Transcription
What if you could go from a single idea to a fully produced, publish-ready video in under five minutes with an avatar, B-roll, captions, and script all generated automatically? What you're currently seeing was fully generated by AI with the help of Hun's new AI video, Agent Flow. And all I had to do is just give it a text prompt. And so, by the end of today's video, you're going to know exactly how to recreate the videos you're seeing here, everything step by step. So, let's get into it.
So, like I just mentioned in the intro, the tool we're going to be using today is called HeyGen. And just in case you don't have an account, by the way, you could get it down below. I'm going to leave a link for you. And so once you're here, you're going to head over to video agent. Now, obviously, they have the regular avatar flow that you might have seen me using on the channel before, but this new feature they recently dropped that is called video agent is something really cool and it's new. Basically, here with this feature, you get to be a creative director. You can prompt advanced things. You can tell the tool exactly what to create for you and it's going to do just that. So to quickly show you some examples, for instance, we can see here this guy. As you can see, it created motion graphics. So this banner with the custom text and then it's got the captions as well. And then it created this kind of a graph. You know, I guess it's relevant to the video's context. And as you can see here on the left, you can see how the concept was explained and the style. But of course, we're going to get into more of that in the video. Just showing you some brief examples of what this new flow is actually capable of.
All right. So before I even prompt anything, let me just show you how this even works and what all of these buttons do. So here obviously pretty straightforward. You can just click on this and you can attach some media elements. So you can just go to your PC and maybe attach particular pictures or videos you actually want AI to include in your generation. Here you can click on knowledge and you can also choose anything from the knowledge hub. You know, you could actually link here your Notion docs or Google Drive, uh, or, you know, your docs from the Google Drive. Uh, here you can also click on brand system and you can customize the colors for your video. I'm not going to do that because I'm not so picky, but, you know, if you are running a professional brand, I guess this could be really useful to you. And so here, if you're not sure exactly how to prompt, I recommend you using this button here that is called prompt idea. So obviously it's going to give you a random idea and it's probably not going to be that useful. But what you should do here is you should look at how the prompt was built and what AI basically communicates and underlines and you should be paying attention to that. So as you can see what I mean is it's giving us this framework where it says the prompt and then style and then avatar and then music. So to say the least, these are the four key elements of the good prompt and uh, these are the things you should keep in mind, right? Um, so, but for now, I'm going to delete this entirely. And uh, here you can also click on options and this drop-down menu is going to appear. And uh, here you can choose the length of the video. So as you can see, at the moment, you can create videos of up to 3 minutes long. And here you can also click on resolution. Um, I'm going to do actually landscape. And for the duration of the video, I'm going to do 1 minute because I guess anything below that is not going to be long enough for us to really test the tool. But as you can see, you can do videos up to 3 minutes long as of now. But I'm sure as they push new updates, this will only get longer. But all right, at this point, let's actually start creating our video.
So, you can first of all click on the avatar and uh actually select one if you wish to. But I'm going to show you how to actually prompt in the avatar selection. And I just want the AI to choose the avatar for us and just to see how it's going to turn out. So, I'm actually not going to select anything, but once again, you could select one if you wish to. So, as you can see, you've got here a bunch of different categories. And in the style here, however, I will make a selection. So, let's go for this one here that is called um watercolor. I really like this style and ambience. So, I'm going to click on that. Now, you could type in your prompt or you could use this uh ideation button as I showed you before. But what you could also do is you could actually use this custom GPT that HeyGen themselves built for you guys to use and uh, just to make prompting easier for you. And basically, I'm going to paste this prompt that I put together in advance. So, what we're going to do is I found this article that says, "Is AI really going to take your job?" I think it's a really relevant topic nowadays. And basically, I copied the link to this article. And then I pasted it here and I said, "Create a 60-second video that explains this article." And then I also gave it a special note. And it's something you should do as well. So, if you want it to make the avatar look a certain way or if you want there to be a certain hook or if you want there to be a certain music or caption style, whatever, you have to, you know, declare that right here. And so, here I did say that the host should be a YouTuber-style avatar in a well-lit studio with a visible microphone. We're going to see if it gets the job done. So, keep that in mind. And I also did say that the opening line should be, "Is AI really going to take your job, or is it all just hype? Let's find out." So, those were my requests and this is what we got back from AI. So, remember this is all a big prompt. So, you're going to go ahead and copy the entire thing. Um, so obviously you could take a minute and read through and make sure it's all good. So, for example, here in the visuals category, it's going to give a prompt for our um character avatar. And here, it's also going to basically explain in the prompt as well what kind of visuals and B-roll we want to see. So, I'm going to go back here to HeyGen and now paste this long prompt. And now, finally, let's click on generate. But before I actually click on generate, I do want to mention that you could also choose here the chat option. And basically, as it says right here, it's going to chat and plan your video. But since our prompt is already extremely detailed, I don't think it's really necessary. So, I'm just going to choose generate here so that it autogenerates the video. And don't worry, later on, you can still edit certain elements, pretty much every element. So uh, let's finally click on generate and see what happens. I'm not going to skip this part. Uh, I want you to kind of see the whole process. And basically, as you can see, it's a very similar experience to ChatGPT where it's just reasoning and thinking through this whole thing in real time. And here you can see, you know, it says connecting the dots, validating, etc. So, I'm just going to speed it up until we get to the next part of this process.
