Hello, nice to meet you. This time, I'm going to introduce how to use the expansion function called ADtailer in Stable Diffusion. ADtailer is a typical expansion function of Stable Diffusion that you've heard of at least once. What it does is, if you create an image without using any expansion function at first, you'll get a lot of images of your face being distorted or your hands being twisted. Especially, if your face is distorted, the whole image doesn't look good, so it's quite annoying and difficult to edit it. But if you use this ADtailer, you just need to check it once, but it automatically detects the face and edits it. It's very convenient and one of the best expansion functions to use. So this time, let's find out how to use and install this ADtailer. Installation of ADtailer is really simple. You can install it by entering the link and clicking Install as we have installed so far. I'll leave this link in the Naver Cafe and YouTube comments. Now that I have a comment I left in the cafe, I'll move to the cafe and install it with that link. I'll copy the installation link after moving to the cafe. Now, you've probably got a sense of how to install it after copying this installation link like me. Go to the settings, go to the extensions again, and there's an install from URL. Click here and paste the copy on the window that inputs the URL. And install it. After a short time like this, the installation will be completed. Now, the installation is complete. There's a comment saying to restart here. You can check if the installation is properly installed by going to the Installed tab. It's done well, and you can check if the checkbox is checked. You can click on this Restart UI. Click. If the UI starts again like this, you can easily see that ADtailer is installed. If you go down a little bit, there's another tab called ADtailer that you can click on. If you click on it, you'll get a few more things, but there's nothing complicated about it. You only need to think about two things we're going to touch. You can use this button to run ADtailer. You can't turn it off. You just need to know this. And this prompt here is how you want to turn off this face when you detect it. You're writing the same prompt. It's different from the prompt you enter above. After creating an image here, the second thing you do is ADtailer. The second thing you do is you detect the face and you change the face. And the way you transform that face is through this prompt. If you don't use a prompt, you can change it to the face that best suits this checkpoint. And this negative prompt is the same as we know. I hope nothing comes out when I transform. I use this. I don't have to use this negative prompt. The last thing to look at is this model. If you click on the model, there's a face, a hand, and a person. I don't know the other things I don't use. I usually use hands and faces, but there's not much difference. There are two models of face, but there's not much difference. If you choose a hand, there are a lot of wrinkled hands. I use it to fix my hands. But I'll explain it later. At first, I used ADtailer to fix my hands. But now I don't think ADtailer is very helpful. I only use ADtailer to fix my face. Please note that. Then I'll briefly introduce you to how to use it. I'll create an image first. Now, even if you spread out the ADtailer like this, I'm going to emphasize it later. When you use this stable diffuser, You have to turn off, turn on, and turn off this button. Just because all the windows are open like this, It's not all the functions. So you have to make sure this button is turned on or off. If this check is disabled, it won't work. If you have a check, this function is a must. If you don't need it, you have to turn it off. So I'm going to check ADtailer. Select a face model. I'll create a masterpiece with this. Now, click. Now that I've created it, I've seen the process. I'll make an image and detect the face like I just did. After detecting, you can see that the face has changed a little more naturally. ADtailer can be used in text-to-image to create images. It's already made. So you can use this image with the face already out there. There are advantages to using it in image-to-image. So if you use ADtailer in text-to-image, Anyway, once you've created an image, Then you can apply ADtailer as a post-process. I don't apply it that much in text-to-image. When I get the angle of the image I want, I often take it to image-to-image and apply it. I'll explain these tips later. Now, let's take a picture that's already been created. I'm going to turn it back to ADtailer and change the result. First of all, turn off ADtailer in text-to-image. Let's move to image-to-image. I don't think this function came out before. As ADtailer has been updated recently, Skip image-to-image has been added to image-to-image. First, let's use the image I made as an example. In image-to-image, unlike other text-to-image, It's got denoising intensity. The good thing about this is that there was no skip image-to-image before. But I know it's been updated recently. Check this function and turn on ADtailer. The overall image of the image-to-image doesn't change. It only detects the face and works. So, denoising strength was very important. If this is less than 0.5, The image of this picture doesn't change completely. It changes with this feeling. And if it's more than that, it's completely changed. ADtailer is in. If you skip image-to-image now, It removes the process completely. I'll just get rid of that process. In this image, I'll just detect the face. Only the face part is in-paint. So let's try it. I think I'm crying too much. I'll change it. With a smile. Here, I'll just create a smile. First, detect the face. You can see how it's created. I think it's changed to a bad image. If you compare it like this, The other part hasn't changed at all. The other part hasn't changed at all. I just detected the face like this. I can see that it's changed from a crying face to a smiling face. In this way, depending on how ADtailer is used in a certain situation, The effect can change by a thousandfold. And it's not all-purpose. It's very important to use it several times and learn this feeling. So if the image we've created in ADtailer is It's an image with a skip image to image. So at first, there was no denoising. So, when you release this skip image to image and increase the denoising intensity, Let's see how it changes. I'll create it the same way. If you create it, you'll just create a different image. And then you just apply ADtailer. So, image to image, I'll explain it again later. This denoising intensity is very important. If the denoising intensity is above 0.5 in image to image, You can see that it changes a lot with this original image. When it's less than 0.5, it's still a form of this image. It's good to think that it's changing to a new image as it's maintained. The fact that it's changing to a new image means that Negative and positive prompts are applied. It's good to think that it's changing to a new image. And ADtailer, not what I explained. Detection, mask processing, in-painting, there are many things. It can be too complicated to handle now. So I'll just turn it off and on briefly. And skip image to image, or not. And what kind of image do you want to change? Once you've done this enough, Once you get used to it, It's better to move on to detection, mask-free processing, or in-painting. So it's better not to touch it now. Once you get used to it, apply it and try it. Today, I used a lot in stable diffusion. I learned about ADtailer, one of the most important things. If you encounter something too difficult at first, you may give up. Especially stable diffusion can be a little difficult. Once you get used to ADtailer, I recommend you to go deeper and deeper in the next step. That's it for today's lecture. I'll see you next time with a little more informative content. Thank you.
