[00:00] In this video, I’ll show you how to use ComfyUI&nbsp; step by step. ComfyUI is one of the most&nbsp;&nbsp; popular free and open-source tools for creating&nbsp; AI-generated content on your own computer. People&nbsp;&nbsp;
[00:12] use it to generate images, create videos, enhance&nbsp; photos, and build powerful AI workflows using&nbsp;&nbsp; a visual node-based interface. In this tutorial,&nbsp; we'll start from the very beginning. We'll install&nbsp;&nbsp; ComfyUI, set up an image model, generate our first&nbsp; AI image, and then learn how workflows and nodes&nbsp;&nbsp;
[00:29] work so you can start building your own projects.&nbsp; I'm Kevin, and let's dive in. To get started,&nbsp;&nbsp; head to the following website. You'll find a link&nbsp; right here at the bottom of the screen. Here on&nbsp;&nbsp; the homepage, you'll notice that ComfyUI offers&nbsp; both a desktop version and cloud version. For this&nbsp;&nbsp;
[00:46] tutorial, we'll use the desktop version since it's&nbsp; free to use, runs directly on your own computer,&nbsp;&nbsp; and gives you full access to your models and&nbsp; workflows. The cloud version can be a good option&nbsp;&nbsp; if you don't have a powerful computer, or maybe&nbsp; you'd rather not manage the hardware yourself,&nbsp;&nbsp;
[01:01] although there is a cost associated with it. Let's&nbsp; click on download and then download again. Once&nbsp;&nbsp; the installer finishes downloading, run through&nbsp; the install process. After installation completes,&nbsp;&nbsp;
[01:13] launch ComfyUI, and you'll see the following&nbsp; screen. For this tutorial, we'll select local,&nbsp;&nbsp; so everything runs directly on your own&nbsp; computer. Down below, I'll agree to the terms,&nbsp;&nbsp; and then let's click on continue. Next, let's&nbsp; give this installation a name. I'll set mine&nbsp;&nbsp;
[01:27] to ComfyUI. For most people, you can leave the&nbsp; remaining settings as is, and then down below,&nbsp;&nbsp; let's click on continue. Once ComfyUI launches,&nbsp; you'll land on the main getting started screen.&nbsp;&nbsp; Now, right away, you can see that ComfyUI supports&nbsp; much more than just the image generation. Over&nbsp;&nbsp;
[01:42] here on the left-hand side, you'll find categories&nbsp; for images. Here we have videos, audio, 3D models,&nbsp;&nbsp; large language models, and more. To help you&nbsp; get up and running quickly, let's go back to the&nbsp;&nbsp;
[01:56] getting started section, and let's start with text&nbsp; to image. I'll click here. ComfyUI now loads a&nbsp;&nbsp; complete image generation workflow. Now, notice in&nbsp; the top right-hand corner that ComfyUI is telling&nbsp;&nbsp; us that we're missing a required model. Think of&nbsp; ComfyUI as the application that allows us to build&nbsp;&nbsp;
[02:13] and run AI workflows. The model itself is what&nbsp; actually creates the image. To see what we need,&nbsp;&nbsp; let's click on show missing models. Right here,&nbsp; ComfyUI shows us the exact model required by this&nbsp;&nbsp;
[02:26] workflow. To install the model, right over here,&nbsp; let's click on download all. Right up on top,&nbsp;&nbsp; we can now check on the download progress.&nbsp; Now that the model has finished downloading,&nbsp;&nbsp; let's generate our first image. Let's take a&nbsp; quick look at the workflow. You can zoom in&nbsp;&nbsp;
[02:39] and out using your mouse wheel, and you can press&nbsp; the middle mouse button to pan around the canvas.&nbsp;&nbsp; You can also use the mini map in the bottom&nbsp; right-hand corner to quickly navigate around&nbsp;&nbsp; larger workflows. Now, notice that this workflow&nbsp; is divided into several different sections. Right&nbsp;&nbsp;
[02:56] over here, we start by loading the model that&nbsp; we just downloaded. Next, we enter a prompt,&nbsp;&nbsp; and this describes the image that we'd like to&nbsp; create. Now, I want to generate an image of a&nbsp;&nbsp; chocolate chip cookie, so right over here,&nbsp; I'll enter in the following prompt. Now,&nbsp;&nbsp;
