Hello, today I heard that there is someone who makes more than 500,000 won a month by creating AI art, so I wanted to check it out. I’m really curious if it's possible to generate automated income of 500,000 won today by creating art with AI. 

Hello, what are you doing right now? I'm currently working on AI art. I’m curious about how you can automate the monetization of AI art and actually make a profit. Is that really possible? 

So, what’s the principle behind generating automated income with AI art? We upload the AI-generated artworks to foreign sites, and once we create them, they keep selling and generating revenue continuously. 

So, besides AI, once you register on that site and start making a profit, how do you receive the money? Do you have to go through a complicated process with PayPal or other services, or can you directly receive it in your bank account, like Kookmin Bank?

You can receive money directly to your Korean bank account from our bank account. So, there's no need to go through a complicated process; you just need to register your Korean account. Today, we will go through how to draw, how to install, and how to register on the sites where you can sell your work. Since we have a computer room, we can practice together there. Let's go.

What is the principle behind this automated monetization? We will create a portfolio using AI. Once we create the portfolio, we will upload it to an overseas site. Once it's uploaded, it will be sold automatically, and we will use free tools for the entire process. So, if we upload one portfolio to the site, you can see that you have earned over $400 within a month. You can confirm that the $400 has come in, and now...

In the middle of it, PayPal didn't fix this and directly received deposits into a Korean bank account. You'll even practice registering your account today. You can learn this in a flash right now. You should be able to implement it within an hour, and we'll even create a portfolio and upload those files, all within that hour. Let's start with the installation method for the program. So, I hope you can follow along well today. Everything is installed for free. We'll do it step by step together. First, we'll install a program called Stable Diffuser. As mentioned before, Stable Diffuser is a free program, but the installation can be a bit tricky. I'll show you how to install it. First, search for GitHub. When you do, the GitHub site will appear. So, what is GitHub?

You can think of it as Google for developers; it's like a search bar. Click on GitHub, and then this will appear. First, you need to sign up for GitHub, so just click on the sign-up button. When you click sign up, does it show something? Oh, it worked! Now that you've been verified, you need to create a password that includes numbers and special characters. Once you do that, click continue, and then you can use your username. Next, you’ll be asked if you want to receive updates and other information via email. You can select yes, but it’s also fine to choose no. You’ll need to complete a quiz; the direction your hand points will indicate the animal. After receiving the email, you’ve successfully signed up. Now, you need to go to GitHub. So, how do you get the stable version? For that, you log in and click on the side button. Please sign in now.

Next, you can search in this square box. If you type Stable Diffuser, you'll find a person named Automatic 111. It’s recommended to download the files uploaded by this user. So, when you click, you will see the files listed. You can download all these files and proceed with the installation. To find the installation method, scroll down to Installation and Running, and just follow the instructions provided there. If you use Google Translate, it will show up in Korean. So, click on Translate to Korean for installation and execution. Once you follow the installation steps as directed, you will be installing the Stable program, right? After completing the installation, there will be some information about the computer specifications and other details.

Is it important? Yes, it is. If you look here, it says that it follows the guidelines for NVIDIA recommended and all AMD GPUs. Basically, when using it, you should think of it as heavily relying on the 2070 graphics card. Now that the installation is complete, what should you do next? When you go into your computer, you'll find a new item called Stable in the C drive. If you enter Stable, you'll find a file with a gear icon labeled Web UI User in the Stable Diffusion section. You need to run this file. When you double-click it, a black screen will appear. You might not understand much about this, so just wait. It will start running right away. You can think of it as the Stable Diffusion program being executed. If you just wait quietly, it will continue to run.

This is how it works. Depending on the computer, it can take different amounts of time. For slower computers, it may take about a minute, while faster ones might only take around 20 seconds. When you first install it, it will take about 3 to 5 minutes. Once you run it, you'll see Stable Diffusion displayed, and the website address will appear as well. This address is the same as the one at the end of the URL. If you copy it from Google using Ctrl+C and paste it with Ctrl+V, it will show up like this. In this case, if you just run the program, it will automatically create images for you. Yes, that's right. This is essentially the basic UI. Depending on the commands you enter, images will be generated. So, if you have a specific image you want to create, you can type it in English. For example, if you want to create an image of a woman, you would type a girl.

