Comfyui lora workflow sdxl example This section will introduce the installation of the official version models and the download of workflow files. And finally, SDXL decided to make all of this slightly more fun by introducing two-model architecture instead of one. Download it, rename it to: lcm_lora_sdxl. The IPAdapter Group is optional, you can bypass or active according your needs. In this example I used albedobase-xl. 1 ComfyUI guide d’installation, workflow et exemple. You can Load these images in ComfyUI to get the full workflow. You will first need: Text encoder and VAE: Dec 8, 2024 · At the moment I'm figuring out how to do add this function into my "Negative Prompt & HiRes. SDXL Default Workflow: Best starting point for generating high-resolution images using SDXL. 24 KB. Follow the steps in the diagram below to ensure the workflow runs correctly. 1 Video workflow. This workflow also contains 2 up scaler workflows. com/models/633553 Crystal Style (FLUX + SDXL) https://civitai. It leverages various techniques, including Lora models, ControlNet, IPAdapter, and image processing nodes, to generate and stylize architectural renderings As you can see, by using a LoRA model, we can generate images in different styles without adjusting the base model. 继续使用 LoRA workflow，首先将 Load LoRA 里的模型，切换成 LCM-LoRA。需要注意，这里需要选择跟你的 checkpoint 一致的 LoRA，比如我这用的 checkpoint 是 sdxl，就要使用 sdxl LCM-LoRA。 设置 KSample. Generates dynamic graphs that are literally identical to handcrafted noodle soup. Stable Zero123 est un modèle de diffusion qui, à partir d’une image contenant un objet et un arrière-plan simple, peut générer des images de cet objet sous différents angles. SDXL基础检查点可以在 ComfyUI 中像任何常规检查点一样使用。 唯一重要的是，为了获得最佳性能，分辨率应设置为1024x1024或具有相同像素数量但不同宽高比的其他分辨率。 Download & drop any image from the website into ComfyUI, and ComfyUI will load that image's entire workflow. ThinkDiffusion - SDXL_Default. 5L_example_workflow. My custom nodes felt a little lonely without the other half. safetensors, stable_cascade_inpainting. Simple SDXL Template. Refiner, face fixer, one LoRA, FreeUV2, Style selectors, and basic image adjustment controls. Instantly replace your image's background. Download Stable Diffusion 3. 0? A complete re-write of the custom node extension and the SDXL workflow. This repo contains examples of what is achievable with ComfyUI. 8（一般大于 1. このガイドは、WindowsコンピュータでFlux. 1: ComfyUI LoRA 用ワークフロー詳細. json format, but images do the same thing), which ComfyUI supports as it is - you don't even need custom nodes. That means you just have to refresh after training (and select the LoRA) to test it! In ComfyUI the saved checkpoints contain the full workflow used to generate them so they can be loaded in the UI just like images to get the full workflow that was used to create them. Click the download button to save it locally. 0 Official Offset Example LoRA Nov 18, 2024 · Whether you're new to SDXL or just want a straightforward way to integrate LoRA, this workflow will help you get started quickly and easily. Now in Comfy, from the Img2img workflow, let’s duplicate Load Image and Upscale Image Nodes. 0, it can add more contrast through offset-noise) Aug 20, 2023 · First, download clip_vision_g. 1を実行するためのComfyUIのセットアップ方法について説明します。 How to use LoRA ComfyUI: 1: Download LoRA 2: Upload LoRA 3: Load the Default Workflow in ComfyUI 4: Add the LoRA Node 5: Connect the LoRA Node 6: Test Your Workflow Unlock next-level video creation with Wan 2. Jan 9, 2025 · This workflow is made for Pony so it may need extra modification to work with other models. Instead of creating a workflow from scratch, you can simply download a workflow optimized for SDXL v1. safetensors; Ensure first Load LoRA loads blindbox_V1Mix. 0 model generation. The native workflow has been detailed in the official documentation I wrote for Comfy, titled ComfyUI Native HiDream-I1 Workflow Example. Buy Me A Coffee. So, I just made this workflow ComfyUI. For the workflow to run you need this loras/models: ByteDance/ SDXL-Lightning 8STep Lora Juggernaut XL Detail Tweaker XL. Purpose: Load the main model file; Parameters: Model: hunyuan_video_t2v_720p_bf16. json , the general workflow idea is as follows (I digress: yesterday this workflow was named revision-basic_example. Dec 11, 2024 · This ComfyUI workflow shows how to use the Visual Area Prompt node for regional prompting control. safetensors. Jul 30, 2023 · Low-Rank Adaptation (LoRA) is a method of fine tuning the SDXL model with additional training, and is implemented via a a small “patch” to the model, without having to re-build the model from scratch. You should try to