Create Your Own AI Avatar in 10 Minutes with Haygen: Step-by-Step Guide
Learn how to create a personalized AI avatar using Haygen in just 10 minutes. Automate your content workflow and enhance your video production effortlessly.
File
How To Create Your Own AI Clone for Videos (No More Shooting)
Added on 09/30/2024
Speakers
add Add new speaker

Speaker 1: This is my AI avatar that looks like me, talks like me, and gestures like me. And I made this in only 10 minutes using this tool called Haygen. And in this video, I'm going to teach you how to do the same as well. This completely automates your content workflow. You no longer have to shoot. Without further ado, let's get started. Alright, so this is the real me and I'm going to walk you through the process of creating your own AI avatar. Let's go. So first of all, I'm going to app.haygen.com and I'm going to create an account. Now, I already have an account, so I'm just going to sign in. Now, once I'm signed in, I can see a bunch of options like instant avatar, photo avatar, template, AI script, etc. You can do a bunch of things with this tool. But in this video, I'm only going to teach you how to create your own AI avatar that can do videos for you. The whole point of this is to automate your videos in some way or you not having to record a message a thousand times and just having to type in a script or just do a simple voiceover. So the first thing you need to do is click on instant avatar. And once you do that, you can actually see this option called free instant avatar. Now, whenever you sign up for HayGen, you get one free instant avatar and you get one free credit out of which you can basically make a 60 second video. After that free version, you have to basically buy credits. I'll quickly walk you over through the pricing plan, but let's first create our avatar and then get into it. So I'm going to click on free instant avatar and this video is going to play. Hey guys, I'm Joshua, co-founder and CEO of HayGen. Let me show you these two videos. By the way, this is not the actual CEO. This is the HayGen version of the CEO. Pretty cool, right? Let's get started. So you can either take text instructions or video instructions. I prefer text because this is like an instruction manual. So I'm just going to click on that. So let's quickly go through the rules and regulations or the do's and the do nots. So the way this tool works is that it takes a two to five minute video footage of yourself, and then it kind of processes it. It understands your gestures. It understands your voice. It kind of stabilizes your backgrounds. It understands. Your hand movements, your facial expressions. It understands all of that. That's what the model does. And then it tries to create an AI replica of you so that whenever you enter a script or you do a voiceover, it can actually emulate you in some way or the other. So there are some rules to ensure that you get the best quality output. I'll just walk you through that. Use a high resolution camera. So the quality has to be top notch. 1080p will actually do the job. Record in a well lit, quiet environment. Try to minimize background noise as much as possible and have good lighting. Look directly into the camera. Look directly into the camera, talk directly into the camera, and try to maintain eye contact. Pause between each sentence with your mouth closed. It will only be able to correctly humanize you in some way if you were to actually close your mouth when you pause. So I know this rule can be a little tricky, but just try to keep this in mind. Use generic gestures and keep hands below your chest. Try to emote your hands in this region. Things to avoid stitches or cuts of your footage. Yes, very important. I did this mistake the first time. I was trying to create my avatar. Don't have a lot of cuts. The best input that you can actually give is a video that is continuously two minutes or five minutes or anywhere between that long. No cuts at all. Talking without pauses, like I said earlier. Changing positions while recording. Don't do this. Stay. Loud background noises. Duh. Shadows and overexposure on your face. Get your lighting correct, basically. Diverting your gaze or looking around. Not too much head movement. Hand gestures above the chest. Don't do that. Use of pointing gesture. Ah, so you're not supposed to do this, this, this. I've seen a lot of these text-to-video, text-to-image tools screw up hands a lot. So I think it can't really get the details of the hands or the fingers correct, which is why it's asking you to do this. All right, let's move on to the next step. So you can either upload an existing footage or you can actually record it with your webcam. Now, obviously, like I mentioned, you need high quality footage. And most webcams. Usually suck. So I wouldn't really recommend the webcam unless you're just trying to fiddle around with this tool and see what it's like. So I already have a footage of myself. I'm going to go here and I'm going to upload that file. So this is the file that I have of myself. OpenAI had their first ever Dev Conference and Sam Altman came on stage to announce some massive updates to ChatGPT and the GPT infrastructure. There were majorly three announcements. And we will be going into a deep dive of all three of them individually. So that's basically the video. But again, guys, I made a little bit of a mistake over here. I wasn't supposed to really keep my hands over my chest. So just keep that in mind. Put it a little down and you'll get much better outputs. But even when I did this, I still got like pretty decent outputs. But if you really want the best output, keep it below your chest. So this is the only video I have. So I'm just going to put it in here. And they ask you a bunch of questions. Your face is visible at all times. Yes. You're looking directly into the camera. Yes. There are pauses between sentences. Absolutely. The environment is well lit and quiet. Yes. My footage looks good. All right, let's go. Now, in the age of AI and deepfakes, it is very common for someone to actually use someone else's face. There was this whole Rashmi Mandana thing that happened on Twitter when somebody did a deepfake video and Amitabh Bachchan retweeted it saying that they can take legal action. In order to prevent this, you are required to provide a legal action. You are required to provide a legal consent that you are consenting Heijan in order to create an avatar of yourself so that you can actually go ahead to record the videos. So I'm just going to quickly record my consent. Turn on mic and cam. It basically has a script. I have to just read the script that's going to come on the screen. Hereby declare that I authorize Heijan to use the footage of me to build a Heijan avatar and use it in my Heijan account. All right, it has my consent. So the tool actually validates. It validates the consent while we're actually uploading it. So that's going to take about 20 seconds and I'm going to hit submit. So it's going to take a little while for my video to get uploaded. Now, while our video is getting ready, it's