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 **

Stable Diffusion 3.5 Tutorial (Cloud-Based)

Scripts