[03:11] below that we have the negative prompt. This&nbsp; tells the model what we'd like to avoid in the&nbsp;&nbsp; generated image. So, over here, I'll enter&nbsp; in my text. If you want to follow along,&nbsp;&nbsp; you'll find these same exact prompts in the&nbsp; description of this video. Next, we can configure&nbsp;&nbsp;
[03:25] the image size. Here we could set the width, we&nbsp; could also set the height, and also the number of&nbsp;&nbsp; images we'd like to generate. Now I'll leave the&nbsp; default settings in place since they work well&nbsp;&nbsp; with this model. Next, let's go up and we'll look&nbsp; at the sampling section. This is where ComfyUI&nbsp;&nbsp;
[03:40] generates the image. You'll notice several&nbsp; different settings here. Now, as a beginner,&nbsp;&nbsp; I recommend leaving these at their default&nbsp; values. Now there is one setting worth knowing,&nbsp;&nbsp; and that's the seed. You'll find that right here&nbsp; at the top of the list. Changing the seed allows&nbsp;&nbsp;
[03:55] you to generate different variations of the same&nbsp; image all while using the same prompt. For now,&nbsp;&nbsp; let's leave all of these settings as is. Then, in&nbsp; the top right-hand corner, let's now click on Run.&nbsp;&nbsp; And there we have our first AI generated image.&nbsp; It also automatically saved the image to your&nbsp;&nbsp;
[04:11] computer. Now, you might be wondering how do you&nbsp; view all of your generated images or assets? Over&nbsp;&nbsp; on the left-hand side, we have this button for&nbsp; assets. When I hover over, you'll also notice that&nbsp;&nbsp; there is an associated shortcut key. You can press&nbsp; the letter A. When I click on that, here I can see&nbsp;&nbsp;
[04:26] all the different images that I've generated.&nbsp; If I right click, over here, I could save it&nbsp;&nbsp; elsewhere on my computer or down below, I could&nbsp; also delete the asset if I'd like. Right above,&nbsp;&nbsp; I'll close this view. Now that we've successfully&nbsp; generated an image, let's take a closer look at&nbsp;&nbsp;
[04:40] how this workflow works and how we could modify&nbsp; it to produce a different result. Now, right&nbsp;&nbsp; now we're generating just a chocolate chip cookie.&nbsp; Let's see what happens when we make a few changes.&nbsp;&nbsp; So right over here, we have the prompt field,&nbsp; and let's update this text to instead create,&nbsp;&nbsp;
[04:55] let's say, a double chocolate chip cookie with a&nbsp; melted chocolate center. That sounds so good. Now,&nbsp;&nbsp; if we scroll down just a little bit, here&nbsp; we have the image size. And previously,&nbsp;&nbsp; we went with a 512 pixel by 512 pixel image, and&nbsp; that produced a square image. But let's actually&nbsp;&nbsp;
[05:11] try a different aspect ratio. So right over&nbsp; here, let's change the width to, let's say 768. I'll enter in that value. And then right up on&nbsp; top, let's try running this. I'll click here.&nbsp;&nbsp;
[05:24] And there we have a double chocolate chip cookie,&nbsp; and now it's also a wider image. Now that we've&nbsp;&nbsp; generated a few different images, let's take a&nbsp; closer look at how this workflow is organized.&nbsp;&nbsp; One of the things that makes ComfyUI unique&nbsp; is that the workflows are built using nodes.&nbsp;&nbsp;
[05:40] Each node performs a specific task. Here's an&nbsp; example. We have our first node, and that loads&nbsp;&nbsp; the checkpoint or the model that we downloaded.&nbsp; And together, all of these different nodes create&nbsp;&nbsp; the workflow that generates this final image. Now,&nbsp; as you look across this canvas, notice how all of&nbsp;&nbsp;
[05:58] these different nodes are connected together with&nbsp; all of these different lines. Data flows from one&nbsp;&nbsp; node to the next until we eventually arrive at&nbsp; the finished image. Now, the nice thing about&nbsp;&nbsp; this approach is that we're not limited to the&nbsp; workflow exactly as it exists today. We could&nbsp;&nbsp;