You just need to press the generator. When you press the generator, a picture will be created like this. So, a picture has been generated here, but if there are any parts that feel lacking or something you want to improve, you can input that. For example, if you want to change the hair color, face shape, or background, you can add those requests. Let's say white shirt. So, if it's a girl in a white shirt and you hit generate, a girl wearing a white shirt will appear. For the command, you just need to press the comma. Yes, that's right. Pressing the comma is an important point. Now, you see, we created it like this, but ultimately, it has to be generated as an image that is in demand on that overseas site we mentioned earlier to be sold.

So, you need to let me know what is in demand. Let's take a look at an actual website where sales can be made. If you search for upload, the first site that comes up has upload written in English. You can think of this site as a place where you can ultimately sell your artwork. Now, if you click on discover, a search window will pop up, showing various products that other people are selling. They have designs and 3D items for sale, but let's try searching for our artwork. I'll search for web portal. It doesn't come up right away. However, if I try again, it will show up. So, the first web magazine comes up, and it appears like this. This is an item that has actually been sold. The image has been sold, and it seems like it automatically generated some income. This means that the earnings have come in. So, this person is now selling something that does not actually exist.

It's a person who created a stable program. Yes, that's correct. If you look at the selling price now, it is set at $5. Is this the price for just one piece, or is it for a collection? Currently, you can see that it's being sold as an AI art collection, bundled with about 50 pieces for around 6,000 won, which is approximately $5. The artworks, however, don't necessarily have to be 50 pieces. It is somewhat referred to as a photobook, but selling it at a missionary market might not be appropriate. Bundling a few works together would enhance its marketability. You can input it in Korean, English, or Japanese because there will be diverse demands. Koreans will also buy it, so you can use Papago for translation. Yes, when you upload it like this, the program you just showed appears quite juvenile. However, here it shows that there are artworks with some degree of maturity.

A photo of a woman has appeared, right? Usually, this means that there is a higher demand for AI-generated images of young women. That's correct; we believe we are adapting to this demand. We generate models that can be used in general photo books or as fashion models, dressing them up and publishing them in photo collections. Now, to achieve the level of transparency we talked about earlier, I would like to show you how we adjust for age and other factors. Currently, the quality is a bit lower than what we previously showed. One way to match these elements is to consider the differences between the painting and the photograph. What do you notice? First of all, the landscape is not visible. When it comes to backgrounds, if you want to add certain elements, you can input keywords, and suggestions will appear like this.

It's also good to include recommended keywords. For example, we have simple and white. The numbers next to them indicate how frequently they have been used. Since simple background has been used quite a bit, let's include that. Next, the quality is a bit low, so let's try to include a higher quality option. We can go with best quality, which also appears. You can use tags like best quality or high quality. However, if you look below, you'll see low quality and worst quality. These serve as negative prompts, meaning we can command to create such qualities but instruct not to create these types. The command for low quality indicates that this has been used frequently. It shows that we can specify such options.

It seems that you can give a hint. This is how it is generated. If you look now, the eyes are slightly uneven. We need to work on these details to make adjustments. Could you try to improve the quality a bit? Even now, it seems to look too youthful, so we need to align the age range with the quality. Although I'm showing this to Jeong Yak-yong, I've organized the quality and prompts like this. The quality is determined based on how we write it. 

These prompts are positive commands, asking for a good outcome, while the ones below are negative commands. Yes, that's right. So now, the representative's commands above and below are being input, and that's how we got this result.

Could you please press it once? The quality is almost at the level where you could call it a photograph. Earlier, we were selling something on the internet that had a similar quality. I don't know what this command is, but it's an extremely important command. It would be good to extract such commands and save them, adjusting the prompt commands based on the well-produced results. That's right. So, you can just copy these and paste them directly. Realistic, just actual, dramatic quality. A 25-year-old Korean woman, a single girl in a cowboy shot, which you can consider a full-body shot. She's wearing a T-shirt, a shirt, and shorts. Then, for the likeness of Laura Korean doll, I inputted facial data during the learning phase. This is the result that came out. And for the negative aspect, I inputted, Please do not produce something like this, referring to drawings or something abstract like that.

