still really really high compared to other AI video generators the arms and legs and anatomy of the character stay pretty consistent even when they roll around but one limitation of fast mode is that you can't generate character dialogue while it's much much cheaper and faster and definitely worth using for the higher quality shots that you want you'll still need to use a regular high-quality model tip number 19 adding camera shots and camera angles to your prompts to get the most cinematic looking videos change up the camera shot you're looking at the character from for example we can look at a close-up shot of the character which focuses in on his face we were a proud people but under Roman rule our traditions are mocked our resources plundered this life is unbearable we can take the full body shot that will show the entire body of the character generate a video of the character from a different camera direction like a side profile shot queen Cleopatra a ruler of unmatched intellect and charisma who captivated the hearts of many and we can pull the camera all the way back for an extreme long shot use all these different camera shots to upgrade your storytelling also think about the camera angle that we look at the character from for example if we use a low angle shot where we're placed below the character and looking up it makes them look bigger stronger more powerful if we use a high angle shot where we're looking down at the character it makes them look smaller weaker we can also go all the way above them to a bird's eye view overhead shot in this case you get a more objective look over the entire environment i'm sure you've noticed for a lot of the videos I just showed it kind of looks like a 3D video game cinematic instead of the lifelike photorealistic characters that I showed earlier the video game style does happen quite a bit using Google V3 if especially if you try to generate more fantasy style characters like this Egyptian warrior tip number 19 is to use a specific keywords muted color cinematic film so here I generated some more videos of that same exact character except I added in the phrase muted colors and cinematic film and you'll see that his face turns into a photorealistic looking person life under Roman rule is unbearable they impose harsh taxes and disrespect our traditions we must reclaim our freedom so this prompt does work in most cases however you are going to run into some situations where the characters just generated into a video game style pretty much no matter what so in those cases I suggest you just change your prompt and try to create something else tip number 21 introduce camera motions into your prompts camera motions help you tell a much more dynamic story in this video I told the camera to tilt up from the female character to reveal an eagle flying in the sky here's an example where I told the camera to pan from the female character and show a abandoned monastery far in the distance by adding camera motions we can relate our character to the environment around them in this scene I told the female character to cut down the knight and for the camera to zoom in on her face during the battle it actually looks like it swaps in between so it starts with the two of them fighting and then it swaps directly to her face this can happen if you try to introduce a lot of camera motion into your videos tip number 22 combine different camera shots camera angles and camera movements together for the most action-packed videos in this case I started with a close-up shot on the character's face and then I told the camera to pull back to reveal her riding on a horse here's another example that combines a bunch of different movements together i asked for a slow crane shot behind the female character walking up the steps of an ancient castle then the camera flies above and slowly tilts down to reveal the entire castle by starting with a shot of the character and then zooming out to reveal the entire castle we're able to demonstrate the scale of the landscape and environment compared to how small that the character is tip number 23 change the camera lens or the device that you're recording the video from for example a fisheye lens that's an extra wide angle camera that's going to create these distorted spherical effects switch up the exact type of camera that you're using depending on the visual aesthetic that you want for the video if you use macro lens that's great for extreme close-up shots especially of very very small subjects with fine details we can also prompt for an infrared camera as well tip number 24 you can create a firsterson point of view short film using Google V3 all we need to do is add the words firsterson POV shot to our prompts in this case I'm going to show a point of view of a character running through a trench in World War I this is a really unique way for storytelling using a video it's a very very immersive film style that puts the viewer directly in the character shoes you can taste how bland these beings are when looking at this video but let's go to some happier times before the war tip number 25 change up the movie genre that you use in your prompts to completely change the style of the videos pi created a group of friends sitting around a table inside a horror movie now I'm not sure why this looks like a Harry Potter movie because those weren't horror movies but if we compare this shot to the same exact scene generated for a comedy movie it's a completely different vibe the lighting and colors are much more lighthearted than a comedy here's what it looks like when I generated the same scene inside a sci-fi movie okay I'm not sure what's going on there maybe they're playing some futuristic game changing up the movie genre is a great way to switch up the visual style and also the colors that we see inside the video tip number 26 you can create the AI videos with different animation styles so here's a video of the robot scavenging through a crash landing in the normal cinematic video style but we can also generate that same exact video using a 3D Pixar style for example this one has a cute looking robot typical of a Pixar movie now when it comes to creating different animation styles for example in an anime scene make sure you put the animation style you want at the start of the prompt this one I specifically told it to create a 2D anime and I put that keyword at the beginning of the prompt so Google Veil 3 does weigh the style keywords you use at the beginning of the prompt more heavily than the ones you use towards the end of the prompt so if you want a specific style try to put them more towards the start of the prompt tip number 27 use different lighting styles and also color palettes in your videos lighting can make a huge difference whether you're using soft sunlight beaming through the trees or if we have harsher moonlight casting down on our characters in the middle of the night adding different light sources for example we can have light shining on him from a fire now the color grading of the video can also make a big difference in the way that the videos look you can use cool blue color tones and it creates a much more cold and dreary feeling video try using warmer colors for something that feels a little softer and more approachable i personally love using desaturated colors which you're seeing in movies like Saving Private Ryan and it gives a more gritty type of feeling to the film monochromatic color schemes where you only use one color shade like green are also popular this green color gives off an uneasy and unnatural feeling to the viewer but we can go with something that's much more colorful for example light pastel color tones that have soft pinks and soft blues tip number 28 making infinite looping videos so inside the app itself there isn't actually a way to create an infinite looping video but let's say I want to take this video I have of the jellyfish in this plastic injection machine which I actually found using FL TV and I want to create an infinite looping video of this specific shot then we can put that video inside a video editor now let's copy that and paste it so we get two shots of that same exact video next to each other and what we're going to do with a second shot is actually reverse it so I'm using the Cap Cut video editor uh this is what the reverse button looks like and what it'll do is play the video clip backwards so now that we've reversed the second video clip when we put them next to each other it's going to create an infinite loop and we can stack as many of these video clips next to each other as we want for a long infinite looping video tip number 29 use the image generator inside Google Veil 3 to create previews for what your video looks like before generating them if we go to the frames to video feature and we click on this add an image we've been uploading the images ourselves so far but it can actually generate an image also so let's tell it to generate a photo photo of a panther in a jungle golden afternoon light and shadows green foliage shot on a Sony camera and it'll generate a set of preview images for us that we can pick and choose from but some of them are a bit closer some of them are taken from a bit farther back um we also see a few different camera angles slightly different color palettes the point is if you use this feature it'll give you an idea of what your video is going to look like before you generate it and now we can go ahead and use this image to turn it into a video uh he quietly stalks in the forest let's create that using a reference image first gives us a little more control over what the final video looks like tip number 30 creating longer videos inside Veil 3 so inside Veil 3 we can use this add to scene feature and inside here we do have the option to extend the video so let's say the totems lapse onto the ground and break apart now if we try to generate that you'll see this error again that says switching into a compatible model for this feature unfortunately this feature also only works for the older Google V2 model but let's see what happens if we generate this okay so if we look at the extended video it does kind of follow my prompt the totems in the distance break apart a little bit and collapse the problem is that it is the older Veil 2 model so the quality won't be as good and also it's not able to generate any sound effects so that means you'll need adding the sound effects by yourself later on if you want a deep dive into how to best use Google Vo3 to create the most cinematic looking AI films go watch this tutorial right here [Music]