Google VEO3 has completely changed the AI filmmaking scene it's not just video generation anymore it's full on cinema realistic motion rich soundscapes character voices that don't sound like robots from 2007 with the right prompts you can even get consistent characters same outfit same face shot after shot no plugins no extra tools just prompt engineering and raw power but let's be real for a second I know how frustrating it is to watch epic tutorials and then your own clips look like Powerpoint presentations that's why I made this video for people who are done with surface level tips you want consistency control character loyalty across scenes no fluff no vague prompts Consistent characters hidden generation tricks and powerful prompt generators we're going deep all the prompts and reference materials I used in this video are available in my Telegram channel you'll find the link down in the description inside Google VEO3 text to video image to video and ingredients to video at first image to video might sound like the better option you upload a photo of your character and the AI animates it right but in reality text to video is way more powerful and gives you better results almost every time here's why when you use text to video the AI has full creative control it generates everything the lighting the motion the camera angles and most importantly it gives you the option to create a full voice over for your character but in practice it's how you get shots that actually look cinematic now compare that to image to video image to video works well for scenes with minimal movement but when it comes to generating dynamic high motion scenes you should always go with text to video by the way image to video doesn't support characters talking so that is a big disadvantage so whenever possible stick with text to video because it consistently delivers far better results than image to video let's start creating our consistent character there's no option to say use this person again no memory no reference tagging it's all on you to keep the character consistent and that comes down to your prompts to generate the consistent character in this video I'll start by using an image of the main character from god of war I'll take a screenshot of that image and upload it here in Chat GPT with the following prompt please provide me with a detailed prompt to recreate this image the prompt should focus on generating the most realistic and cinematic version possible once you click generate you'll receive a fully detailed prompt now we will get another prompt and combine them together let me show you this is whisk it's a tool from Google where you can generate still images and then turn those into videos and much more we will use one powerful feature that it offers you can upload any image and get a full prompt of how Google AI World sees this image we will combine this prompt and the prompt that we generated by Chat GPT and we will make a new template prompt for our consistent characters movie I copied both prompts into Chat GPT and I told it the first prompt is the prompt that you generated me earlier the second prompt is how Google AI sees this image I would like a detailed VEO3 description of just this man that I can use in a template for building prompts where I will try to place him in a consistent looking way don't worry about his clothes just focus on his face as you can see chat GPT generated a base prompt for our character then I said suggest how I should call him and how would you write a full voice prompt for him so it suggested a few voice styles for him here's an example of what it came up with OK now we will ask Chat GPT please provide me a core prompt for Kael Varn 2nd a core prompt for his voice and 3rd a core prompt for a 50 millimeter cinematic shot it gave us the example of how all three prompts should look like now the last thing is to combine that into the template this is what I told Chat GPT okay I want you to generate me a full template format where I can just paste a scene description and the characters and everything else will remain the same it gave us the same prompt template I showed you earlier here's how it's structured the first part is the full description of the character the second part is where you'll insert your scene description then comes the third part which covers the cinematic setting this is a powerful and easy way to create full movies with consistent characters just add your description and your character dialogue and you are good to go and boom this is how I made all the shots in this video now let me show you how to use it I told Chat GPT OK now according to all information that you have I want you to generate me the scene description prompts for new scenes I want you to generate me a scene description where he is near a frozen lake make sure that you are using the same chat GPT chat all the time when generating this prompt generator as you can see it generated too many details and we don't want that this is too much when you include too much detail and use very long prompts you end up confusing the AI making it harder to achieve the best possible quality so I asked chat GPT please generate simpler scene descriptions this is too long next one should be kale putting his armor on and getting ready for a fight this one is perfect not too long not too short now let's head over to Google Flow or Gemini to start working with VEO3 personally I'll be using flow because it gives us way more control over how our generations turn out once you're inside start a new project then take the template prompt we got from Chat GPT and paste it directly into the main prompt box after that grab the description prompt this is the one that defines how the scene should look and paste it in place of the placeholder text that's written inside the brackets now here's the part where we bring the character to life scroll all the way down to the end of the prompt and replace the sample line of dialogue with your own