Can you make a workflow that changes the weather in images? ... You have UNTIL April 28th to participate! Info on the #Challenges Discord Channel. Prize: $50 PayPal Credit! 

Weather Changes

Coco

Realistic Image Transformer

matt3o

This is an experiment for virtually "prompt free" outpainting.It is generally able to outpaint 128px per side without touching the positive/negative prompts. Depending on the image it also works pretty well up to 256px per side, sometimes adding a few keywords in the prompts helps drastically as it's impossible to foresee every scenario.It requires the ComfyUI essentials updated to the latest.Consider this a beta. Suggestions and improvements welcome.Model links:Clip vision encoder for IPA - &nbsp;<a href="https://huggingface.co/h94/IP-Adapter/blob/main/sdxl_models/image_encoder/model.safetensors" target="_blank">https://huggingface.co/h94/IP-Adapter/blob/main/sdxl_models/image_encoder/model.safetensors</a> Clip vision encoder for Style - &nbsp;<a href="https://huggingface.co/openai/clip-vit-large-patch14/resolve/main/pytorch_model.bin" target="_blank">https://huggingface.co/openai/clip-vit-large-patch14/resolve/main/pytorch_model.bin</a> Style model - &nbsp;<a href="https://huggingface.co/TencentARC/T2I-Adapter/blob/main/models/coadapter-style-sd15v1.pth" target="_blank">https://huggingface.co/TencentARC/T2I-Adapter/blob/main/models/coadapter-style-sd15v1.pth</a> RealisticVision 5.1 <a href="https://civitai.com/models/4201?modelVersionId=130072" target="_blank" style="text-align: left;">https://civitai.com/models/4201?modelVersionId=130072</a>RealisticVision 5.1 inpainting <a href="https://civitai.com/models/4201?modelVersionId=130090" target="_blank" style="text-align: left;">https://civitai.com/models/4201?modelVersionId=130090</a>

Outpainting

Peter Lunk (MrLunk)

This ComfyUI workflow by #NeuraLunk uses Keyword prompted segmentation and masking to do controlnet guided outpainting around an object, person, animal etc. aso. <a href="https://openart.ai/flow-apps/neuralunk/replace-any-background/GLKn34FnxsSwwx6QFraX" target="_blank">Check out the Flow-App here.</a> Flow-App instructions: 🔴 1. Upload a starting image of an object, person or animal etc. \n 🔴 2. Use one or two words to describe the object you want to keep. 🔴 3. Describe what you would want as a new background, start with the word(s) used in step 2. for best results! 🔴 Now press Press CREATE to see the result in a minute ! 💚 ------------------ This workflow uses nodes for image loading, edge detection with Canny, latent space encoding for inpainting with VAEEncode, and applies neural control networks to guide the modifications. The workflow concludes with nodes to preview and save the generated image after the outpainting process <a href="https://openart.ai/workflows/profile/neuralunk" target="_blank">#NeuraLunk</a>

Flow-App info: 
🔴 ① Upload a starting image of an object, person or animal etc. 
🔴 ② Use one or two words to describe the object you want to keep. 
🔴 ③ Describe what you would want as a new background, start with the word(s) used in step 2. for best results! 
🔴 Press CREATE to see the result ! 💚 And hit Like! if the results are satisfying ! 

Replace any background.