click on each one of those model names in the ControlNet stacker node and choose the path of where your models But captions are just half of the process for LoRA training. The Wan2. It seamlessly combines these components to achieve high-quality inpainting results while preserving image quality across successive iterations. Apr 27, 2024 · This is a small workflow guide on how to generate a dataset of images using ComfyUI. Img2Img works by loading an image like this example image, converting it to latent space with the VAE and then sampling on it with a denoise lower than 1. LCM Examples. Detailed guide on setting up the workspace, loading checkpoints, and conditioning clips. Please try SDXL Workflow Templates if you are new to ComfyUI or SDXL. What it's great for: This What this workflow does. 1 Workflow - Complejo - for Base+Refiner and Upscaling; 1. SD+XL workflows are variants that can use previous generations SD models to provide initial scene concept. In this example, we will learn how to load and use a LoRA model in ComfyUI, covering the following topics: May 19, 2024 · These two files must be placed in the folder I show you in the picture: ComfyUI_windows_portable\ComfyUI\models\ipadapter. Hidden Faces (A workflow to create hidden faces and text) View Now. Complete the Workflow Step by Step. Aug 20, 2023 · In part 1 , we implemented the simplest SDXL Base workflow and generated our first images. Some more use-related details are explained in the workflow itself. Step 3: Download and Import the Workflow File into ComfyUI Download SD1. 3 Workflow - LORA - Including LORA (instead of using the VAE that's embedded in SDXL 1. The only important thing is that for optimal performance the resolution should be set to 1024x1024 or other resolutions with the same amount of pixels but a different aspect ratio. Created by: C. => Place the downloaded lora model in ComfyUI/models/loras/ folder. Works with bare ComfyUI (no custom nodes needed). safetensors and put it in your ComfyUI/models/loras directory. Mar 19, 2025 · 🟢 1️⃣ Workflow Overview. Download it from here, then follow the guide: An All-in-One FluxDev workflow in ComfyUI that combines various techniques for generating images with the FluxDev model, including img-to-img and text-to-img. 3 Workflow - Face - Face restoration/refacing; 1. Starting workflow. It then uses an SDXL checkpoint model with a LoRA and the Fooocus CPDS ControlNet (similar to Depth) to put the text into the image. It allows for the dynamic adjustment of the model's strength through LoRA parameters, facilitating fine-tuned control over the model's behavior. step = 5（一般大于 4 都可以） cfg = 1. This can be useful for systems with limited resources as the refiner takes another 6GB or ram. json which has since been edited to use only one image): May 12, 2025 · Stable Diffusion 3. 1) May 12, 2025 · Flux. Flux Redux is an adapter model specifically designed for generating image variants. I tried both. ” I tried to find a good Inpaint workflow and just found a bunch of wild workflows that wanted a million nodes and had a bunch of different functions. All the images in this page contain metadata which means they can be loaded into ComfyUI with the Load button (or dragged onto the window) to get the full workflow that was used to create the image. The denoise controls the amount of noise added to the image. Notably, the workflow copies and pastes a masked inpainting output, ensuring that image degradation is May 12, 2025 · ComfyUIでのLoRAの使用方法を解説する初心者向けガイド。基本的な単一モデルの使用方法から、複数のLoRAを組み合わせてより豊かなAIアート効果を実現する高度な使用方法まで。モデルのインストールやパラメータ調整に関する詳細な手順を含みます。 SDXL Prompt output examples for inspiration: Simply drag or load a workflow image into ComfyUI! See the "troubleshooting" section if your local install is giving Aug 16, 2023 · ComfyUI LoRA scheduling with hook keyframes nodes; ComfyUI workflow with Visual Area Prompt node; Install missing Python modules and update PyTorch for the LoRa resizing script; Cordova Recaptcha Enterprise plugin demo; Cordova Recaptcha v2 plugin demo; Generate canny, depth, scribble and poses with ComfyUI ControlNet preprocessors Animation: Workflows like AnimateDiff enable the creation and customization of animations in ComfyUI. We will demonstrate how to use a LoRA model. 