This content is related to requests for removing elements in a comic format. It also needs to address improvements in image quality. Additionally, we should avoid anything that appears faint or has falling quality. Now, regarding NSFW, this is an important topic; it refers to explicit content. To regulate the level of explicitness, think of it as a warning for NSFW. Therefore, it's important to emphasize that this should be avoided. When highlighting things that are truly disliked, it’s best to place them at the beginning and use double parentheses to stress them.

There are often cases where bad hands appear strangely, so we should remove those. We should also specify that there should not be two women appearing. I will briefly explain these details. It mentions a sampling method, which indicates a certain style of the author.

You can think of it in terms of the illustration style. The sampling steps can be considered as a way to increase the detail. However, increasing them will also prolong the time it takes to generate the images. You can set it to 100, 200, or even 15, but the optimal range is between 20 and 34. Depending on the quality you want to achieve, you can increase the number. As the numbers go up, the quality improves. For our photo book, we used a size of 512x768 and made the height a bit longer. Regarding upscalers, it's about how to enhance the image quality, and you can generally set it to a fixed value. This is currently the most optimized size based on our research. The images displayed on the screen represent the level of results you can achieve.

You can think of the adjustments below as a kind of post-processing. Depending on this, you can make various adjustments, but it's best to upload the exact settings I captured and shared with you. As for downloading AI photos, you can find the option to save here. If images have already been created, there's a folder icon that says Open Image Output Directory. If you click on this, you can see the photos taken today, July 31, 2023. So, in this way, you can download the images as PNG files. Currently, everything is set to PNG format. You can download them as PNG files, compress them, and then upload them to the sales site. That's right. How can you download the files? If you want to get the Laura files, you can search for 12 types here.

There is a Stable Diffusion model available for download. As you can see, many images come up. So, what should you search for? Here, you will see the LoRA. Our current model is version 1.5, which you can think of as the most stable option. If you click on it, it will filter the results. Depending on the type of image style or facial style you want, you can find different LoRAs. Let me show you how to download it. When you click, the LoRA will be available for download. You need to download it and place it in the Stable Diffusion folder, right? Yes, that's correct. If you look behind the LoRA, you will see that the file extension is safe or something similar. This indicates that it is a secure file name. So, do we need to place this file name in the folder where we installed it earlier?

Once you go into the computer and install Stable Diffusion on the C drive, you can access the web UI from the installed software. In the models section, you'll find LoRA. If you go into LoRA, you'll see that I just had the Korean doll lightness model. You need to input this model here. While doing this, you can transfer it over. Then, if you go back to Stable Diffusion, you’ll be able to select those settings. Yes, that's correct; it has been applied immediately. If you click here, the data values from the model you just received will be input right away. Since we have generated a lot, if we have more than 50 points, we can compress these and upload them. Yes, that's right. When we compress, we usually create a new folder here. Let's create a new folder and name it Web Gallery. Then, we can put in 50 photos that we like from here.

I have created a compressed folder titled Photo Album Part 1 here. This is how it turned out. Now that the folder is created, I will place it on the desktop. I will also upload the cover photo we created, which is the main representative cover. Let's name it Cover. 

Now, regarding the upload, we registered earlier, right? After registering, what happens? Once we log in, the dashboard appears. Please click on the dashboard. Next, we should go to Creator, right? Yes, here, we can click on Create My Work. 

Next, we need to enter a name. I will write Web Butter 1. I will select Digital Product. Since this is just a test, I am doing it like this for now, but for the name, we can use more diverse and creative names.

If you write down the name of the work based on what you want to express or emphasize, that would be great. Now, let's click on the digital product. After clicking, we can set the price. I think we should set it around 5 to 10 dollars. Let’s set that and then click on 'Next' to customize. 