All right. Right. So it did reason for like 51 seconds and it generated the plan. So we now have the video plan. We can expand it. And as you can see, we have the concept, the style, the avatar, the voice. Pretty much everything is now systematized. And it's even created a scene-by-scene sequence where it's pretty much, it's kind of a script, if you will, where there is no randomness. There is no AI slop involved, but a clear plan for your video, right? Which is a super cool thing because later on, when what I'm going to be showing you how to actually edit certain parts or elements of your video, this is really going to come into clutch. And now is the next stage of this whole process. It is now actually generating video elements. So as you can see, it is quite literally generating video elements in real time. So I guess this is going to take another few seconds or maybe minutes.
All right. So as the generation was ongoing, a cool thing popped up which I wanted to show you. It's called suggested memories. Basically, the AI wants to understand, as it says right here, uh, memories help video agent learn your preferences and create better videos. And so, we can now review these suggestions and either accept them or reject them. And so, in case you're going to be making consistent videos, I don't know, for example, you're running an agency, a marketing agency, and you're consistently making ads in a, you know, similar manner for a similar product, and you want your style to be consistent. In that case, if it gets the style right, you actually could accept it and then click on done here. And in the future, you could expect your videos to have more consistency because just like ChatGPT, it has like a memory feature, as you just saw. So that's how it works. And in the meantime, our video has been generated. Now, I'm going to play the entire video at the end of my video, but for now, let's just give it a high-level overview. So, obviously, here's the AI avatar. He's talking and uh, we can see the graphics popping up. We got like the custom graphics that are perfectly matching the script. And so now it's about time to proceed to the AI studio uh, to where I can show you how to potentially edit this in detail. And so to do that, to edit your video in AI studio, you can click here on the right on edit a copy in AI Studio. Uh, just heads up, you could do so in this uh chat box. So literally just chat to it and tell it what you want to have edited. But I would recommend going directly into AI Studio because it's just a better and more uh, pinpointed way to edit whatever you want because as you can see, we're going to have all this uh interface in front of us where you could really get scientific with your edits.
Now, let me just walk you through the whole AI studio because this could be overwhelming at first, but it's actually fairly easy to use. Now, on the left here, you know, fundamentally the script is what makes or breaks your video, right? So, you could first of all just change the actual text. So you could click into each of the lines and for example, you don't like this line at all. Although this is the hook we prompted to have. But let's say you don't like it at all, you could just literally select it and uh, then delete it and then just, you know, type in whatever you want to have. And then also you could click on this uh triple dot button and then you could regenerate the audio. You can actually listen here to the particular audio piece and then if you don't like it, obviously you can regenerate it or delete the scene overall. And additionally, what I really like here, and I've never seen that in any other video tool, is uh, you can direct the voice. So, you could choose the emotion that it's going to be pronounced with, or you can just, you know, type in the prompt that would be used to describe your emotion. But I'm actually going to do um excited or actually it's rather a um, you know, like a philosophical provoking hook. So, let's do something like serious. Let's definitely do serious. And uh, click on apply. Okay, so we will need to rerender the video, but we'll have to do it anyway. So we can click on confirm here.