rui400

Thanks for your attention. After being released for 25 days, &lt;CuteYou&gt; Workflow has received over 30,000 views and 7,000 downloads. I'm very thrilled about this!Link: <a href="https://openart.ai/workflows/rui400/cute-youcreate-your-cartoon-likeness-using-1-photo/2vwb81MasZTJWU7Zko6L" target="_blank">https://openart.ai/workflows/rui400/cute-youcreate-your-cartoon-likeness-using-1-photo/2vwb81MasZTJWU7Zko6L</a> I am thankful for the work of OpenArt. Now, it has become a FlowApp that can run online. I am extremely happy about this.However, the previous workflow was mainly designed to run on a local machine, and it's quite complex. Users of the workflow could simplify it according to their needs. But for the online version, users cannot simplify it, resulting in a long running time, and some steps are not necessary. In order to make CuteYou run faster and more stably, I have now created a Simplified version. This mainly simplifies the background processing part - based on my tests, the running time has been halved. This is more suitable for online running. Also, if your graphics card performance is not very good, this is equally applicable. I hope everyone enjoys using it! Love from China! Happy New Year! <img src="https://cdn.openart.ai/workflow_thumbnails/MOsZBUmh14mRk71paQfu/image_KJqWHuHt_1704271649922_raw.jpg" alt="" data-href="" style=""/>

CUTE YOU!

An infinite number of options. Check the positive prompt and comment/un-comment the lines.More testing is needed especially to find the optimal negative.Better results can be obtained with an additional IPAdapter conditioning. I'll work on that too.<ul><li>bronze statue of a woman, highly detailed, 4k, high resolution</li><li>marble statue of a woman, highly detailed, 4k, high resolution</li><li>wood statue of a woman, highly detailed, 4k, high resolution</li><li>old silver statue of a woman, highly detailed, 4k, high resolution, weathered, 1700</li><li>(oil painting:1.2) portrait of a woman, textured, masterpiece, (artistry:1.2), brush strokes</li><li>(pop-art painting:1.1) of a woman, stylized, vibrant</li><li>clip art illustration of a woman</li><li>(watercolor artistry painting:1.2) portrait of a woman</li><li>action figure superhero woman</li><li>(anime action figure:1.1) woman</li><li>chibi action figure anime woman (doesn't work very well)</li><li>comic woman portrait. graphic illustration, comic art, graphic novel art, vibrant, highly detailed</li><li>(medieval:1.2) book drawing of a woman, illustration, low details</li><li>charcoal quick drawing of a woman, artistry, stylized</li><li>caricature drawing of a woman, artistry, stylized, illustration</li><li>videogame rendering of a woman, highly detailed, 3d, resident evil</li><li>videogame rendering of a woman, highly detailed, 3d, tomb raider</li><li>videogame rendering of a woman, highly detailed, 3d, cyberpunk 2077</li></ul>

Draw Me

In this first Part of the Comfy Academy Series I will show you the basics of the ComfyUI interface. What are Nodes? How to find them? What is the ComfyUI Manager? How to Link Models from Automatic 1111 into ComfyUI? How to connect Nodes? How to create Node Groups? and much more.

Lesson 1: Using ComfyUI, EASY basics - Comfy Academy

In this Part of Comfy Academy we build our very first Workflow with simple Text 2 Image. But then I will also show you some cool tricks that use Laten Image Input and also ControlNet to get stunning Results and Variations with the same Image Composition.

Lesson 2: Cool Text 2 Image Trick in ComfyUI - Comfy Academy

Let's explore Latent Upscaling in ComfyUI. It is super easy to do and can greatly improve your image quality. This is also a needed step for many other things like better img2img render, UltimateUpscale or FaceDetailer

Lesson 3: Latent Upscaling in ComfyUI - Comfy Academy

Using a very basic painting as a Image Input can be extremely effective to get amazing results. In this Lesson of the Comfy Academy we will look at one of my favorite tricks to get much better AI Images. This gives you control over the color, the composition and the artful expressiveness of your AI Art.

Lesson 4: Img2Img Painting in ComfyUI - Comfy Academy

I built a magical Img2Img workflow for you. In the first workflow, we explore the benefits of Image-to-Image rendering and how it can help you generate amazing AI images. In the second workflow, I created a magical Image-to-Image workflow for you that uses WD14 to automatically generate the prompt from the image input. This is fantastic!

Lesson5: Magical Img2Img Render + WD14 in ComfyUI - Comfy Academy

Model Switching is one of my favorite tricks with AI. We render an AI image first in one model and then render it again with Image-to-Image in a different model. This allows us to use the colors, composition, and expressiveness of the first model but apply the style of the second model to our image. 
Then we also explore Image Masking for inpainting in Comfyui, a hidden gem that is very effective.

