인생 초상화: 정적 이미지를 애니메이션 동영상으로 변환하기

### 인물 사진에 생기를 불어넣는 라이프 포트레이트: 이미지를 움직이는 말하는 초상화로 변환하기
* 👨 인생 초상화는 **모든 인물 이미지를 애니메이션으로 말하는 이미지**로 변환하여 인물 사진에 생동감을 불어넣을 수 있는 새로운 도구입니다.
* 🖼 인생 초상화를 설치하려면 **픽 서포터용 원클릭 설치 프로그램 또는 수동 설치 프로세스**의 두 가지 옵션이 있습니다.
* 💻 이 도구를 사용하려면 **Python, Windows용 Git, FFMpeg**와 같은 특정 소프트웨어 종속성을 설치해야 합니다.
* 🚀 설치 과정에는 **인생 초상화 저장소를 복제**하고, **가상 환경을 설정**하고, **사전 학습된 모델을 다운로드**하는 과정이 포함됩니다.
* 🎭 사용자는 **소스 이미지와 소스 동영상을 업로드**하고 애니메이션 설정을 조정한 다음 **'애니메이션'을 클릭**하여 애니메이션 초상화를 생성할 수 있습니다.
* 🎙 고급 사용자 정의의 경우, 사용자는 **비디오에서 오디오를 분리**하고, **RVC를 사용하여 음성을 수정**하고, **수정된 오디오를 애니메이션 초상화와 결합**할 수 있습니다.
* 🌐 이 도구는 **로컬 컴퓨터에서 실행**하거나 **컴퓨터 성능이 낮은 사용자를 위해 RunPod에서 실행**할 수 있으며, 두 설정 모두에 대한 자세한 지침이 제공됩니다.
* 🔮 향후 업데이트를 통해 사용자가 직접 동영상 소스를 사용할 수 있게 되면, 이 도구를 통해 **더욱 창의적인 작업**을 할 수 있게 될 것입니다.

라이프 포트레이트는 정적인 인물 사진을 움직이는 움직이는 이미지로 변환하여 디지털 아트와 상호 작용하는 방식을 혁신합니다.

Are you tired of seeing the same old static lifeless portraits? Do you wish you could bring any portrait to life? Well, now you can, and today I'm going to show you how. Hello, humans! My name is SK Overlord, and oh boy, I am back with a banger because today I'm going to show you an incredible new tool called Life Portrait that allows you to perfectly transform any portrait image into an animated talking image. Just to put a video of someone talking and a source image, and it will then transfer the video onto the image, giving you this absolutely amazing quality. Just, yeah, this is insane. And with a bit of editing and RVC, you can give your character a new voice that fits perfectly. And in this video, I'm going to show you how to do something like this from scratch. So, that being said, sit back, relax, and let's go. And to install Life Portrait, you have two ways. The first is, of course, by using the one-click installer that is available for my Pict supporters. Just download the files onto

your computer. If you don't have FFM Peg installed already onto your computer and it's not added to Puff, first make sure to run this file as an administrator so that it automatically installs FFM Peg. And once this is done, you can double-click on the live portrait onto install the bat file, which will automatically install live portrait onto your computer. And after a few minutes, you should now see a local URL that if you press control, then left-click, it will automatically launch the web UI in your browser, ready to be used. Simple as that. You don't need to do anything. Oh, and also do not forget to put the launcher. bot file inside the live portrait folder so that the next time you want to launch live portrait, you can just double-click on the launch of the bath file to start the web UI. And the second way to install this is, of course, the manual way. So first, make sure that you have Python, Git for Windows, and FFM Peg installed. Then you're going to come to the live portrait GitHub page, click on the little

button, and then click here to copy the URL. So then, in your folder, you're going to click on the folder path, type CMD, press enter. This will bring the command prompt window, and here you're going to type git clone, followed by the URL that we copied previously, which will then clone the repository onto our computer. So next, we're going to go inside that folder, then create a virtual environment for Python, then activate that environment. Next, we're going to install PyTorch with this command, then we're going to install the requirements. So then, for the next step, I made things a little bit easier for you when it comes to models, so you can download them easily. But for this, you need to delete the pre-trained weights folder. So basically, just select the folder and then press delete. So you can then click the link in the description down below. You're going to arrive on my Hugging Face repository, where here you're going to click on this little icon, then clone repository. Then here, you're going to click on this little icon to copy this

entire line. Then you're going to paste this line right here, and then you're going to do the same with the second line. So just copy and paste to automatically download all the models, and then finally to launch the web UI. You're going to type python app.py, which will then give you a local URL that if you click on it, it will open to web UI. And now we can finally have some fun. So then how do we do this? Well, this actually frightly easy. So first, you're going to upload your Source image. So for example, something like this. Then you're going to upload your Source video. I'm simply going to be using like one of the examples right here. Now this one might seem a little silly, but it is actually a really good example of what is possible because that woman makes some very weird-looking faces, which really displays a huge range of motion, which is really, really good for this demo. So then you scroll down, you click on animation instructions and options. So here basically in the instructions, it