1 ComfyUI インストールガイド、ワークフローと例. May 12, 2025 · After installation, restart or refresh ComfyUI to let it recognize the new models. It can be used with any SDXL checkpoint model. safetensors, put them in your ComfyUI/models/loras/ folder. SDXL Workflow for ComfyUI with Multi-ControlNet This node is designed to dynamically load and apply LoRA (Low-Rank Adaptation) adjustments to models and CLIP instances based on specified strengths and LoRA file names. Purpose and Function: This workflow is specifically designed for architecture design and rendering tasks using Stable Diffusion XL (SDXL) as the base model. This example is an example of merging 3 different checkpoints using simple block merging where the input, middle and output blocks of the unet can have a The SDXL base checkpoint can be used like any regular checkpoint in ComfyUI open in new window. 0. For these examples I have renamed the files by adding stable_cascade_ in front of the filename for example: stable_cascade_canny. Example workflows for how to run the trainer and do inference with it can be found in /ComfyUI_workflows; Importantly this trainer uses a chatgpt call to cleanup the auto-generated prompts and inject the trainable token, this will only work if you have a . 5 Lora with SDXL, Upscaling Future tutorials planned: Prompting practices, post processing images, batch trickery, networking comfyUI in your home network, Masking and clipseg awesomeness, many more. In this example, one B-LoRA model is used for style and another for content, allowing for more complex and customized results. ControlNet Depth Comfyui workflow: Use ControlNet Depth to enhance your SDXL images: View Now: Animation workflow: A great starting point for using AnimateDiff: View Now: ControlNet workflow: A great starting point for using ControlNet: View Now: Inpainting workflow: A great starting point for using Inpainting: View Now: Using LoRA's: A All LoRA flavours: Lycoris, loha, lokr, locon, etc… are used this way. These are examples demonstrating how to do img2img. 3) This one goes into: ComfyUI_windows_portable\ComfyUI\models\loras. ControlNet (Zoe depth) Advanced SDXL Template Maîtrisez Stable Diffusion avec ComfyUI Wiki! Explorez les tutoriels, les nœuds et les ressources pour améliorer votre expérience avec ComfyUI. env file containing your OPENAI key in the root of the repo dir that contains a single line: OPENAI_API_KEY=your_key_string Everything will Sorry man, workflow ain't perfect :P Most of the cases, you don't get better hands than that. Complete flexible pipeline for Text to Image Lora Controlnet Upscaler After Detailer and Saved Metadata for uploading to popular sites Use the Notes section to learn Aug 30, 2024 · OpenArt AI で公開されている FLUX + LORA (simple) で使用されているノードについて説明します。名前のとおり ComfyUI による FLUX. Study this workflow and notes to understand the basics of ComfyUI, SDXL, and Refiner workflow. Jun 18, 2024 · Load B-LoRA Node; Workflow Examples Single Load B-LoRA Node. Examples. Dec 17, 2023 · ComfyUI-Background-Replacement. json. LCM loras are loras that can be used to convert a regular model to a LCM model. It generates a full dataset with just one click. It can generate high-quality 1024px images in a few steps. They can be used with any SDXL checkpoint model. 1 Model Loading Nodes. 0 with SDXL-ControlNet: Canny Part 7: Fooocus KSampler Custom Node for ComfyUI SDXL 4 days ago · v9 tested on Better Days with Touching Grass, and this PornMaster lora. SDXL_V2_0. Files to Download. Jan 6, 2024 · Introduction to a foundational SDXL workflow in ComfyUI. 5/SDXL. What comfyUI workflows can use some speed up? Of course it’s making videos! So here you go: A ComfyUI AnimateDiff prompt travel workflow sped up by LCM-LoRA! Step 1: Load the workflow. A good place to start if you have no idea how any of this works 1. safetensors Weight Type: default (can choose fp8 type if memory is insufficient) Oct 5, 2024 · The images in the examples folder have been updated to embed the v4. ComfyUI seems to work with the stable-diffusion-xl-base-0. 5 Large Created by: CgTopTips: In this video, you'll see how, with the help of Realism LoRA and Negative Prompt in Flux, you can create more detailed, high-quality, and realistic images. May 12, 2025 · ControlNet et T2I-Adapter - Exemples de workflow ComfyUI Notez que dans ces exemples, l’image brute est directement transmise à l’adaptateur ControlNet/T2I. I used to work with Latent Couple then Regional prompter module for A1111, which allowed me to generate separate regions of an image through masks and guided with ControlNets (for instance, generate several characters using poses derived from a preprocessed picture). Oct 5, 2024 · These workflow templates are intended as multi-purpose templates for use on a wide variety of projects. 