Lesson 6: Model Switch and Masking in ComfyUI - Comfy Academy

Wei Mao

This ComfyUI workflow is designed for swapping faces with high similarity using a trained LoRA model. It starts with setting up essential custom nodes in ComfyUI, followed by configuring a sequence of operations including loading checkpoints, detecting faces, and executing the face swap using specialized nodes like "Eff. Loader SDXL", "KSampler SDXL", and "FaceDetailer". Each step is meticulously designed to ensure the generated face matches the trained subject accurately and integrates seamlessly into any desired image. For a detailed, step-by-step visual guide on how to use this workflow, check out the corresponding video tutorial at https://youtu.be/EwBE4oKC1QY. <div data-w-e-type="video" data-w-e-is-void>
<video poster="" controls="true" width="auto" height="auto"><source src="https://youtu.be/EwBE4oKC1QY" type="video/mp4"/></video>
</div>This resource is particularly helpful for visual learners and those new to ComfyUI or face swapping technology.

Swap Faces with LoRA in ComfyUI

homer_26

Workflow based on the ComfyCouple modelI made a few changes to make the interface more user-friendly, along with mixing two characters into one image.The workflow has an upscaler and I set the default to 1024x1024.For the positive prompts, don't forget to add the “couple” tag so that there is interaction between the prompts.This workflow also has metadata included when saving images in case you want to publish them on sites like Civitai.In my tests, I was able to mix colors between the personages using the BREAK tag. Be careful with the number of tags, as depending on the size you may get an error when trying to render the image. If you have any doubts or need more information, I recommend the author of the original workflow (https://github.com/Danand/ComfyUI-ComfyCouple). Example prompts:Prompt Positive (primary)1girl, couple, retropunk background, hair multicolored, looking in the camera, pink punk clothes, perfect pose, sharp focus, volumetric lighting,(detailed face:1.2), subsurface scattering, f2, 35mmPrompt Positive (secondary)1boy, couple, retropunk background, hair punk, hair multicolored, looking in the camera, neon punk clothes, perfect pose, sharp focus, volumetric lighting,(detailed face:1.2), subsurface scattering, f2, 35mm The following additional nodes will be needed:- UltimateSDUpscale (upscale)- SaveImageWithMetaData (metadata)- Comfy Couple (couple generator)

Couple Generation + Upscaler

eogana

DESIGNER 

wang vince

The use of animatelcm's cyclic sampling resulted in the best video rendering effect，You can freely describe the background prompt words you need, or you can also disable background removal

vid2vid-The fastest video rendering workflow-AnimateLCM

Aki Hung Ác

Just give a few words about location, I'll take care of the rest Checkpoint: https://civitai.com/models/397741/robmix-cosxl-edit ---------------------------Discord: &nbsp;<a href="https://discordapp.com/users/339415040308412418" target="_blank">https://discordapp.com/users/339415040308412418</a> Any support is appreciated, thank you very much :)patreon.com/akihungacbuymeacoffee.com/akihungacpaypalme/manhtuanphoto