Here, we have the T-shirt and the photo. Next, I will write a description. This one will be for women's wear. The URL will be generated. In fact, it would be good to write it in various languages like Korean, English, and Japanese, just like you did when selling earlier. But since this is just a test, we can keep it simple for now. Ultimately, there should be diverse content included here. I will write something roughly for now. 

You can also write in both English and Korean together. This can be done well using Pop. Next,

I will upload a cover photo. Click on the computer file and select the one that appears first, the one that stands out the most, and we have saved it on the desktop. The cover will be uploaded, and it should also allow for additional content below. Now, it is set up like this. You will need to work on the thumbnail as well. Here, the detailed information is provided. It mentions a size of 600 by 600 for jpg or png files, or you can decide on the details from here. Typically, this is referred to as a call to action, which is a box that consumers click on before making a purchase. You can use a variety of options for this; it doesn't need to be very specific. So, please double-check the price and the version, and make sure everything is correct. If there are no major issues, you can go ahead and click save and continue. Next, we will look at the content and discuss what to include. Earlier, we compressed it.

해놨죠. 그럼 여기다가 구매했을 때 다운로드 가능한 콘텐츠를 넣는 거네요. 맞습니다. 그래서 인서트 파일을 해서 파일을 넣어줍니다. 컴퓨터 바탕화면에서 우리가 압축해놨던 거 있죠? 여기 웹 화보집 1편을 더블클릭해주겠습니다. 아, 이러면 이제 결제한 분들이 받을 수 있는 압축파일이 올라가네요. 네, 맞습니다. 그리고 바로 퍼블리셔에서 발행하고 그대로 이어갑니다. 네, 그러면 실제 프리뷰도 다 나와 있고, URL을 복사한다고 하면 판매 사이트들을 공유할 수도 있겠죠. 그래서 카테고리도 정해줍니다. 카테고리에서는 우리가 포토, 포토 관련된 거죠. 포토를 치니까 사진에서는 코스프레가 뭐가 돼 있지만, 저는 코스프레론 태그에서는 어떤 게 있을까요? 웹 포토, 웹 포토 한번 쳐볼게요. 어, 왜...

There is a photo book. Then, let's click on the web photo book and add tags. You can add multiple packs, right? Yes, you can add several. Now, the changes have been saved. Let's check if the things we've published are showing up correctly. Here, it shows that it has been published properly. If we go to the current status of the prototype, we can see what we just created. 

Here is the phone number. Let's confirm the account number once more. Below, there is a section for individuals and corporations, but since I am just an individual, I can select individual. Yes, most people are individuals, so please click on that. Now, you can write your name in English and fill in the required fields. There are sites available for that, so you can use those. 

Yes, that's right. And once registered, you can receive the profits directly to your Korean bank account without having to change anything to PayPal or Payoneer. That's correct. However, there is a fee listed here: a 10% upload fee plus an additional 2.9%.

Then, the card fees are added, so you can think of it as about 15% being paid as a commission. Ultimately, creating each item consumes time. However, if we do this, we need to use it as automatic income, otherwise, the merit decreases. If we raise the prices, we have to keep raising them. So, if we have something like A that is automatically sold, we need to keep creating more like A, B, C, D, E, F, G. Is there a way to make this process more efficient in the future? Yes, there is. You could consider it a unique feature of Stable Diffuser. We've been using it consistently, and when you generate images, if you right-click, it says Generator Forever. This means it will keep generating as long as the computer is on. So, before going to bed...

You can just set this up and let the generator run while you sleep, and it will keep producing. That's right. The setting we practiced earlier took about 20 seconds to create one photo. If we consider that, we can produce about three images in a minute, which means we can generate around 180 images in an hour. From those, we can select the pretty photos, edit them, and create a compressed file, which would result in a photo book. So, by continuously uploading these photo books, we could summarize a kind of pipeline. Exactly, since products are continuously being produced, if we can maintain a quality that satisfies buyers, repeat purchases can happen. In some ways, this method is more active abroad, but it hasn't been widely used domestically. The ability to generate income using AI art is something you've just pointed out.