5 with ComfyUI. However, since the official documentation currently only supports Chinese and English, I will also provide corresponding examples in this guide, considering the multilingual support of ComfyUI Wiki. Step-by-step guide Step 1: Load workflow. - Ling-APE/ComfyUI-All-in-One-FluxDev-Workflow May 12, 2025 · ComfyUI Workflow Examples This repo contains examples of what is achievable with ComfyUI . It uses the Instant ID IP-adapter model to copy the face and the pose. example of the variants: Installation Steps 1. In each style block, you can adjust the settings to dial in the style Here is an example of 3 characters each with its own pose, outfit, features, and expression : Left : woman wearing full armor, ginger hair, braids hair, hands on hips, serious Middle : girl, princess dress, blonde hair, tiara, jewels, sitting on a throne, blushing 这些是展示如何使用 Loras 的示例。所有的 LoRA 类型：Lycoris、loha、lokr、locon 等 都是以这种方式使用的。 您可以在 ComfyUI 中加载这些图片以获得完整的工作流程。 Loras 是应用在主 MODEL 和 CLIP 模型之上的补丁，因此要使用它们，将它们放在 models/loras 目录中并像这样使用 Lo May 12, 2025 · ComfyUI SDXL Turbo Examples; Resources. Since the workflows are available in the repository of the models introduced earlier, let’s start with those files. The workflow consists of 3 major Group Nodes. 1 workflow. 5,” and then copy your model files into “ComfyUI\models\loras\SD1. I guess this means that your CUDA out of memory issue is unlikely to happen and hence the memory of your GPU is use more efficiently. Ajout d’un sujet au centre inférieur de l’image en ajoutant une autre invite de zone. The prompt for the first couple for example is this: Aug 18, 2023 · Part 1: Stable Diffusion SDXL 1. Thanks!--Reply. 0 Base and Refiners models downloaded and saved in the right place, it should work out of the box. From here, we will explain how to use Stable Diffusion 3. In this guide, I'll use the popular Sytan SDXL workflow and provide a couple of other recommendations. 1 ComfyUI Workflow. Part 2 - we added SDXL-specific conditioning implementation + tested the impact of conditioning parameters on the generated images. Globby J's Simplified SDXL Text to Image + Upscale Workflow. 3B (1. So I gave it already, it is in the examples. Now it’s time to incorporate the LoRA node into your workflow for SD1. json - Refiner, face fixer, one LoRA, FreeUV2 (which I don't like and bypass), Style selectors, Latent upscale (weird), and basic image adjustment controls. For this demonstration, we’ll use the following model: Access and download the model: Squish Effect (Wan2. Most of these changes stemmed from someone mentioning lora clip weights and concat conditioning. 1. Sadly, I can't do anything about it for now. Starts at 1280x720 and generates 3840x2160 out the other end. Feb 21, 2025 · How does this workflow work? This ComfyUI workflow uses an SDXL checkpoint model, with an optional LoRA, to transform a photo into a new style. 0 with ComfyUI Part 2: SDXL with Offset Example LoRA in ComfyUI for Windows Part 3: CLIPSeg with SDXL in ComfyUI Part 4: Two Text Prompts (Text Encoders) in SDXL 1. This basic workflow runs the base SDXL model with some optimization for SDXL. 0 release includes an Official Offset Example LoRA . If you get a message saying something about missing nodes, go to the ComfyUI manager and click "install missing custom nodes" anyways, feel free to ask any questions in the discussion of this page! Enjoy my workflow! :> May 12, 2025 · Img2Img Examples. Layout changes primarily on the left side of the workflow. If you have the SDXL 1. I'm not sure what's wrong here because I don't use the portable version of ComfyUI. No complicated settings—just plug in your SDXL checkpoint, apply your LoRA, and . Links to SDXL 1. Single Load B-LoRA Node. This workflow should work with any SDXL, Pony, Illustrious, and NOOB (NAI) models. 9 fine, but when I try to add in the stable-diffusion-xl-refiner-0. 1 Canny. 0 Workflow. Flux. Loras are patches applied on top of the main MODEL and the CLIP model so to use them put them in the models/loras directory and use the LoraLoader node like this: ComfyUI Examples. 5, SDXL, PonyXL, Flux and any other models. May 12, 2025 · 4. Pinto: About SDXL-Lightning is a lightning-fast text-to-image generation model. A Oct 27, 2024 · Understanding SDXL Model Types SDXL comes in several variants: Base SDXL 1. It is licensed under the Apache 2. 5! 🚀 Join Discord And Contact Admin For 50GB Free! Feb 6, 2024 · ComfyUIを立ち上げて、LoRAを導入する方法をご紹介します！そしてComfyUIでLoRAを使う方法は勿論、複数のLoRAを適用して画像を生成する方法についても解説しています。とっても簡単にLoRAを導入できますので、ぜひ本記事を参考にしてください。 