for this example I want him to say for our ancestors we fight next it's time to select your model just click the model button and you'll get two options you can go with VEO3 quality which costs 100 credits per generation and gives you the highest possible detail and cinematic finish or if you're trying to save credits go with VEO3 fast it only uses 20 credits and still gives you really strong results just with slightly lower fidelity alright let's take a look at what we've just generated not bad at all definitely gives off that cinematic vibe now let's move on and generate another one so I told Chat GPT next scene should be him taking his two swords from behind his back in an epic pre fight moment and he says the monster you've created has returned to kill you the result well it didn't quite hit the Mark visually definitely not as clean or precise as we hoped but honestly still pretty epic the mood the energy it's all there and here's something cool you can actually generate multiple variations at once I asked chat GPT amazing now create me 6 new versions but in a warlike fashion I want kale with a massive crowd behind him like they're all heading into battle I made a few videos with the scene prompts that chat GPT gave us let's check out the results just amazing the quality the atmosphere even the ambient sound design it all just clicks and best of all the character stays consistent throughout every shot let's do it one more time this time I will ask Chat GPT now generate 6 different prompts for kale as he and his army are approaching the big mountain cliff and a stunning view over the horizon I just copy pasted the scene descriptions into our prompt template and generated a few versions let's see the results this is absolutely next level just think about where video creation technology was just one year ago it's almost unrecognizable and honestly I can't even imagine what this will look like in another year you can use this exact blueprint for any character you want just get creative remix these techniques and start building entire cinematic universes the results are not 100% consistent but for now this is the best way so far we've covered text to video but we can also do frames to video in which case we upload a reference image first and then Google will create a video for us based on that reference image you might be thinking if the text to video is so good why do we still need image to video that's because text to video is still limited a lot of the time it struggles to generate the exact characters or scenes you have in mind for example this is a character from Halo franchise called Master Chief if you're not familiar with him he's basically a super soldier who is trained from childhood for combat but if I try to generate him inside Google VEO3 using only the text box it gives me a character that doesn't look like Master Chief so much Cortana please find me the new location the voice is a perfect match but the appearance is not 100% the same so let's go to frames to video we'll have an option to upload some images we can either upload our own image or generate one directly inside VEO3 I created these images using a similar method as for crafting the consistent character prompts for text to video I started by asking Chat GPT to give me the best possible prompt to recreate an image of Master Chief from Halo after that I took the prompt and asked Chat GPT to create a template one where I can simply swap out the scene description but keep the character completely consistent every time it generated a nearly perfect template prompt for us after that I told Chat GPT now I want you to focus on this prompt and not change a single detail I'll be giving you different poses motion variants and scene ideas to generate but the character prompt must always stay exactly the same then I told it to generate a brand new scene this time with Master Chief pointing a gun at an alien creature it generated a full scene description prompt all you have to do is copy that prompt paste it into whisk and just click generate you've successfully created your consistent character ready to appear in any scene you can imagine just ask Chat GPT to create a brand new prompt based on the scene you've imagined if you want a more in depth breakdown of that process check out the Google Docs LinkedIn my Telegram channel we can go ahead and use this image to generate a video of it and inside the prompt we can ask for anything for example the camera zooms in as he aims his gun let's try and generate that and it'll create a decent looking video from the reference image that we uploaded now when it comes to using the frames to video feature I don't recommend it as much as just trying to use text to video let me explain there are some decent features you can get from image to video for example if we upload a reference image let's say we'll put in another photo of our consistent character there's a bunch of options here now if we look inside the settings you can see that I have selected the Google VEO3 model for the video generation however if we try to actually run this prompt you'll see this pop up switching you to a compatible model for this feature this is because a lot of the image to video features only work with the VO2 model for example adding in the camera motions which means you won't be able to use the most advanced video model for this feature but let's generate it anyway and see what it looks like the camera movements do work pretty well depending on the scene that you need I think a lot of these will work perfectly fine the camera movements do look pretty smooth and follow what you ask for you'll need to directly add them into the prompt by yourself you do have a decent amount of control over the camera motion in just the prompt itself here's one video where the camera slides up there's one more method I want to mention while we're talking about image to video and that's the green screen hack in this case what we'll need is a single