tells you what you need to do. Basically, the only two checkboxes that you need to pay attention to are this one: do crop source and do crop driving video. So, basically, if your source image is not a square, you need to enable this option, and if your driving video is not a square, you need to check this option as well. Now, in our case, the driving video is a square, so we don't need to do anything. But I will show you later in the video an example of a video where we need to use this option. But basically, for now, the only thing that we need to do is just click on the button animate and wait for everything to be finished. Now, surprisingly, it actually doesn't really take a lot of VRAM to run, but it still uses a lot of computation and power, like RAM and CPU power, to do the job. But don't worry, even if you don't have a powerful computer, I will show you later in the video how to install this on RunPod. So, even if you don't have a

power from computer. You can rent the GPU for a few cents an hour and do the exact same thing. So that in the end, we get something like this. Yeah, I mean, this is just crazy. And if you just look at the before and after with half the source image in the middle. I mean, the result is just absolutely insane. I mean, just absolutely incredible because the precision vision of the movement and just like the overall quality of the generation is really, really impressive. And once you're done, if you just want to save the video, you can just click on the little button and it will download the video onto your computer. Okay, so now then, how do we actually take this further? How do we, for example, take a clip of someone talking and then apply it to another person's face and then use another voice for the video so that the character talks in a custom voice? Now, all you need to do is just do some very basic video editing so that you get a clip. So like, for example, I got something like this weeks after President Biden's disastrous debate.

performance just a clip of a woman talking that I'm going to put right here. Then here you're going to put your source for an image, let's say about this image of Donald Trump. Then here in the options, since our video is not corrupted as Square, we need to enable this option this time and leave everything else by default. And now if I click animate so that in the end we get something like this weeks after President Biden's disastrous debate performance. He today faced the world's media in an appearance labeled make or break for his election campaign. So and yeah there we go. This looks pretty good. Unfortunately, this is not the right voice. Although we still see Trump, we definitely still hear the voice of the woman. So how can we change that? Well actually, if you separate the audio from the video with a very basic video editing tool, we can then use sof like RVC to convert the audio into something else. And if you don't know what RVC is and you don't know what I'm talking about, you should definitely

watch my RVC video that I did a few months ago because otherwise you're not going to understand what's coming next. And basically from this, it is very, very simple. Basically, just select your Trump voice that you downloaded somewhere else, decrease the transfer's value a little bit to maybe sub L minus 7. Then here, enter the path of the audio file. You can leave everything else by default and then click convert. And then after a few seconds, you will get something like this: Wakes up to President Biden's disastrous debate performance. He today faced the world's media in an appearance labeled make or break for his election campaign. So did he make it or did he break it? And yeah, that definitely sounds a little bit better. And in the end, once you put the video and audio together, it should give you something like this: Who wakes after to President Biden's disastrous debate performance. He today faced the world's media in an appearance labeled make or break for his election campaign. So did he make it or did he.

break it. I mean, yeah, there you go. The results are pretty crazy. Now, obviously, you can do this with every character you want. You can do it with cartoon voices, non-realistic portraits. You have here a bunch of examples that you can use. But yeah, I mean, just from a single image and a video, you get some very, very interesting and very cool results. Now, don't worry, even if you don't have a powerful computer to run the life portrait, I will show you how to use runp pod to do the exact same thing. Now, first, all you have to do, if you don't have a runp pod account, just click the link to the description down below and create a new account on runpod. Then click on deploy. Then, to save money, you're going to select Community cloud and then scroll down until you see the RTX 390. So, select it. Then click change template. Make sure that you have selected rumo py 2.1 and then click click Deploy on demand. Then you're going to click my pods. Wait for everything to

be installed. Then you're going to click connect, connect to Jupyter Lab, and then here, if you are one of my picture supporters, this is going to be very easy for you because I made a one-click installer for you to use. So just drag and drop the Run Pod installer inside the workspace right here, then click on this little terminal icon, and then copy and paste these two lines of code that you will find in the Patreon post, and then press enter, and it will install Life Portrait automatically. And once this is done, in the end, you should get a public URL that, if you click on it, will open the web UI as if it was running on your local computer. Simple as that. You really don't need to do anything, but don't worry, I will also show you how to do it the manual way. Okay, so first, what you want to do is to click on this terminal icon, then before doing anything, we're going to install FFM Peg with these two commands: apt-get update and apt-get install -y ffmpeg. Then we're going to