LoRA selector, (for example, download SDXL LoRA example from StabilityAI, put into ComfyUI\models\lora\) VAE selector, (download default VAE from StabilityAI, put into \ComfyUI\models\vae\), just in case in the future there's better VAE or mandatory VAE for some models, use this selector May 12, 2025 · SDXL 示例. 0_fp16. Step 2: Upload to ComfyUI Then move it to the “\ComfyUI\models\controlnet” folder. May 12, 2025 · Learn about the LoraLoaderModelOnly node in ComfyUI, which is designed to load LoRA models without requiring a CLIP model, focusing on enhancing or modifying a given model based on LoRA parameters. The only important thing is that for optimal performance the resolution should be set to 1024x1024 or other resolutions with the same amount of pixels but a different aspect ratio. safetensors; Ensure second Load LoRA loads MoXinV1. The SDXL 1. Models. Support for Controlnet and Revision, up to 5 can be applied together May 12, 2025 · ComfyUI-Hunyuan3DWrapper Hunyuan3D 2. 1 Depth and FLUX. Advanced Merging CosXL. This node can be used with SD 1. May 12, 2025 · If you want do do merges in 32 bit float launch ComfyUI with: —force-fp32. That means you just have to refresh after training (and select the LoRA) to test it! May 12, 2025 · Since models need to be versioned, for your convenience, I suggest renaming the model files with a version prefix like “SD1. May 12, 2025 · Wan2. This workflow perfectly works with 1660 Super 6Gb VRAM. Aug 13, 2023 · The simplest example would be an upscaling workflow where we have to load another upscaling model, give it parameters and incorporate the whole thing into the image generation process. 0 with SDXL-ControlNet: Canny Part 7: Fooocus KSampler Custom Node for ComfyUI SDXL Here is an example for the full canny model: They are also published in lora format that can be applied to the flux dev model: flux1-canny-dev-lora. Some examples done under 3 sec. safetensors; Click Queue or press Ctrl/Cmd + Enter to generate Thanks. This group node consists of the Checkpoint Loader, VAE loader (for if you're not using a checkpoint with baked in VAE), Positive and Negative Prompt Text Boxes, and 4 LoRA Stack (Reduce strength values to 0 to disable that LoRA) Oct 9, 2023 · Crisp and beautiful images with relatively short creation time, easy to use. Here is my way of merging BASE models and applying LORAs to them in non-conflicting way using the ComfyUI (grab the workflow itself in the attachment to this article): A collection of workflow templates for use with Comfy UI These workflow templates are intended as multi-purpose templates for use on a wide variety of projects. Created by: AILab: Lora: Aesthetic (anime) LoRA for FLUX https://civitai. Fix" workflow. ai/workflows/openart/basic-sdxl Oct 5, 2024 · If you don't need LoRA support, separate seeds, CLIP controls, or hires fix - you can just grab basic v1. You don't understand how ComfyUI works? It isn't a script, but a workflow (which is generally in . We name the file “canny-sdxl-1. Community Flux Controlnets Aug 18, 2023 · Part 1: Stable Diffusion SDXL 1. All the images in this repo contain metadata which means they can be loaded into ComfyUI with the Load button (or dragged onto the window) to get the full workflow that was used to create the image. for - SDXL. It can generate variants in a similar style based on the input image without the need for text prompts. It uses a depth ControlNet to fix the character’s pose. LCM Lora. 1 model, open-sourced by Alibaba in February 2025, is a benchmark model in the field of video generation. Introduction of refining steps for detailed and perfected images. Feb 20, 2025 · Ideally, it would happen before the proper image generation, but the means to control that are not yet implemented in ComfyUI, so sometimes it's the last thing the workflow does. Jun 30, 2023 · My research organization received access to SDXL. === How to prompt this workflow === Main Prompt ----- The subject of the image in natural language Example: a cat with a hat in a grass field Secondary Prompt ----- A list of keywords derived from the main prompts, at the end references to artists Example: cat, hat, grass field, style of [artist name] and [artist name] Style and References 💡Education. The SDXL base checkpoint can be used like any regular checkpoint in ComfyUI. It's simple and straight to the point. Ensure Load Checkpoint loads dreamshaper_8. Chaque adaptateur ControlNet/T2I nécessite que l’image qui lui est transmise soit dans un format spécifique comme les cartes de profondeur, les cartes de contours, etc. I used the same model for both the base and the refiner. com Sep 9, 2024 · Step 4: Add the LoRA Node for SD1. In this section, we’ll use the ComfyUI-Hunyuan3DWrapper plugin to implement Hunyuan3D 2. 0 license and offers two versions: 14B (14 billion parameters) and 1. Intermediate SDXL Template. First, search for an example workflow on Civitai. 