going to take about two to three minutes. I'm going to quickly take you through the pricing if I go in the monthly prices. So in the free tier, you can basically get one free credit, which is a 60 second video and also one instant avatar. But there are different scale as you go plans over here. So you can just decrease it or increase it. I've opted for the lowest. Plan, which cost me $30 a month with $30 a month, I get 15 credits a month, which means I can make 15 one minute videos, which is more than enough for me. And the video can actually be five minutes long. You also get three instant avatars, which means you can either film yourself wearing three different outfits or you can film three different creators as well or three different backgrounds, whatever that works for you. And you get access to a bunch of other premium features that I'm not going to talk about in this video. So this is a pretty decent plan. And if you want more, you can basically. Just click on the slider and basically the only thing that changes is the number of credits. Everything else is the same. All right. Let's see if our video is ready. Yeah, I actually have to wait for this. The most annoying part about AI is that it makes you wait for so long. One eternity later. All right. It's done. Let's see our model. Congrats. Your AI avatar is ready. All right. Hey, Sridev Ramesh, your instant avatar is ready. Feel free to create videos with it. Also, click the feedback button to share what you think. Hope you enjoy. Wow. It actually caught all my gestures, right? My voice is right, but my accent is really weird. So one problem that I've actually encountered with this tool is that it doesn't really get Indian accents really well. So what I essentially would recommend is that if you don't have like an American or British or like a Western accent, what you should do is just record a voiceover and turn that into a video rather than write a script and turn that into a video. You can write a script. But you'll sound fake. So it's better to just record it in your normal voice so that you don't have to shoot in front of a camera all the time and set up a studio and things like that. You basically can just record a voiceover and convert that into a video. So I'm going to show you what that's actually like. Let's hit on create video. So here you see the Hagen editor. So this is the model that has been selected. And here I can basically write whatever script I want. Now, you guys actually saw how bad the audio quality in the script was. So let's actually try going for an audio script. You can upload a local audio file. You can probably record it on your audio software or on your iPhone or something. Or you can directly just record it from here. I'm just going to directly record it from here. Hey, guys, this is an official tutorial of Hagen. I'm going to teach you how to make your own AI avatar. And this is available on 100X. If you like this video, subscribe to 100X right now. All right, let's see how that sounds like. Hey, guys, this is an official tutorial of Hagen. I'm going to teach you how to make your own AI avatar. And this is available on 100X. If you like this video, subscribe to 100X right now. All right, I'm going to hit submit. So essentially it's going to cost me 0.5 credits. So anything below 30 seconds is 0.5 credits. Anything between 30 seconds to 60 seconds is one credit. That's the math. Hey, guys, this is an official tutorial of Hagen. I'm going to teach you how to make your own AI avatar. And this is available on 100X. If you like this video, subscribe to 100X right now. Right now, I'm going to show you how to make your own AI avatar. Right now, what I've created here is a pretty basic avatar, which does a pretty decent job of emulating your gestures, et cetera, et cetera. But one important thing that I want to cover over here is fine tuning. If you're really serious about putting out videos with Hagen, what I would recommend is to fine tune your models. It understands the small nuances. It is a much better output than the normal version, and it definitely costs a little more in order to generate a fine tune model. But if you're serious about this, it's actually worth it. All right. Now, in order to fine tune it, you can just click the video and you can see this button called Fine Tune. I'm going to click on Fine Tune. And like I said, you have to pay a little bit extra in order to fine tune your model. Now, you can only choose to fine tune your video or you can choose to fine tune your video plus your audio. But since this tool is not that great at capturing Indian accents, I wouldn't want to actually fine tune my voice. So I'm going to click no on the voice, which means it's basically going to cost me $40 in order to fine tune my actual video. Fine tuning, what it basically does is your video basically gets a higher resolution, which means you can get your videos in 4K. You also get a better lip syncing in the fine tune model, and it gets all the finer details of your gestures. The thing about fine tuning is that it actually takes 8 to 12 hours, which means there is actual training from your 2 to 5 minute footage video that is being done. So one suggestion is if you want to fine tune your video model, I would suggest putting a 5 minute clip instead of just a 2 minute clip because that's bigger data. So you can keep it overnight. You can leave it overnight. Next morning you come, your fine tuned model will be ready. Now, I've already fine tuned my model, so I'm just going to close this and I'm going to show you what my fine tuned model looks like. The audio still sucks. The accent still sucks. And that's just a normal problem that's there with Hagen. But let's actually create a video with this and see. So here's what I'm going to do. I'm going to get my AI avatar in order to record the ending of this video. And that is how you create your own personalized AI avatar. You no longer have to shoot. I've actually used my AI avatar in my Instagram reels and those reels have pulled over a million views. It actually works. If people have seen you for long enough, they'll understand that it's not you. But in about a few months, this is only going to get better and better. Thank you for watching. And if you want more AI stuff, subscribe to 100x.

ai AI Insights
Summary

Generate a brief summary highlighting the main points of the transcript.

Generate
Title

Generate a concise and relevant title for the transcript based on the main themes and content discussed.

Generate
Keywords

Identify and highlight the key words or phrases most relevant to the content of the transcript.

Generate
Enter your query
Sentiments

Analyze the emotional tone of the transcript to determine whether the sentiment is positive, negative, or neutral.

Generate
Quizzes

Create interactive quizzes based on the content of the transcript to test comprehension or engage users.

Generate
{{ secondsToHumanTime(time) }}
Back
Forward
{{ Math.round(speed * 100) / 100 }}x
{{ secondsToHumanTime(duration) }}
close
New speaker
Add speaker
close
Edit speaker
Save changes
close
Share Transcript