to close the repository, then go inside the folder. Then we're going to install the requirements. Then just like for the local install, we're going to delete the pre-train W folder. Then we're going to install G lsf with these three commands. Then we're going to download the model's repository. And then finally, we're going to launch the web UI with the command python app.py D-Share. So that once again in the end, you get a public URL that if you click on it will launch the web UI ready to be used. And there you go. Now you can finally have some fun. Select an image, then the video, and then click animate. And in the end, you get something like this. Yeah, that's kind of weird, but hey, at least it works. Oh, and also what's really cool is that maybe very, very soon we might be able to actually use his source videos instead of a simple image. So basically, we'll be able to merge two videos together. Like for example, look at this. You have like a

video of a woman and a video of a man, and when you put them all together, you get something like this, which is just insane. I mean, look at this. You have here a woman with an open mouth looking to the side, and a video of a man smiling. And when you put all of them together, you have a man smiling with an open mouth looking to the side. I mean, this is just absolutely insane. And this should probably come in the next few days, so maybe by the time that you're watching this video, you might be able to use a video instead of an image for the source. So yeah, this life portrait software has a lot of potential. Oh, and also, don't forget that I provide priority support for my Pat supporters. So if you have any questions whatsoever, do not hesitate to send me a DM, and I will try to answer your question as soon as possible. The link for my Patreon will be in the description down below. So yeah, there you go. This was life portrait, an absolutely amazing software that allows you to

generate an absolutely amazing array of portraits from a video source, and really with this technology, the sky is the limit, especially when the ability to use a video as a source will be made available. So once we have that, this is going to be very, very fun. So yeah, there you go, definitely try this out yourself and have some fun. And there we are with folks. Thank you guys so much for watching. Don't forget to subscribe and smash the like button for the YouTube algorithm. Thank you also so much to my P supporters for supporting my videos. You guys are absolutely awesome. You people are the reason why I'm able to make this video, so thank you so much, and I'll see you guys next time. Bye-bye.

RIP DEEPFAKES! FREE AI Brings ANY PHOTO TO LIFE!

Welcome to Aitrepreneur, I make content about AI (Artificial Intelligence), Machine Learning and new technology.

Thank you so much for watching and don't forget to subscribe! Thanks!



Aitrepreneur

인생 초상화는 동영상을 오버레이하여 정적인 인물 이미지를 움직이는 말하는 이미지로 변환하는 새로운 도구입니다.

픽트 서포터용 원클릭 인스톨러를 통해 설치하거나 Python 및 Git을 사용하여 수동으로 설치할 수 있습니다.

이 과정에는 저장소를 복제하고, 가상 환경을 설정하고, 원활한 변환을 위해 필요한 모델을 다운로드하는 과정이 포함됩니다.

애니메이션용 웹 UI를 사용하려면 먼저 소스 이미지와 주행 동영상을 업로드합니다.

소스 이미지와 주행 동영상이 정사각형이 아닌 경우 자르기 옵션에 주의하세요.

애니메이션 버튼을 클릭한 후 컴퓨팅 성능을 위해 RunPod에서 GPU를 대여할 수 있으며, 완료되면 애니메이션 동영상을 다운로드할 수 있습니다.

보이스오버 변환을 사용한 동영상 제작

이 과정에는 여성의 음성 클립과 도널드 트럼프의 이미지를 동영상에 사용하는 것이 포함됩니다.

오디오는 비디오에서 분리된 다음 RVC를 사용하여 트럼프의 음성으로 변환됩니다.

마지막으로 비디오와 오디오가 결합되어 매끄럽고 인상적인 결과물이 만들어집니다.

인생 초상화 설치하기: 단계별 가이드

라이프 포트레이트를 설치하려면 Jupyter Lab에서 Run Pod 설치 관리자를 끌어다 놓은 다음 터미널에서 제공된 코드를 실행하여 자동으로 설치합니다.

수동 설치의 경우 FFMpeg를 업데이트하여 설치하고, 리포지토리와 설치 요구 사항을 설정한 후 지정된 명령어로 웹 UI를 실행합니다.

설정이 완료되면 애니메이션을 적용할 이미지와 동영상을 선택할 수 있으며, 나중에 두 개의 동영상을 병합할 수도 있습니다.

인물 사진 소프트웨어의 잠재력

이 소프트웨어는 여성과 남성의 비디오를 결합하여 남성은 미소를 짓고 여성은 입을 벌리고 옆을 바라보는 독특한 초상화를 만듭니다.

곧 동영상을 초상화 소스로 사용할 수 있는 기능이 추가되어 소프트웨어의 잠재력이 더욱 향상될 예정입니다.

제작자는 Patreon 후원자에게 우선 지원을 제공하고 시청자가 소프트웨어를 사용해보고 피드백을 제공하도록 권장합니다.