9. Here is an example of how to create a CosXL model from a regular SDXL model with merging. Techniques for utilizing prompts to guide output precision. Ce guide explique comment configurer ComfyUI sur votre ordinateur Windows pour exécuter Flux. Download Required Files Feb 21, 2024 · SDXL-Lightning: Progressive Adversarial Diffusion Distillation. safetensors from the control-lora/revision folder and place it in the ComfyUI models\clip_vision folder. Here is an example for the depth lora. If you get a message saying something about missing nodes, go to the ComfyUI manager and click "install missing custom nodes" anyways, feel free to ask any questions in the discussion of this page! Enjoy my workflow! :> Here is an example for the full canny model: They are also published in lora format that can be applied to the flux dev model: flux1-canny-dev-lora. One guess is that the workflow is looking for the Control-LoRAs models in the cached directory (which is my directory on my computer). I’ve reorganized the corresponding workflows and added model download links to the workflow files. Community Flux Controlnets 4 days ago · v9 tested on Better Days with Touching Grass, and this PornMaster lora. 0, this one has been fixed to work in fp16 and should fix the issue with generating black images) (optional) download SDXL Offset Noise LoRA (50 MB) and copy it into ComfyUI/models/loras (the example lora that was released alongside SDXL 1. The images above were all created with this method. Loras are patches applied on top of the main MODEL and the CLIP model so to use them put them in the models/loras directory and use the LoraLoader node like this: May 12, 2025 · SDXL Turbo is a SDXL model that can generate consistent images in a single step. Stable Diffusion; Flux; If you need to download LoRA models, ComfyUI Workflow Examples; fp8 support; requires newest ComfyUI and torch >= 2. So I created another one to train a LoRA model directly from ComfyUI! By default, it saves directly in your ComfyUI lora folder. Models For the workflow to run you need this loras/models: ByteDance/ SDXL-Lightning 8STep Lora Juggernaut XL Detail Tweaker XL Created by: Etienne Lescot: This ComfyUI workflow is designed for SDXL inpainting tasks, leveraging the power of Lora, ControlNet, and IPAdapter. I'm glad to hear the workflow is useful. Download the workflow JSON file. 5 without lora, takes ~450-500 seconds with 200 steps with no upscale resolution (see workflow screenshot from ver 1. you're good to go! ===== ComfyUI에서 SDXL과 LoRA를 가장 쉽게 사용하는 방법을 찾고 계신가요? See full list on github. Upcoming tutorial - SDXL Lora + using 1. All LoRA variants: Lycoris, loha, lokr, locon, etc… are used in the same way. It facilitates the customization of pre-trained models by applying fine-tuned adjustments without altering the original model weights directly, enabling more flexible and targeted model behavior modifications. There is a “Pad Image for Outpainting” node to automatically pad the image for outpainting while creating the proper mask. May 12, 2025 · Exemples 3D - ComfyUI Workflow Stable Zero123. The proper way to use it is with the new SDTurbo Feb 21, 2024 · Saved searches Use saved searches to filter your results more quickly May 12, 2025 · In this example, we will use LoRA models in the Wan2. It'll load a basic SDXL workflow that includes a bunch of notes explaining things. LCM models are special models that are meant to be sampled in very few steps. 5 Original FP16 Version ComfyUI Workflow. Nobody needs all that, LOL. Feb 14, 2025 · In this simple example, I added stars to the sky, changed the hair and eye colors, and made the dress slightly darker. 5-ModelName,” or without renaming, create a new folder in the corresponding model directory named after the large model version like “SD1. The requirements are the CosXL base model, the SDXL base model and the SDXL model you want to convert. Separate Style and Content B-LoRA Nodes. However, if you want a particular style such as pastel colors but your base model struggles with it, you can use a different model specialized in that style as the Created by: OpenArt: What this workflow does This workflow loads an additional LoRA on top of the base model. What's new in v4. Advanced sampling and decoding methods for precise results. SDXL Turbo is a SDXL model that can generate consistent images in a single step. Region LoRA/Region LoRA PLUS Created by: MentorAi: Download Lora Model: => Download the FLUX FaeTastic lora from here , Or download flux realism lora from here . For more information check ByteDance paper: