If you look at the image, you can see that it is almost the same as the original image. Now, with the lines of this Kenny, I combined the prompt I wrote and the method step, and this image came out. So, you can think of it as a new image created by maintaining the original picture frame I provided. Hello, nice to meet you. Today, I will introduce how to use Kenny in Control Net. If you have ever used Control Net, you will definitely come across Kenny. Kenny is one of the very useful Control Net functions used in various places. Kenny is, to put it simply, a Control Net function that can extract lines from the image and combine them with the prompt based on the lines to create a new image. So, it would be good to think of one of the functions that extracts lines as Kenny. And from the time you use Control Net, the boundaries between text-to-image and image-to-image become somewhat vague. So, before moving on to Control Net, I would like to emphasize that you should know the concept of text-to-image and image-to-image to some extent so that you will not be confused. Now, let me explain how to use Kenny. After running Stable Diffusion like me, if you go down, there is a Control Net tab. Click it. Click. If you click like this, you can see several tabs and select them. I'll explain step by step. First, come down to the Control Type and select Kenny. Select Kenny and check if the preprocessor is set to Kenny. Click it. If the model is set to Kenny like this, it is successfully set. Now, what image are you going to extract lines with? So now I need to bring a picture. I brought a picture I prepared in advance. So how does this Control Net work? I posted this picture now. Then, through the function called Kenny in this picture, Kenny's function is to extract lines. After extracting the line, you can combine the line and the prompt to create a new image. To explain a little more about the function, press Kenny in the Control Type like this. After checking that the preprocessor model was selected well, you can leave the control weight and the rest of the steps as they are. This preprocessor resolution can be viewed in advance. If you press this bomb button, you can view this Kenny in advance. When this picture is extracted as a line, I'm going to combine it with a prompt later with this image. I just showed you an example. This resolution is 512. If it gets bigger than this, for example, it's 1060 now. Now, an image is created to match the resolution of 1062. When you first use it, leave it at 512 and use it. I usually use Kenny with 512. So at first, leave it at 512 as the default value. If you don't like it later, you can change it anytime. When you first use it, just follow it. Leave it as it is. There are a few more settings below. There are Kenny Low Threshold and Kenny High Threshold. It's also good to just leave this as the default value at first. If you raise this value, when you raise the Low Threshold, If you press the preview once, you can see that the line is a little shorter than when you did it at the default value. On the other hand, if you lower the Low Threshold back to 0, you can see that there are too many lines. The High Threshold probably plays a similar role. You can adjust the value yourself like this. First, try using it as the default value. If you don't like this value, then you can adjust this value yourself. Now, let's give the Kenny Low Threshold and the Kenny High Threshold the default values of 100 and 200. Let's take a look at the preview. Let's create an image with this feeling. Leave the Control Mode at Balanced. I'll leave the Resize Mode at Just Resize. And you said that the Prompt is applied. So, you can think of it as a new image created by adding my prompt to this control net. First, apply the Negative Prompt. And then, analyze this image first. Let's see what characteristics it has and how to get a similar image. First, since it's a female image, I'll use the images described in the One Girl, City, and Tree images. I'll give the Sampling Method to SD-Carras, and I'll give the Step to 30 and create it. Now, the image is created. If you look at the image, you can see that it's almost the same as the original image. Now, with these Kenny lines, I've combined the prompt I wrote and the method step, and this image came out. So, you can think of it as a new image created by maintaining the original picture frame I provided. If I want to change my outfit a little bit here, then the prompt will be of course, so I can use a different outfit for the prompt. If you say Wear Suit, you can change it to a woman in a suit, but what did I say was important from the control net? If you download it, there's something called Control Weight in the control net. This control weight can be up to 2, and even if you just give it 1, this control net frame is quite solid. So, if you just use this Wear Suit in 1 person, it probably won't work properly. Let's make it in this state and see the results. Now, the results are out, and as I said, I can't get out of that frame. So, the black color of the suit is in this frame, but the overall outfit is not worn. Because the control weight is 1, and when you check the image extracted from this Kenny line, the line that doesn't fit the suit is dressed like this. So, unless you break a strong frame, it's hard to wear a new outfit. So, how can I wear a new outfit? First, you can modify this line, or you can lower the control weight a little bit. Modifying this line is a bit difficult, so I'll explain it later if I have a chance. Just lowering the control weight can easily mix the feeling a little more. I'll briefly introduce you to that method. Now, if you look down, the control weight is 1, so you can think of it as following this line. So, even if I want to wear a bear costume in this state, I can't wear it. Because this frame is so strong. So, in order to break this frame a little bit, you have to lower the number. Now, the intensity of the clothes looks quite strong, so I'll lower the number to about 0.4. Now, what happens then? It follows this line, but it will change overall. So, even if the original image doesn't follow perfectly, there may be some changes. Let's take a look. Now, I'll leave the prompt as it was and create it again. Now, the result is out. Now, you can see that it's very different from the original image. I don't know if it feels similar overall, but by lowering the control weight, you can see that it has a lot of influence on the background or the costume. Until the previous image, the strong sleeveless costume has now changed to a coat. Now, I think it's because it says wear suit and followed the color of the suit. So, when you use the control net, it's important to find the optimal value for yourself by adjusting the intensity to some extent yourself. And because this control mode is also balanced, it's a balance between the control net and the prompt. If I chose one of these two, another result could come out. So, there are quite a lot of variables, so it's most important for you to touch this and that and find the right value for yourself. So, today, we learned about Kenny, one of the functions of the control net. Through the extraction of Kenny's lines like this, you can use it a lot when you do a lot of work such as compositing. If you master this function well through practice, it can be quite helpful when making products or drawing illustrations. I think it's important to master it to suit your style by repeating the content I explained today. So, that's it for today's lecture. I'll come back next time with more useful content. Thank you.