Now here on the left, you also have another cool feature. Basically, if there is a particular scene, so let's just say for example, in scene number six, I don't like this uh B-roll piece and I really want our avatar to be here. So you could click here on this human-like button and then uh, this menu is going to appear on the right and then you can choose to have an avatar. Now, it doesn't need to be your avatar, like the actual avatar we use here. You could go here to public avatars and use any other avatar, or I mean, it makes more sense to use the avatar you're already using in the video. Maybe you could have actually a dialogue like simulation and you could have multiple avatars. And uh, because I do believe you could also choose like different voiceovers for the audio. So, I'm pretty sure it's possible. Yeah, actually, right here on the right, you could uh choose a different voice. So go here and uh click into all voices and uh choose whatever fits the context, but I'm obviously not going to do that. But yeah, that's just how you do it. So you could simulate a um conversation between, you know, avatars if you wanted to. But all right, now you more or less understand how the avatar or script flow works.
Here on the left, we can actually click into AI tools button here on the right. And in my opinion, this one is pretty cool. So the first thing here is called script writer. You can basically generate scenes with the help of this tool. But um, you know, I don't really see the point to add any other scenes right now. But a cool thing here is motion designer. So you can actually go through these uh motion templates and uh customize one uh that fits your style. So for example, actually here in the bottom, you can click on landscape. So make sure to have that selected. And this one looks like something out of a video game from the '90s, you know, the synth wave. So, I do like this style. So, I'm just going to say add this motion graphic style to my video in the third scene. And so, now our piece of motion graphic got generated and it's going to appear here in my designs. So, again, let's say you don't like this B-roll clip, you can just click onto the actual clip and then just click on delete and then it's going to disappear. And then you can click on add scene or add to scene and then just adjust it um, you know, up to your screen. Uh, just like so. And uh, now you can click on play. Actually, before I click on play, as you can see, you could be adding your motion graphics into the video. But as you can see, if I uh make it fit it up to screen, it basically overlaps with this text. And uh, you know, let's imagine that you have this uh text and it's important, right? Because it's also another motion graphic. It's not actually the caption. Like it doesn't conflict with the caption, but whatever, whatever other elements you might have, it does conflict with them. So to fix it, basically you still make it full screen and then you click here at the bottom on set as background and it's basically going to fix that. And so now if I click on play, as you can see, everything is uh the way it's supposed to be. So now we can back out and we can also click on image generator. So if there is a particular image that you want to see or to have in your video and you just know that stock library doesn't have your images, uh, you could just prompt it and uh have it included in your video. Same goes for the actual video generator. So and so let me actually put the video editor to use. So for example, in the scene number two, I want to have a custom video AI video going on. So, what I'm going to do is I'm going to literally go to scene 2, copy the line, and then uh type something like this. So, it's going to go like this: "Generate an appropriate illustrating video for this line," and then just paste whatever you copied from this line. And then click on generate. Uh, as it's generating, you can see here, you can actually select the model that is going to be uh generating your video. As you can see, they've got a bunch of models. I would imagine that uh depending on the model, it's going to be consuming your credits uh slower or quicker. So, but VO3.1 is pretty much the best one you could go with. So, it will take a minute. So, let's see what it generates. So, as you can see, it did produce this clip right here. So, you can just simply click on this clip and uh make sure to fit it uh, you know, make it fit the screen and then it's going to be a part of your video just like so. We can now play back the video. And as you can see, it is now a part of your video. Perfect.
I do prefer stock footage over uh the media. By the way, the media is right here. So, if you want to insert a regular piece of B-roll, you could do so. But those are kind of boring to be honest. Even though they're here, uh most of the people are used to seeing them. Uh, well, if it's something super simple, you could of course use a B-roll. But in my opinion, using AI-generated stock footage is much cooler. It is much more original because you can really express your imagination uh with AI generations like with nothing else. Uh, and so here also you can have music. So I've already had uh selected a music track here. But as you can see, it's uh pretty straightforward. So you can have your music right and then uh click on whatever song you have here. By the way, I'm pretty sure the music is royalty-free so you can safely use it on YouTube. But then yeah, you can choose whatever song you wish to and then it's going to appear here on the timeline. And then lastly, caption style. You can also choose the actual caption style. So currently it's the brand kit style. Um, and uh, but we could go with something that is simpler like uh this you can see here on screen. And then lastly, we can go with the layout. So you can do avatar only with no motion graphics. Uh, but, you know, that kind of kills the point of uh what we've been making here. But as you can see, there is a lot of layouts. So, it's kind of a preference thing. And here at the bottom, you can also click on layers. So, another way to, you know, what we were dealing with here in the previous scene, you can actually go through each scene and make sure that everything is in order and nothing is overlapping because once again, if I just go ahead and switch the order here, or I can't really do that because it's background. But if I go here for example, and if I place the avatar above the motion graphic, you can see how it now blocks the graphic. So that's why this button is here. But I think I'm pretty sure I've shown you every single uh part of this interface and now how everything works. In case anything is still confusing, make sure to comment down below. But I'm pretty sure it's self-explanatory.