SDXL-Lightning: Progressive Adversarial Diffusion Distillation . Drag and drop the workflow file to ComfyUI to load it. That's all for the preparation, now we can start! May 12, 2025 · All LoRA flavours: Lycoris, loha, lokr, locon, etc… are used this way. You can use more steps to increase the quality. May 12, 2025 · Outpainting is the same thing as inpainting. Nov 25, 2023 · Inpainting workflow (A great starting point for using Inpainting) View Now. 0: The standard model offering excellent image quality; SDXL Turbo: Optimized for speed with slightly lower quality; SDXL Lightning: A balanced option between speed and quality; Eg. 3 Worfklow - Complejo - Complex workflow, two passes; 1. Download the ComfyUI JSON workflow below. They can be used with any SDLX checkpoint model. once this all works accordingly it'll be added to the Instant LoRA workflow as well! Found Issue: Tying in the Regional Prompting is not a problem but it seems the IPadapter for Flux doesn't like playing together with it. 0) Base model + MoXin LoRA (weight 0. Here is an example for how to use the Canny Controlnet: Here is an example for how to use the Inpaint Controlnet, the example input image can be found here. Based on the revision-image_mixing_example. Run ComfyUI, drag & drop the workflow and enjoy! Feb 1, 2024 · Ah, ComfyUI SDXL model merging for AI-generated art! That's exciting! Merging different Stable Diffusion models opens up a vast playground for creative exploration. No, for ComfyUI - it isn't made specifically for SDXL. Workflow Node Explanation 4. The proper way to use it is with the new SDTurboScheduler node but it might also work with the regular schedulers. You will need to customize it to the needs of your specific dataset. Bypass things you don't need with the switches. safetensors”. In this file, I provide three workflows: Base model; Base model + MoXin LoRA (weight 1. Connect the upscale node’s input slots like previously. 3 billion parameters), covering various tasks including text-to-video (T2V) and image-to-video (I2V). 5. Using LoRA's (A workflow to use LoRA's in your generations) View Now. safetensors and flux1-depth-dev-lora. In this example this image will be outpainted: Using the v2 inpainting model and the “Pad Image for Outpainting” node (load it in ComfyUI to see the workflow): Mar 18, 2024 · This ComfyUI workflow is designed for SDXL inpainting tasks, leveraging the power of Lora, ControlNet, and IPAdapter. How to use this workflow ** LORAs can be daisy-chained! You can have as many as you want ** Apr 30, 2024 · Load the default ComfyUI workflow by clicking on the Load Default button in the As an example in my workflow, Was able to quickly integrate a LoRA into my SDXL workflow. , selon le Yes, 8Gb card, ComfyUI workflow loads both SDXL base & refiner models, separate XL VAE, 3 XL LoRAs, plus Face Detailer and its sam model and bbox detector model, and Ultimate SD Upscale with its ESRGAN model and input from the same base SDXL model all work together. After you're all set up, you'll be able to generate SDXL images with one click: Apr 27, 2025 · How to Use MimicPC to Create SDXL Workflow? Here are the steps to use MimicPC to create an SDXL workflow using ComfyUI: Step 1: Browse and Download a Workflow. 1, Hunyuan, and LTXV 0. 1 (decreases VRAM usage, but changes outputs) Mac M1/M2/M3 support; Usage of Context Options and Sample Settings outside of AnimateDiff via Gen2 Use Evolved Sampling node; Maskable and Schedulable SD LoRA (and Models as LoRA) for both AnimateDiff and StableDiffusion usage via LoRA Hooks Created by: OpenArt: What this workflow does This workflow adds a refiner model on topic of the basic SDXL workflow ( https://openart. com/models/274793 May 12, 2025 · Voici à quoi ressemble le flux de travail dans ComfyUI : Cette image contient les mêmes zones que la précédente mais dans l’ordre inverse. I recently switched from A1111 to ComfyUI to mess around AI generated image. Replace your image's background with the newly generated backgrounds and composite the primary subject/object onto your images. The metadata describes this LoRA as: SDXL 1. A Basic Text to Image template is included with the extension, and can be loaded from ComfyUI's May 12, 2025 · Detailed Tutorial on Flux Redux Workflow. 🛟 Support Feb 27, 2025 · How does this workflow work? The workflow first generates a control image using the Text Overlay node. This custom node lets you train LoRA directly in ComfyUI! By default, it saves directly in your ComfyUI lora folder. 2 Workflow - Simple - Easy to use and simple with Upscaling 4K, just Base+VAE; 1. In this example, a single B-LoRA model is used to apply both style and content to an image. 