Anyone can give it a try through this method. That's right. I know that you're busy today, but thank you for sharing such valuable information and precious methodologies with us. So, what I learned today is about Stable Diffusion version 1.5, correct? Yes, that's right. And what about the latest version? It doesn't have a version number, but the official name has changed. You can think of it as Stable Diffusion, maybe around version 3.0. It can produce much higher quality results compared to 1.5, right? Yes, that's correct. For those who are somewhat satisfied, they can definitely achieve good results with the version 1.5 that we learned today. That's right. Using free tools like Black and White AI, it actually feels like a half-hearted attempt, but once you try it, it becomes clear that this is...

I think there is definitely a sense of realism in this. By using tools that automatically generate content, we can select the best options from them, compress the files, and then sell them in a way that allows for automatic downloads. This seems like a pipeline that anyone can try out. 

As for my rating, I would give it about 4.0 out of 5 stars. There is a bit of a learning curve when installing the diffusion program. Once you complete the installation and set up the prompts perfectly, it becomes quite easy from there. The initial setup can be a bit tough, which is why I deducted 0.8 points. 

Today, I will input the prompts that were shared, extract the content, and try uploading it. Since it doesn't cost anything just to set it up and upload, I believe it deserves a rating of around 4.2.

I think this is not a side job that can be taken on with the level of difficulty. I’m really tired right now, so I feel like I need to go rest soon. Thank you for your hard work. I will be going now.

[수익 인증] 부업인데 일 안하고 월 50만원 자동으로 번다고!? 못믿어서 찾아갔더니 새로운 방법 100% 다 알려줌 ㄷㄷ (무료 AI로 새 부수익 만들기)ㅣ검증왕

"가난하게 태어났다고, 평생 가난해야 하는 건 아닙니다. 과거가 불행했다고, 미래까지 불행할 순 없습니다. 우리에게 주어진 현실을 냉정하게 인식하고, 삶이 조금이라도 나아질 수 있도록, 아주 작은 것이라도 실천해 봅시다. 한 번에 대박을 내겠다는 신기루를 좇지 말고, 매일의 꾸준함으로 단단하게 쌓아 올리는 삶을 살아갑시다. 저도, 구독자님들도 모두 행복해질 권리가 있습니다. 우리, 행복 합시다!" by 절약왕 정약용

하나,
누구나 실천 가능한 생활 재테크 정보를
쉽고 재밌게, 숨김 없이 100% 알려드릴게요.

둘,
다양한 스승들이 알려준 자기계발 가르침 중
제 삶에 실제로 적용해, 검증이 끝난 내용만
알기 쉽게 들려드릴게요.

셋,
하루라도 빨리 알았으면 좋았을,
동기부여에 도움되는
유익한 이야기를 들려드릴게요.

넷,
다양한 분야의 재테크 고수를
초대하여, 구독자님들께
도움되는 정보를 제공할게요.

재미있고 진솔한 꿈, 재테크, 자기계발, 동기부여 유튜버 정약용의 이야기를 통해, 우리가 함께 대화를 나누고 모두가 성공했으면 좋겠습니다 ^^

비즈니스 관련 문의 : isadhappy@us-all.co.kr
출연 신청 / 사연 제보 : starkimyh57@naver.com

***중요 공지사항***
[저작권 고지] 절약왕 정약용 채널이 제공하는 영상, 이미지, 텍스트 등 각종 콘텐츠는 저작권, 퍼블리시티권 등 관련 법의 보호를 받습니다. 절약왕 정약용 콘텐츠를 사전 허락없이 무단 복사, 개작, 전재, 배포, 판매할 경우 민, 형사상의 책임이 따를 수 있다는 걸 알려 드립니다. 절약왕 정약용 콘텐츠 사용에 대해서는 이메일로 문의하시기 바랍니다

절약왕 정약용은 종목 추천, 주식 리딩, 카톡방, 텔레그램방 등을 운영하지 않습니다. 모두 사칭방이니 유의해주세요!

제가 나누는 모든 이야기들은.. 베풀기 위해 아껴 썼던 정약용 선생님에게 영감을 받아 시작되었습니다.

"...절약해서 쓰는 것이야말로, 즐겁게 베푸는 것의 근본이다"
by 정약용 (목민심서 中)