[06:14] easily extend it by adding new nodes. I'll zoom&nbsp; in right here on this last stage. To add a node,&nbsp;&nbsp; the option to add a node. I'll click on that, and&nbsp; that opens up a menu with hundreds of different&nbsp;&nbsp;
[06:30] nodes that you can use to customize your workflow.&nbsp; Now, I'd like to add an image upscaling node. So&nbsp;&nbsp; here within the image at the very top of the list,&nbsp; we have the option for upscaling. And right here,&nbsp;&nbsp; we can upscale the image by. I'll click on that.&nbsp; Right over here, it's now added a new upscaling&nbsp;&nbsp;
[06:45] node. And let's try upscaling the image, let's say&nbsp; by 2X. Here I'll enter 2 and click on OK. However,&nbsp;&nbsp; right now, this node is not connected to the&nbsp; overall workflow. So over here, let's take the&nbsp;&nbsp;
[06:57] output of the decoding node, and here I'll feed&nbsp; it into the upscaling step. And over here, I'll&nbsp;&nbsp; take the output of the upscale step, and let's&nbsp; feed that into the save image step. That way,&nbsp;&nbsp; we'll upscale the image before we end up saving&nbsp; it. Now, let's run the workflow again. In the top&nbsp;&nbsp;
[07:12] right, I'll click on run. And right there, you&nbsp; can see that we've now enhanced the image simply&nbsp;&nbsp; by adding a single node to the workflow. Right&nbsp; down below, you'll now see the new dimensions&nbsp;&nbsp; of this image. We've effectively upscaled it&nbsp; by 2. This is the core idea behind ComfyUI. By&nbsp;&nbsp;
[07:27] combining and connecting different nodes, you can&nbsp; build workflows that generate images, edit photos,&nbsp;&nbsp; create videos, process audio, and much more.&nbsp; Now that we've customized this workflow, let's&nbsp;&nbsp; save it so we can use it again later. Right up on&nbsp; top, you'll see all of your different workflows.&nbsp;&nbsp;
[07:42] I'll right click on this one, and right here,&nbsp; we see the option to save. I'll click on this.&nbsp;&nbsp; Right here, we can enter in a file name. I'll call&nbsp; mine the upscale image generator, and down below,&nbsp;&nbsp; let's click on confirm. To get back to workflow,&nbsp; over on the left-hand side, click on workflows,&nbsp;&nbsp;
[07:56] and there you can see the workflow that we just&nbsp; saved. You can also share workflows with other&nbsp;&nbsp; people, which is one of the reasons ComfyUI has&nbsp; such a large and active community. Instead of&nbsp;&nbsp; building everything from scratch, you can start&nbsp; from an existing workflow and modify to fit your&nbsp;&nbsp;
[08:11] exact needs. Here, I'll close out of workflows.&nbsp; Before we wrap up, let's take one more look at&nbsp;&nbsp; One of the things I like about ComfyUI is&nbsp; that you have a large collection of starter&nbsp;&nbsp;
[08:23] templates to get you up and running quickly. Now,&nbsp; in addition to text to image, which we just used,&nbsp;&nbsp; here you'll find other fun ones, like image to&nbsp; video. Here, you have image editing. Here we&nbsp;&nbsp; have image to video again, image to 3D. You even&nbsp; have text to song, so lots of different options&nbsp;&nbsp;
[08:38] and fun things that you can play with. Now, I&nbsp; recommend clicking in, downloading the models, and&nbsp;&nbsp; then exploring how they're built. Every workflow&nbsp; here is made up of nodes, and you could use these&nbsp;&nbsp; templates as a starting point and then customize&nbsp; them to fit your own needs. And that's how to use&nbsp;&nbsp;
[08:53] ComfyUI. Now that you understand the basics, go&nbsp; ahead and experiment with the built-in templates&nbsp;&nbsp; and start building workflows of your own. If you'd&nbsp; like to learn more about AI tools and how to get&nbsp;&nbsp; the most out of them, consider subscribing. Thanks&nbsp; for watching, and I'll see you in the next video.