Vultr Cloud GPU에서 Stable Diffusion 3.5 설정하기

### Vultr에서 Stable Diffusion 3.5 설정하기: 단계별 가이드
* 🌟 스테이블 디퓨전 3.5는 사용자가 멋진 이미지를 만들 수 있는 오픈 소스 텍스트-이미지 변환 모델로, 이 가이드는 **Vultr라는 클라우드 GPU 서비스**를 사용하여 쉽게 설정할 수 있도록 도와줍니다.
* 💻 시작하려면 Vultr 계정에 무료로 가입하고 **Burman300 코드를 사용하여 첫 달 동안 $300 크레딧**을 받으세요.
* 🚀 로그인한 후 "제품" 탭으로 이동하여 "클라우드 컴퓨팅"을 선택하여 새 인스턴스를 배포하고 **Nvidia L4s 요금제 및 Ubuntu 운영 체제**를 선택합니다.
* 🔑 클라우드 GPU가 실행되면 제공된 IP 주소와 비밀번호를 사용하여 서버에 SSH로 접속한 후 **ComfyUI 도커 이미지를 다운로드**하여 안정적인 확산 설정을 용이하게 합니다.
* ⚙ 스토리지 폴더를 생성한 후 Docker 이미지를 실행하고 방화벽에서 **포트 8188을 허용**하여 ComfyUI 인터페이스에 액세스할 수 있도록 합니다.
* 🔑 허깅 페이스 CLI를 설치하고 생성된 토큰으로 로그인하면 **필요한 모델 파일을 다운로드**할 수 있으며, **파일 크기에 따라 몇 분 정도 소요**될 수 있습니다.
* 🌐 브라우저에서 서버의 IP 주소 뒤에 **":8188"을 입력하여 ComfyUI에 액세스**하면 이미지 생성을 위한 인터페이스가 준비됩니다.
* 🎨 "불독이 사람 쓰기" 또는 "용 날개를 가진 고양이"와 같은 다양한 프롬프트와 워크플로우를 실험해보고 **Vultr의 강력한 GPU가 제공하는 빠르고 인상적인 결과물**을 즐겨보세요.

안정적인 Diffusion 3.5를 통해 사용자는 Vultr의 클라우드 GPU 서비스를 통해 멋진 이미지를 쉽게 제작할 수 있습니다.

Stable Diffusion 3.5 is an incredible open-source text-to-image model, and I'm going to show you how to set it up and use it today. You can create incredible images like this one, this one, this one, and this one. This is completely

open source and free, so you can download it on your local computer. But what we're going to do today is get it set up using a cloud GPU service called Vultr. If you've watched my channel for any period of time, you know I love Vultr; they've

sponsored a bunch of different videos, and they're partnering with me on this video as well. You can follow many of these steps to get it set up locally, but if you want it done easier and you don't have a beefy GPU to power it, this is a

great and inexpensive alternative. So, if you don't already have a Vultr account, go ahead and sign up; it is free to sign up. And remember, during sign-up, use the code Burman300 to get $300 of credits for free during your first month of use. So,

log in, go to the left side, click on this "Products" tab. You're going to come to "Cloud Compute" right at the top here. So, on the right side, click "Deploy," and then you're going to deploy a new instance. Choose type; we're going to click "Cloud

GPU." You can select whichever location makes sense for you, whichever one's closest to you. So we're going to select the Ubuntu operating system right there, just like that. Scroll down; we're going to choose our plan. The Nvidia L4s is sufficient for what we need today for

Stable Diffusion 3.5. We're going to select one; that's all we need. So the next thing is we're going to deselect auto backups because we don't need that. So go ahead, click there. I understand the risk, and we are ready to

go. We click 'Deploy Now'. So just wait till it's done installing, and then we'll go to our next step. All right, so now you can see our new Cloud GPU is up and running, with the green play button status right there. So we click on the

three dots right here. We go to 'Server Details', and what we're going to be doing next is SSHing into the server. Next, open up your terminal, and we're going to do all of this through Docker. So you're going to come down here; you're going to

click the 'Copy IP Address' button right next to your IP address. So you're going to do `ssh root@` and then the IP address, and then hit Enter. And then when you're asked for the password, go back to Vultr, and you'll find your password

right here. You click 'Copy Password', and then you just paste it in and hit Enter. And there we go; we're in. So we can see `root@vultr`. Now that we're SSH'd in, we are going to download the Docker image. So we're going to do `docker pull`

and then we're going to download the ComfyUI Docker image. And you don't have to remember any of this; I'll drop all of the instructions down below. So we hit Enter, and then we can see I've already downloaded it, so I won't need to do it.