0 workflow. Let's pick the "SDXL Text Image Enhancer" workflow for this guide. 1 I2V LoRA) Save the model to the ComfyUI/models/loras directory Aug 17, 2023 · This workflow template is intended as a multi-purpose templates for use on a wide variety of projects. 1 Models. You should try to click on each one of those model names in the ControlNet stacker node and choose the path of where your models This node is designed to dynamically load and apply LoRA (Low-Rank Adaptation) adjustments to models and CLIP instances based on specified strengths and LoRA file names. 2) This file goes into: ComfyUI_windows_portable\ComfyUI\models\clip_vision. 5 model files SDXL Turbo is a SDXL model that can generate consistent images in a single step. The workflow files and examples are from the ComfyUI Blog. 2 Workflow - Face - for Base+Refiner+VAE, FaceFix and Upscaling 4K; 1. 1 is a family of video models. If you need perfection, like magazine cover perfection, you still need to do a couple of inpainting rounds with a proper inpainting model. 0 model files: Control LoRA and prompt scheduling, advanced text encoding, regional prompting, and much more, through your text prompt. . 5/SDXL: Load LoRA Node: In ComfyUI, double-click on the workspace or right-click May 12, 2025 · SDXL Examples. List of Templates. Wan 2. Examples of ComfyUI workflows. I am using a base SDXL Zavychroma as my base model then using Juggernaut Lightning to stylize the image . SD1. The example workflow utilizes SDXL-Turbo and ControlNet-LoRA Depth models, resulting in an extremely fast generation time. 5 就行） sampler Examples of what is achievable with ComfyUI open in new window. SD and SDXL and Loras models are supported. The LCM SDXL lora can be downloaded from here. That means you just have to refresh after training (and select the LoRA) to test it! Making LoRA has never been easier! I'll link my tutorial. The usage description is inside the workflow. UNETLoader. The TL;DR version is this: it makes a image from your prompt without a LoRA, runs it through ControlNet, and uses that to make a new image with the LoRA. 5 Model Files. You will need the included LoRA, place it in ComfyUI/loras folder like usual, it's converted from the original diffusers one (that won't work in Comfy as it is) Example workflow is in the examples folder: Dec 3, 2024 · Using Stable Diffusion 3. My ComfyUI workflow was created to solve that. But captions are just half of the process for LoRA training. Dec 19, 2023 · ComfyUI won't take as much time to set up as you might expect. 1 を使用した LoRA のタスクに対応しています。 FLUX. SDXL Default ComfyUI workflow. under an RTX 4070 12GB: Updates 设置 Load LoRA. Highly optimized processing pipeline, now up to 20% faster than in older workflow versions. Notable Examples . json, which is an example workflow for Stable Diffusion 3. This time, we will use SD3. 作成 : Masashi Okumura (@ClassCat) 作成日時 : 08/30/2024 Jun 12, 2024 · ComfyUI LCM-LoRA animateDiff prompt travel workflow. The template is intended for use by advanced users. Flux Models: Ideal for reducing memory usage without compromising image quality significantly. They are intended for use by people that are new to SDXL and ComfyUI. 0 Part 5: Scale and Composite Latents with SDXL Part 6: SDXL 1. Detailed Guide to Flux ControlNet Workflow This tutorial will guide you on how to use Flux’s official ControlNet models in ComfyUI. Just load your image, and prompt and go. 2) + QingYi LoRA Mar 3, 2024 · This workflow is just something fun I put together while testing SDXL models and LoRAs that made some cool picture so I am sharing it here. 9, I run into issues. SDXL Examples. We will cover the usage of two official control models: FLUX. 5 Basic LoRA Workflow Workflow File Description. This workflow can use LoRAs, ControlNets, enabling negative prompting with Ksampler, dynamic thresholding, inpainting, and more. Bypass things you don't need. You can also easily upload & share your own ComfyUI workflows, so that others can build on top of them! :) Why I built this: I just started learning ComfyUI, and really like how it saves the workflow info within each image it generates. Top Left: SDGEN1. For example: 896x1152 or 1536x640 are good resolutions. Part 3 (this post) - we will add an SDXL refiner for the full SDXL process Aug 22, 2023 · StabilityAI have release Control-LoRA for SDXL which are low-rank parameter fine tuned ControlNet for SDXL which should support the usage of consumer level GPUs. Ending Workflow. mpgbxquso yhbuxby smzpen mwbtesx cse lsh vxjea fwcjgxla ixdep yvqngdt