Flexible location photoshoot workflow

kejun

首发地址：<a href="https://sanhua.himrr.com/news/1f550f3b-effb-4145-b12d-017732384d59" target="_blank">https://sanhua.himrr.com/news/1f550f3b-effb-4145-b12d-017732384d59</a> 演示合集（欢迎贡献）：<a href="https://github.com/mrrhq/fluffy-style-logo" target="_blank">https://github.com/mrrhq/fluffy-style-logo</a> Overview小红书爆火的毛茸茸风格图标，无需抽卡，非常稳定The fluffy style icons that have become extremely popular on Xiaohongshu can be generated very stable.只需一张图片即可生成，速度非常快，只需要 9 steps You can create them with just one image, and the process is very fast, requiring only 9 steps.Control Net<ul><li>depth: 用于控制 3d 程度，提供了 depth_midas 或者 depth anything，哪个效果好用哪个</li><li>canny: 用于控制造型，可以用 lineart 替换，哪个效果好用哪个，权重越小越毛绒</li><li>tile: 用于控制颜色，权重越大颜色越还原，模糊越大边缘越毛绒</li><li>color：也是用于控制颜色，起辅助作用</li><li style="text-align: start;">Depth: Used to control the degree of 3D effect. You can choose between 'depth_midas' or 'depth anything'. Use whichever one gives better results.</li><li style="text-align: start;">Canny: Used to control the shape. It can be replaced with 'lineart'. Choose whichever one provides better results, as a smaller weight makes it fluffier.</li><li style="text-align: start;">Tile: Used to control the color. The greater the weight, the more accurate the color reproduction, and the larger the blur, the fluffier the edges.</li><li style="text-align: start;">Color: Also used to control the color, serving an auxiliary role. </li></ul>download<a href="https://civitai.com/models/43331/majicmix-realistic" target="_blank">https://civitai.com/models/43331/majicmix-realistic</a> <a href="https://huggingface.co/h94/IP-Adapter/resolve/main/models/ip-adapter-plus_sd15.safetensors" target="_blank">https://huggingface.co/h94/IP-Adapter/resolve/main/models/ip-adapter-plus_sd15.safetensors</a> https://www.liblib.art/modelinfo/c17051f8b3b74f29955bffaedc649fcchttps://www.liblib.art/modelinfo/77be9311a31b463da2af682c2c849052https://www.liblib.art/modelinfo/da0bb8cf61094f32bc36607e58fb3aa1more result thumbnails<a href="https://www.xiaohongshu.com/user/profile/637f0b23000000001f018fc5" target="_blank">https://www.xiaohongshu.com/user/profile/637f0b23000000001f018fc5</a>

furry&fluffy style icon V1.1 毛茸茸风格图标 V1.1 

AI KSK

<pre style="text-align: left;"><code>A collection of various redrawing techniques
如果你能看懂中文，可以查阅我本人制作的使用教程:https://www.bilibili.com/video/BV1ip421S7jM/</code></pre>

Inpaint Anything: clothes, hats, hair, eyes, background, anything

M.D. Programming

Simple and neat weather changer using CLIP Interrogator and a simple Canny ControlNet while staying true to the original composition. I've included 4 different weather conditions in the workflow for test, connect the desired positive prompt and its corresponding negative prompt. You can add more weather conditions, just write a simple prompt with simple keywords and set their weights to 1.3.And also add a few keywords to the negative prompt.

Weather Changer XL (Weekly Challenge)

andiamo

This workflow has been adapted from <a href="https://github.com/TencentQQGYLab/ComfyUI-ELLA/tree/main/examples" target="_blank">https://github.com/TencentQQGYLab/ComfyUI-ELLA/tree/main/examples, I just added an additional step for a simple HiresFix.</a>ELLA models use LLMs for better prompt adherence and understanding.

OpenArt

Flow

|

Builder

Home

Active Sessions

New Featured ComfyUI Workflows

Swap Faces with LoRA in ComfyUI

Couple Generation + Upscaler

DESIGNER

vid2vid-The fastest video rendering workflow-AnimateLCM

Flexible location photoshoot workflow

furry&fluffy style icon V1.1 毛茸茸风格图标 V1.1

Inpaint Anything: clothes, hats, hair, eyes, background, anything

Weather Changer XL (Weekly Challenge)

ELLA SD1.5 + LORA + HiresFix

FACEID+CN

Top Creators of the Week

ComfyUI Academy

Lesson 1: Using ComfyUI, EASY basics - Comfy Academy

Lesson 2: Cool Text 2 Image Trick in ComfyUI - Comfy Academy

Lesson 3: Latent Upscaling in ComfyUI - Comfy Academy

All ComfyUI Workflows

Loading...