Again, but you'll need to go through that download process. Now we're going to make a new folder for storage, so `mkdir DP_storage`; hit Enter. And now we have to run the Docker image, so we're going to run this command again. I'll drop all of

these in the description below. So we run it like so; hit Enter. And that's it! Now the Docker image is going to run. Now I just ran `docker logs <docker_ID>` with the Docker ID. And if you need to find the Docker ID, just type `docker ps`; it'll give

you the list of Dockers running. Got the logs; I see it's working. Now we want to allow the port in the firewall, so we go back to root and we type `ufw allow 8188`. There we go. Next, we need to install the Hugging Face CLI, so we're going to type

`pip install 'huggingface_hub[cli]'`; hit Enter, and then it should download just like that. Next, we're going to log in to Hugging Face, so `huggingface-cli login`; hit Enter. At this point, you need

to generate a token, so you can grab the URL from right here: huggingface.co/settings/tokens. Copy it, go into your browser, create a new token just like so. We're going to call it `comfyui-vulture`, and we're going to have it as a read

token. Create token. Let's copy it, and here let's enter it and hit Enter. And I don't need it as a Git credential. Okay, login successful! Now we are going to download the model files just like so; hit Enter, and there it goes. So we can

See, it's downloading right now, and it's relatively fast, but it is multiple gigabytes worth of files, so obviously it's going to take a couple of minutes. All right, and it looks like it's done. So now, go into your server information; you're

going to grab this IP address right there, and you're simply going to put it in your browser and then ":8188" as the port. Hit enter; it's going to direct you to the ComfyUI, and there we go, and it's loading up now. Look at that! Boom! So

we have ComfyUI with Stable Diffusion 3.5 loaded up. Awesome! So there are just a couple of last things we need to do to get this ready. First, we're going to move the files that we just downloaded into the storage folder, into the correct location.

So we're going to use this command right here. Hit enter. Okay, and now we're going to move the text encoders, just like so. Hit enter. All right, so refresh the page, and it should just work out of the box. So right here, where it says "CLIP"

text encoder prompt: "Bulldog writing a person." Then you click "Q prompt," and here is the output. Not quite what we asked for, but still really good-looking. And you can import other workflows as well. So let's try another one: "Cat with dragon wings." Q prompt, and look how fast this is!

I mean, this is a very beefy GPU on Vultr, and yeah, it's super fast. There we go. I mean, it looks fantastic! This looks really good. And like I said, you could try different workflows; you can play around with all of these different settings; you could try Flux Stable

Diffusion, large language models... all of these things are possible now that you know how to set it up. So that's it. Hopefully you enjoyed this. I want to say another thank you to Vulture for being just such an amazing partner. And Vulture right now has a

limited-time promo offer on the GH20, so go to getvulture.com/gh20. It's $2.99 an hour; that is 25% off their normal rate. And like I said, all the commands are going to be down in the description in a gist file. So if you

enjoyed this video, please consider giving it a like and subscribe, and I'll see you in the next one.

Stable Diffusion 3.5 Tutorial (Cloud-Based)

Artificial Intelligence (AI), Open Source, Generative Art, AI Art, Futurism, ChatGPT, Large Language Models (LLM), Machine Learning, Technology, Coding, Tutorials, AI News, and more

** Exclusive Pinecone Partner **

Matthew Berman

Stable Diffusion 3.5 및 Vultr 설정 소개

오픈 소스 텍스트-이미지 변환 모델인 Stable Diffusion 3.5를 소개합니다.

Vultr 클라우드 GPU 서비스를 사용하여 설정합니다(로컬 설정보다 쉽습니다).

강력한 GPU가 없는 분들을 위한 저렴한 대안으로 Vultr을 추천합니다.

할인 코드를 포함한 Vultr 가입 안내.

Vultr에 새 클라우드 GPU 인스턴스를 배포하는 단계.

Ubuntu 운영 체제 및 Nvidia L4s 요금제 선택하기.

자동 백업을 선택 해제하고 인스턴스를 배포합니다.

서버 세부 정보에 액세스하고 서버에 SSH로 로그인합니다.

Vultr의 IP 주소와 비밀번호를 사용하여 SSH를 통해 연결합니다.

컴피유 도커 이미지를 다운로드합니다.

저장소용 새 폴더 만들기(`mkdir DP_storage`).

컴피유 도커 이미지를 실행합니다.

방화벽에서 Docker 로그를 확인하고 포트 8188을 허용합니다.

허깅 페이스 설정 및 모델 다운로드

허깅 페이스 CLI 설치하기(`pip install 'huggingface_hub[cli]'`).

허깅 페이스에 로그인하고 토큰을 생성합니다.

안정적인 확산 3.5 모델 파일 다운로드.

IP 주소와 포트 8188을 사용하여 브라우저를 통해 ComfyUI에 액세스합니다.

다운로드한 파일을 올바른 위치로 옮기기.

예제 프롬프트를 사용하여 ComfyUI를 테스트하고 출력을 관찰합니다.

ComfyUI의 추가 기능 및 워크플로 살펴보기.

Vultr과 기간 한정 프로모션에 감사드립니다.

동영상 설명에서 명령어와 지침을 확인할 수 있습니다.

좋아요를 누르고 구독을 요청하세요.