But anyway, obviously before you exit out, make sure to click here at the top right on generate uh, just so that new video gets rendered. Otherwise, your your uh changes will not be saved. So go ahead and uh, actually you can choose the name for your video also resolution. Make sure to go with the 1080p at least so that your video looks good and then uh FPS, whatever you want, and then format. And um, yeah, now let's click on submit.
And so while our render is being processed and before you get to see the final generation which I'm going to play at the end of this video, I do want to quickly mention the whole technology that powers uh, this particular AI avatar and it's pretty much one of the best ones I've seen in the market as when it comes to talking avatars. And anytime I use a talking avatar in my videos, uh, both a fictional one or even cloning myself, Hun is the tool I always use. And so recently they released a uh, new model for their avatar which is called Avatar 4. So it used to be before Avatar 3, but now it's four. So it does have realistic movements now and the face and body and background do adapt to a script. And so I already made the avatar here which we're going to see in a moment. But I did give it this simple prompt which is uh, you know, "In today's video, we're going to be talking about cutting-edge AI tools, automations, other AI-related advancements." And so that was kind of the idea for the script. And then uh, in order to actually create your avatar, you could uh go here and uh select one from the library. They obviously got a few ones. You can also build an avatar if you want to. So you could uh, uh, go ahead and basically engineer your avatar here. Also, you could clone yourself. So you could literally, as you can see, my webcam is going to appear here. So you could clone yourself just like so. Uh, or you could, you know, uh, record your video. So you can actually connect with your phone and literally just take a video of yourself, which is uh, quite crazy in my opinion. Uh, I've genuinely never seen that in any other AI tool. And obviously you can just uh, upload a picture. So yeah, that's when it comes to choosing your avatar, but I'm going to go with the one from the library. And uh, as far as your script, once again, I'm going to paste it. And you could also uh, for the script, you can record it as well. Just dictate it or upload a uh, actual audio that you have in case, you know, you're just lip-syncing your uh, video clone. And now here at the bottom, obviously make sure to select the voiceover that is going to be used. So, you know, select whatever one you prefer. And obviously here when it comes to the model, make sure to select the four and uh, finally click on generate. And um, I've already done it. So I actually generated two versions of this avatar and I'm going to play them next to each other. Uh, so I'll put together like a version four and version three. So let's take a look at them.
"In today's video, we are going to be talking about cutting-edge AI tools, automations, other AI-related advancements. Stick around until the end to ensure you don't miss out on what could potentially be life-changing information."
All right, so now you've seen what avatars actually power this technology and uh, what specifically we used in this video. Uh, in the meantime, AI Studio finished rendering our final output. So, let's give it a watch.
"Is AI really coming for your job? Let's separate the hype from the reality. AI is incredible at automating repetitive tasks. But most jobs are complex collections of many different responsibilities. Automation doesn't mean replacement. It means your role is evolving to focus on higher value work. Creativity, critical thinking, and emotional intelligence. Those are still 100% human. In fact, history shows that technology creates more roles than it destroys. It's about transformation, not loss. The key is adaptation. Those who learn to collaborate with these tools will be the ones who lead the future. Here is the truth. AI won't replace people. [music] But people who use AI will replace those who don't. So start learning today. Your future self will thank you."
All right. So to sum it up, as you just saw yourself, we can now create complete end-to-end AI videos thanks to Hen's new AI video agent. Remember, if you want to check out Hen, it's going to be linked down below. So it really is a cool technology in my opinion. I mean, we can now genuinely create an entire video end-to-end with B-roll, music, captions, videos like B-rolls, custom AI images if we wanted to, and everything in between, all within a single tool. And remember, it's only going to be getting better and better. I mean, imagine what's it going to be like in 12 months from now. So, I really cannot wait to see that. But thank you guys for watching today's tutorial. That's all I got. And if you want to see more videos from my channel, click right here. You're going to see a complete YouTube automation course that I made recently. So, go there and I'll see you in that.