image of our character with a green screen behind here's what we're going to do head over to frames to video and upload that green screen image then in the prompt start with the phrase instantly jump cut to on frame one that specific prompting method is how you actually get the consistent character in the scene you want after that just describe what you want to happen in the scene for example he is walking forward and looking around now let's take a look at some of the results I got using this method as you'll see the first frame always starts with our character and the green screen background but when we hit play the video breaks free from that static frame and transitions smoothly into the new scene just like we described in the prompt the overall quality of these clips won't be as polished as what you'd get using the pure text to video feature in some cases the lighting might be overly intense or a few visual details might feel off but still it does a really solid job of integrating our character into the environment and that's what makes this technique so powerful because with just one reference image you can generate a consistent character that you can reuse over and over across entirely different scenes alright let's move on to subtitles because one of the biggest questions I get is if you've used Google VEO3 before there are a few options to remove them CapCut has a great feature for this I use CapCut all the time it's one of the best free AI powered editors out there so I dropped the video into the timeline selected the clip went to the video tab scrolled down to AI Remove and checked the box after a second it scanned the video I selected the brush tool did a quick swipe over the subtitles and Bam gone in some cases I even boxed out the entire area and told it exactly what to remove and honestly it cleaned it up really well even around tricky spots like the character's collar it blended everything seamlessly CapCut did an amazing job here now just a heads up this AI remove option isn't available in all countries but if you're using a VPN and connect through a US server you'll see the option appear inside CapCut personally I'm using fast VPN and it is working wonders for me fast VPN is one of the most affordable VPN services out there you can check the link in the description to get your first month for only $1 simply open your VPN application select us and click connect that's it you're now ready to use the AI Remove feature just make sure to restart CapCut after connecting your VPN if that's not an option you can also try an external website just open a new tab and type Vmake AI Subtitle Remover into the search bar click on the very first link that appears once you're on the site simply upload your video file v make will automatically process it and remove the subtitles no complicated tools no editing experience required it literally does all the hard work for you but keep in mind there is one limitation you can only download five second preview videos since VEO3 videos are 8 seconds that means we're getting cut off three seconds short but if you need more than that you can always upgrade your plan on V make and unlock the ability to download full length videos no cuts no limitations there's another feature I think is definitely worth pointing out in VEO3 it's called the ingredients to video feature this tool allows you to combine several characters or elements into a single scene it is an amazing way that you can use for generating multiple consistent characters for instance if I head down to ingredients to video I can upload three images directly into the video generator first I'll upload my two characters after that I'll use an image to generate a background that fits perfectly with them for the prompt I might write something like big guy with a beard and a futuristic soldier walk together side by side when using this image based feature it won't allow you to work with the latest VEO3 model instead it defaults to an older VO2 version the visual quality is lower and you won't get sound effects either the first result wasn't great so I tried running the same prompt again but this time I added a note at the end I described it as a cinematic film and asked for muted colors that gave me a more acceptable output let's try one more this time I will give it another prompt and use another landscape image it's a fun method to create scenes featuring several of your custom characters keep in mind the final result won't always perfectly resemble your original reference for example Master Chief didn't turn out 100% exactly like the image I uploaded but still it is a very powerful and easy way to create scenes with multiple consistent characters just keep in mind that each generation costs 100 credits last part I wanna touch on is the voice of our character one issue I had is that when you generate your consistent character sometimes the voice is a little bit off and changes slightly for example now I like the voices I got in some scenes so I took a few of those clips and stitched together About 23 seconds of Kratos speaking I exported that audio and uploaded it into 11 Labs with just 23 seconds 11 Labs was able to create a cloned voice I named it Kratos when I tried to add the voice over it didn't work 100% of the time some clips still had slight variations so I typed out the same lines as text from the video and generated multiple files until I found one with the right timing then I use that generated audio and swap the voice from the video it's not the same generated audio from VEO3 but it is still AI generated and I think keeping both the face and voice consistent is a real art this will be it for this video now you've got the blueprint the prompts the tools and the power to build an entire film studio from your keyboard if you found this helpful hit that subscribe button and don't forget to join my Telegram community where you can find a lot of gems I'll see you in the next one inside veo3 flow interface